The news business does not do well with numbers. But, some would say the news is full of numbers. And technically that is true. Many news stories feature graphs and charts that are full of very precise very accurate numbers. But the numbers in these graphs and charts aren't really important. Instead what is going on is a "cool graphic". The modern news business is almost entirely about pictures and a "cool graphic" is an effective kind of visual. But so is a picture of a scantily clad starlet. Now I like to gaze fondly at scantily clad starlets so I am all for this sort of thing. But I don't confuse pictures of scantily clad starlets with news. And I bet that if a news producer has the choice between a scantily clad starlet and a cool graphic to illustrate a story the starlet will win 100 times out of 100. So what's important about the cool graphic is not the numbers its the coolness.
And thus I introduce numbers. I use the number "100" above, twice. How important was the specific number I picked? Not very. What I needed was a number that was "big" but not incredibly big. And I am using the word "incredibly" in the sense where it measures how believable something is. In this example "big" was important, I was going for "impact". But credibility was also important. I wanted you to believe what I said. If I had used 1000 instead of 100 I would have gained bigness but lost believability. 100 seemed like the right balance between a number big enough to get impact but not so big to lose believability. And this is a long windy way of demonstrating that the psychological impact of a number is important.
Amping up the psychological impact is incredibly important in the news business. One simple strategy is to use a big number instead of a small number. This plays out by consulting an "expert" who has an ax to grid and who will provide you with an exaggerated estimate of how likely, big, or important something is. If expert #1, who actually is an expert, says "nothing to worry about" and expert #2, who is more interested in shilling for his cause than in enlightenment, says "be afraid, be very afraid", guess who gets lots of air time and who does not. And news producers go into orgasms if they can get two dueling "experts", one saying "it's very very very red" and the other saying "its very very very blue". It doesn't matter that a real expert might say "its mildly green".
This strategy works best in areas where the average audience member doesn't know much about the subject. So we have seen a lot of this surrounding the tragedy in Japan, particular the nuclear problems. Radiation exposure is such a subject. Scientists have gotten very good at measuring radioactivity very accurately. Theoretically, people should know a lot about this subject. It has been a matter of intense public interest since at least 1945, when A-bombs were dropped on Japan. But most of the public discussion has been on a level with my red/blue example above. Over the years the pro-nuclear camp has been saying "no danger here" and the anti-nuclear camp has been saying "any tiny amount of radiation is extremely dangerous". The facts support the green position. We live in a sea of low level radiation. It is literally everywhere. So there are things to worry about but a lot of the "scare" coverage is exaggerated.
The "radiation" story out of Japan is one where there is at least some justification for a difference of opinion. And at least some segments of the media are trying to clarify the situation rather than obscure it. But it is an aspect of the story where these is some justification for the media's actions. Unfortunately, there are many other aspects of the Japan story that illustrate the media's complete inability to deal properly with numbers.
This comes out most strikingly with respect to casualty numbers. The media is very good at conveying the difference between zero and one. A story in which no one dies is covered very differently than a story in which one person dies. The first story becomes a "miracle rescue" story, like the miners in Chile. The second story becomes a "Law and Order" story; who died, who did it, etc. That's OK. But what about where in one case one person dies and in the other case two people die. Given our zero/one example one would think that the coverage would be completely different. But the media coverage is only slightly different. Now it's an "individual" versus a "group" story. But the coverage of the two stories will be not very different.
Moving on, what about a two casualty story versus a ten casualty story? The difference should be large. In the latter case, eight extra people are dead. And remember the death of the one person in the "one death" story was important enough to justify coverage. But the coverage is almost identical. Recently a bus crashed in New Jersey killing 2. This happened within a few days of the Bronx bus crash that killed more than ten. There has been a little more coverage of the Bronx crash but the media approach to each crash has been more similar than different.
This inability to differentiate gets even more pronounced as the numbers get bigger. What if 100 people are killed? Is this different than only 10 people getting killed. No! The media will either choose to cover the story or it won't. If both stories are covered they will both be "a lot of people were killed" stories. The death toll in Japan has crossed 10,000, as I write this. It is expected to continue to rise. The final total will likely pass 20,000. But in the end about one person will be killed in the Japan tragedy for every ten people who perished in Haiti under roughly similar circumstances. The media is completely incapable of differentiating in any meaningful sense between these two numbers. But the difference is hundreds of thousands of lives.
The media has fallen into the "up close and personal" trap. For disasters they show "devastation" video. With their close in emphasis it all looks pretty much the same. They have not figured out how to convey the extent of the devastation. If they have enough devastation video to fill up a "clip loop" it all looks the same. You get a number of short clips, usually about 10, each showing a piece of devastation. The same clip loop is run over and over. So once you have enough devastation to make up the ten short shots all disasters look the same. All disasters are visually boiled down to the clip loop and all clip loops end up looking pretty much the same.
The human toll is handled in a similar manner. We get interviews of survivors or people who knew a victim. Again about 10 of these interviews is all the media can absorb. So if a disaster generates 10 interviewees or millions of interviewees it is all the same. But a disaster which creates 10 interviewees is not the same as a disaster that generates millions of interviewees. But it will be pretty much impossible to tell one disaster from the other based on the media coverage.
Several years ago an earthquake hit my city. Within a few hours the media had put together their clip loop of the event. That's when the calls and e-mails started coming in. Was I OK? Was my home, car, or place of business wiped out in the terrible devastation? How many friends had been killed or seriously injured? That sort of thing. In fact, no one I knew was killed, injured, or even had suffered any property damage. Because no one was killed and only a few people were injured. There was serious damage to a few specific areas but 99% of the city was completely undamaged. You couldn't tell any of this because the national media picked up the same highlight loop showing the few instances of damage and ran it over and over.
Shortly thereafter 9/11 happened. 9/11 was a much bigger event involving thousands of deaths, and much more property damage than my little earthquake. 9/11 looked like a bigger event than the earthquake in my town. And it has received vastly more coverage, partly because it happened in a media mecca. But it is literally impossible to accurately gage the size and scope of the two events based solely on the media coverage. Numbers might help. But, as I have shown, the media is not good with numbers.
Saturday, March 19, 2011
Thursday, March 10, 2011
Pensions
Pension plans, particularly the one in the state of Wisconsin, are in the news these days. I don't know the details of the Wisconsin pension plan and neither do most of the people arguing about it. But I do know that it is what's called a "defined benefit" plan. There has been a lot of talk over the years as to whether defined benefit plans are better or worse than the other general class of pension plans called "defined contribution" plans. As with a lot of things there are pluses and minuses associated with each type of plan. And that's what this piece is about. Not the pluses and minuses of the Wisconsin plan in particular but the pluses and minuses of the two classes of plans in general.
So what is a defined benefit plan? That's a plan where an employer contracts to provide a specific level of benefit at a specific time. I have been a participant in a defined benefit plan in the past. It was typical of the breed. The way it worked was that if I continued working for the company for 40 years they would provide a pension that would cover 60% of what I was getting paid when I retired. The way it worked was that my "guarantee" would go up on average 1 1/2% per year. So if I stayed with the company for a year my pension would be 1 1/2%. If I stayed 10 years I would get 15%, and so on. There were "vesting" rules so I didn't get a thing unless I was with the company for at least 5 years. And the percentage wasn't a flat 1 1/2% per year. Sometimes it was 1% for a year, sometimes, 2%, but on average it went up 1 1/2%. Why the variation? The company was interested in retaining employees so the percentage was designed to encourage employees to stick around (high percentage per year) until they were stuck then the percentage was lower.
This sounds like a pretty good deal and I thought so at the time. And most people like defined benefit plans like this one. So what's not to like? Defined benefit plans are a bad deal if they don't deliver on the promise. For instance, in my case the company was bought by another company after I had more than 15 years in. The new company honored the old plan but they folded it into their similar plan. The problem was that they had a different pattern of percentages. The old plan had higher percentages in the later years and lower percentages in the earlier years. The new plan had the reverse. So if you added up all the percentages in my case (lower percentages in the earlier years and now lower percentages in the later years) it no longer added up to the full 60%. And I got laid off after just under 20 years. In my case I did end up with something. But it was a low percentage of my salary as it was when I was laid off many years before retirement. But I at least got something.
The cost to the employer of a defined benefit plan depends on a lot of things. How many people will go the full 40 years. How much will they be earning when they retire. Many other things. And there is something called "present value". My old company made contributions to the plan each year and those contributions were invested. So when I worked my 15th year for the company they put in some amount of money that was supposed to cover what my pension would eventually cost. That money would have many years to earn interest. How much money did they need to put in that year? Well, you make a bunch of assumptions and then you perform what is called a "present value" calculation and that's how much money the company put in. Actuaries, usually employed by insurance companies, specialize in doing this kind of thing. And once all the assumptions are made there is a precise mathematical way to perform the present value computation. But it all depends on the assumptions in the end. And that's where the trouble comes from.
There is no way to get the assumptions right. If you make one set of assumptions the present value calculation says put a smaller amount of money in. If you make a different set of assumptions the present value calculation says put a larger amount of money in. Which set of assumptions are right? No one knows so honest and ethical people can disagree. And then there's the real world. As time goes by things can turn out more like the "small money" assumptions. Then it is likely that the pension plan is over funded. And, of course, the real world can turn out more like the "big money" assumptions. Then the pension plan is underfunded. For many years the stock market has grown faster than the historical trend. This has resulted in a lot of over funded pension plans. Then when the stock market crashed a couple of years ago the value of many pension fund investments dropped a lot and many pension funds instantly became underfunded. Neither of these outcomes are the result of anyone doing anything wrong.
For many years most large companies had a pension plan and it was almost always a defined benefit plan. Defined contribution plans are a recent invention. The first thing to notice about defined benefit plans is that how much money a company needs to put into the pension fund is a matter of opinion. To a company pension contributions are just another kind of expense. If you lower expenses you increase profits. So companies always want to put in as little as they can. So the first way a company can game the system is to use the "less cost" assumptions when calculating the pension contribution even if they are not justified. That's not good. But it can get worse, way worse. Back when there were many companies around with defined benefit pension plans a common tactic for an unscrupulous "take over artist" would be to borrow a lot of money, take the company over, then raid the pension fund to pay back all the money that had been borrowed, and finally to pay themselves a large "fee". After that, they didn't even care if the company stayed in business. And many companies were literally driven out of business by these tactics.
And what happens to the pension fund if the company goes out of business? Well, lots of times the fund is raided before the company goes completely under. And even if this doesn't happen there will be no more contributions from the company. So the fund is likely underfunded. It used to be that the retirees just got completely screwed in these kinds of situations. Now there is a federal agency called the Pension Benefits Guarantee Corporation and a law to go with it. It can contain the damage to some extent but it does not pay the full contracted benefit. And it is perennially short of money. It is currently dealing with the GM and Chrysler pensioners, for instance.
So, if everything works out well a defined benefit pension is a good deal for employees. But things frequently don't work out. And many of the things that can go wrong with a private corporation do not apply to governments like the Wisconsin state government. But some of them do. A good way to reduce current spending is to underfund the state pension. And state legislatures do this all the time. It is one of the standard "budget gimmicks" you hear about. The theory is that the state will make up the shortfall later when the economy is better. But later there is always something more fun to spend the money on that a pension contribution. As far as I can tell (the press coverage on this sort of thing is always poor) Wisconsin was fine before the market crashed and will be fine again if the market goes up enough. But in the same way financial considerations affect business leaders, political considerations affect politicians. This can result in bad behavior that jeopardizes retirees ability to get what they were promised.
So now we know the bad news about defined benefit plans. So what's the bad news about "defined contribution" plans? Before getting into that let me explain how a defined contribution plan works. The best place to start is with a 401k plan. In a 401k plan (applicable to a for profit corporation, there are similar plans with different designations for non-profits and governments) the employee makes a contribution to the plan, say 5% of salary. This money is taken off the top before taxes are calculated and put into a fund. Frequently there is an employer match. The most common match is that the employer will match at a rate of 50 cents on the dollar anything the employee puts in up to 6%. So in our case the employee would put in 5% and the employer would contribute an additional 2 1/2% so the total amount going into the fund would be 7 1/2%. None of this so far is a pension. But the employer can contribute in a similar manner to a fund on behalf of the employee that is not dependent on how much the employee kicks in. This is a pension plan. It is called "defined contribution" because there is a formula that determines how much the employer is going to put in (typically a fixed percentage of the employee's salary) but the employer does not guarantee how much income this will translate into when the employee retires.
Before going into the problems let me go into the benefits of this type of plan. The first benefit is that the employer can't game the system by coming up with some optimistic assumptions and underfunding the plan. There is a rule and the employer is bound by law to put in the exact amount the rule specifies. There is no wiggle room. Secondly, the employer looses control of the money. The money goes to the plan administrator, typically someone like Fidelity or Vanguard. The money is no longer under the employer's control. The employer can't raid the fund. Third, and this is not obvious, all of the "defined contribution" pension cost for a particular year is paid in the year the cost is incurred. With a "defined benefit" plan extra money may need to be put in (or possibly can be taken out) because previous years now look underfunded (or over funded). This means that a company can accurately predict the cost of doing business in the current year. In summary, with a defined contribution plan the money is reliably there and corporations get a more predictable cost structure. So what's the down side?
The down side is that the employee does not know how much income he will get at retirement. If enough money is put into the defined contribution plan and the money is invested well an employee can do as well or better than with a defined contribution plan. But if the money is invested badly there may be little or nothing left at all at retirement time. This is not as far fetched as it sounds. Let's say you were an Enron employee and Enron had a defined contribution pension plan. (I don't know what kind of plan Enron actually had). With many companies the 50% match on the 401k is in company stock. And many employees believe in the company they work for so they invest their portion of the 401k money in company stock too. And some companies will contribute company stock in lieu of cash as their pension contribution. So in this case it would be possible for all of an employee's pension money to be tied up in Enron stock. Enron stock is now worthless so people who were 100% in Enron were completely wiped out. Enron was a high flier. But for most if its life Washington Mutual was a well run and staid bank, technically a mutual savings bank. For most of the life of the company, employees would be justified in believing that WaMu stock was a conservative investment. Many WaMu employees had a large part of their retirement tied up in WaMu stock. That investment is now worth a few cents on the dollar and may eventually be worthless. So the down side of defined contribution plans is the investment risk.
In summary, defined benefit and defined contribution pension plans both have risks. With the defined benefit plan you are betting on the company being reasonably ethical and surviving for a long time. It should be noted that the only stock listed in the 1900 Dow Jones Industrial average that was still there 100 years later was GE. You are also betting you can stay with the same company for your entire adult carrier. I worked for three companies as an adult. I worked for the same company for 19, 4, and 15 years. And I worked hard to stay with a company after I got there. In no case did I quit, although I did retire from the last one. Most people are much less successful at longevity than I have been. My 4 year stint would be a dead waste to a defined benefit plan. I do get a nice to have but very modest pension from my 19 year stint. By the time I retired from my 15 year stint the company had converted to a defined contribution plan. So it is a lot harder than it looks to cash in on a defined benefit plan.
I also was able to avoid the "put all your eggs in one stock" problem I have described above. It was part of a pretty conservative investment strategy that has stood me in good stead. But there's no doubt about it. It would have been easy to screw up some time over the years and lose all or a lot of my retirement money. So it is also harder than it looks to cash in on a defined contribution plan. Most people are more aware of the risks inherent in a defined contribution plan so they characterize it as the more risky of the two. I hope I have convinced you that it is not necessarily more risky and is frequently less risky. And the fact that both options are risky is why I strongly advocate leaving Social Security pretty much as it is. Assuming it can survive the current political attack on it, it represents the only truly low risk retirement option available to people. Everyone needs a low risk component in their retirement plan.
Finally, what should be done about government pension plans like the one in Wisconsin? Probably the biggest reason private defined benefit pension plans are risky is bad behavior by company management. They either abuse the pension plan or they run the company badly. There is no reason to believe that state legislatures are likely to do a better job. The current activities of the legislatures in Wisconsin and several other states strongly supports this contention. The money pot invariably associated with a defined benefit plan just represents too tempting a target. So I think that governments should follow the lead of private industry. Private industry has almost entirely shifted over to defined contribution pension systems. I think it is both inevitable and likely that in the coming years governments will switch over to defined contribution systems too. The switchover needs to be watched carefully. It can be done properly so that the employee at least starts out with an equivalent benefit. But it can also turn into a license to steal if done badly.
But once done opportunities for mischief are greatly reduced. It now becomes a discussion between the employer and employee about total compensation. The employer is now indifferent as to how much of the compensation is in salary and how much is in the form of a pension contribution. He only cares about the total amount. And there is no longer an opportunity to raid the pension fund or to underfund the plan in the current year. Pensions would be off the table completely in Wisconsin if the state had a defined contribution plan. And that would be better for state employees and for the rest of us.
So what is a defined benefit plan? That's a plan where an employer contracts to provide a specific level of benefit at a specific time. I have been a participant in a defined benefit plan in the past. It was typical of the breed. The way it worked was that if I continued working for the company for 40 years they would provide a pension that would cover 60% of what I was getting paid when I retired. The way it worked was that my "guarantee" would go up on average 1 1/2% per year. So if I stayed with the company for a year my pension would be 1 1/2%. If I stayed 10 years I would get 15%, and so on. There were "vesting" rules so I didn't get a thing unless I was with the company for at least 5 years. And the percentage wasn't a flat 1 1/2% per year. Sometimes it was 1% for a year, sometimes, 2%, but on average it went up 1 1/2%. Why the variation? The company was interested in retaining employees so the percentage was designed to encourage employees to stick around (high percentage per year) until they were stuck then the percentage was lower.
This sounds like a pretty good deal and I thought so at the time. And most people like defined benefit plans like this one. So what's not to like? Defined benefit plans are a bad deal if they don't deliver on the promise. For instance, in my case the company was bought by another company after I had more than 15 years in. The new company honored the old plan but they folded it into their similar plan. The problem was that they had a different pattern of percentages. The old plan had higher percentages in the later years and lower percentages in the earlier years. The new plan had the reverse. So if you added up all the percentages in my case (lower percentages in the earlier years and now lower percentages in the later years) it no longer added up to the full 60%. And I got laid off after just under 20 years. In my case I did end up with something. But it was a low percentage of my salary as it was when I was laid off many years before retirement. But I at least got something.
The cost to the employer of a defined benefit plan depends on a lot of things. How many people will go the full 40 years. How much will they be earning when they retire. Many other things. And there is something called "present value". My old company made contributions to the plan each year and those contributions were invested. So when I worked my 15th year for the company they put in some amount of money that was supposed to cover what my pension would eventually cost. That money would have many years to earn interest. How much money did they need to put in that year? Well, you make a bunch of assumptions and then you perform what is called a "present value" calculation and that's how much money the company put in. Actuaries, usually employed by insurance companies, specialize in doing this kind of thing. And once all the assumptions are made there is a precise mathematical way to perform the present value computation. But it all depends on the assumptions in the end. And that's where the trouble comes from.
There is no way to get the assumptions right. If you make one set of assumptions the present value calculation says put a smaller amount of money in. If you make a different set of assumptions the present value calculation says put a larger amount of money in. Which set of assumptions are right? No one knows so honest and ethical people can disagree. And then there's the real world. As time goes by things can turn out more like the "small money" assumptions. Then it is likely that the pension plan is over funded. And, of course, the real world can turn out more like the "big money" assumptions. Then the pension plan is underfunded. For many years the stock market has grown faster than the historical trend. This has resulted in a lot of over funded pension plans. Then when the stock market crashed a couple of years ago the value of many pension fund investments dropped a lot and many pension funds instantly became underfunded. Neither of these outcomes are the result of anyone doing anything wrong.
For many years most large companies had a pension plan and it was almost always a defined benefit plan. Defined contribution plans are a recent invention. The first thing to notice about defined benefit plans is that how much money a company needs to put into the pension fund is a matter of opinion. To a company pension contributions are just another kind of expense. If you lower expenses you increase profits. So companies always want to put in as little as they can. So the first way a company can game the system is to use the "less cost" assumptions when calculating the pension contribution even if they are not justified. That's not good. But it can get worse, way worse. Back when there were many companies around with defined benefit pension plans a common tactic for an unscrupulous "take over artist" would be to borrow a lot of money, take the company over, then raid the pension fund to pay back all the money that had been borrowed, and finally to pay themselves a large "fee". After that, they didn't even care if the company stayed in business. And many companies were literally driven out of business by these tactics.
And what happens to the pension fund if the company goes out of business? Well, lots of times the fund is raided before the company goes completely under. And even if this doesn't happen there will be no more contributions from the company. So the fund is likely underfunded. It used to be that the retirees just got completely screwed in these kinds of situations. Now there is a federal agency called the Pension Benefits Guarantee Corporation and a law to go with it. It can contain the damage to some extent but it does not pay the full contracted benefit. And it is perennially short of money. It is currently dealing with the GM and Chrysler pensioners, for instance.
So, if everything works out well a defined benefit pension is a good deal for employees. But things frequently don't work out. And many of the things that can go wrong with a private corporation do not apply to governments like the Wisconsin state government. But some of them do. A good way to reduce current spending is to underfund the state pension. And state legislatures do this all the time. It is one of the standard "budget gimmicks" you hear about. The theory is that the state will make up the shortfall later when the economy is better. But later there is always something more fun to spend the money on that a pension contribution. As far as I can tell (the press coverage on this sort of thing is always poor) Wisconsin was fine before the market crashed and will be fine again if the market goes up enough. But in the same way financial considerations affect business leaders, political considerations affect politicians. This can result in bad behavior that jeopardizes retirees ability to get what they were promised.
So now we know the bad news about defined benefit plans. So what's the bad news about "defined contribution" plans? Before getting into that let me explain how a defined contribution plan works. The best place to start is with a 401k plan. In a 401k plan (applicable to a for profit corporation, there are similar plans with different designations for non-profits and governments) the employee makes a contribution to the plan, say 5% of salary. This money is taken off the top before taxes are calculated and put into a fund. Frequently there is an employer match. The most common match is that the employer will match at a rate of 50 cents on the dollar anything the employee puts in up to 6%. So in our case the employee would put in 5% and the employer would contribute an additional 2 1/2% so the total amount going into the fund would be 7 1/2%. None of this so far is a pension. But the employer can contribute in a similar manner to a fund on behalf of the employee that is not dependent on how much the employee kicks in. This is a pension plan. It is called "defined contribution" because there is a formula that determines how much the employer is going to put in (typically a fixed percentage of the employee's salary) but the employer does not guarantee how much income this will translate into when the employee retires.
Before going into the problems let me go into the benefits of this type of plan. The first benefit is that the employer can't game the system by coming up with some optimistic assumptions and underfunding the plan. There is a rule and the employer is bound by law to put in the exact amount the rule specifies. There is no wiggle room. Secondly, the employer looses control of the money. The money goes to the plan administrator, typically someone like Fidelity or Vanguard. The money is no longer under the employer's control. The employer can't raid the fund. Third, and this is not obvious, all of the "defined contribution" pension cost for a particular year is paid in the year the cost is incurred. With a "defined benefit" plan extra money may need to be put in (or possibly can be taken out) because previous years now look underfunded (or over funded). This means that a company can accurately predict the cost of doing business in the current year. In summary, with a defined contribution plan the money is reliably there and corporations get a more predictable cost structure. So what's the down side?
The down side is that the employee does not know how much income he will get at retirement. If enough money is put into the defined contribution plan and the money is invested well an employee can do as well or better than with a defined contribution plan. But if the money is invested badly there may be little or nothing left at all at retirement time. This is not as far fetched as it sounds. Let's say you were an Enron employee and Enron had a defined contribution pension plan. (I don't know what kind of plan Enron actually had). With many companies the 50% match on the 401k is in company stock. And many employees believe in the company they work for so they invest their portion of the 401k money in company stock too. And some companies will contribute company stock in lieu of cash as their pension contribution. So in this case it would be possible for all of an employee's pension money to be tied up in Enron stock. Enron stock is now worthless so people who were 100% in Enron were completely wiped out. Enron was a high flier. But for most if its life Washington Mutual was a well run and staid bank, technically a mutual savings bank. For most of the life of the company, employees would be justified in believing that WaMu stock was a conservative investment. Many WaMu employees had a large part of their retirement tied up in WaMu stock. That investment is now worth a few cents on the dollar and may eventually be worthless. So the down side of defined contribution plans is the investment risk.
In summary, defined benefit and defined contribution pension plans both have risks. With the defined benefit plan you are betting on the company being reasonably ethical and surviving for a long time. It should be noted that the only stock listed in the 1900 Dow Jones Industrial average that was still there 100 years later was GE. You are also betting you can stay with the same company for your entire adult carrier. I worked for three companies as an adult. I worked for the same company for 19, 4, and 15 years. And I worked hard to stay with a company after I got there. In no case did I quit, although I did retire from the last one. Most people are much less successful at longevity than I have been. My 4 year stint would be a dead waste to a defined benefit plan. I do get a nice to have but very modest pension from my 19 year stint. By the time I retired from my 15 year stint the company had converted to a defined contribution plan. So it is a lot harder than it looks to cash in on a defined benefit plan.
I also was able to avoid the "put all your eggs in one stock" problem I have described above. It was part of a pretty conservative investment strategy that has stood me in good stead. But there's no doubt about it. It would have been easy to screw up some time over the years and lose all or a lot of my retirement money. So it is also harder than it looks to cash in on a defined contribution plan. Most people are more aware of the risks inherent in a defined contribution plan so they characterize it as the more risky of the two. I hope I have convinced you that it is not necessarily more risky and is frequently less risky. And the fact that both options are risky is why I strongly advocate leaving Social Security pretty much as it is. Assuming it can survive the current political attack on it, it represents the only truly low risk retirement option available to people. Everyone needs a low risk component in their retirement plan.
Finally, what should be done about government pension plans like the one in Wisconsin? Probably the biggest reason private defined benefit pension plans are risky is bad behavior by company management. They either abuse the pension plan or they run the company badly. There is no reason to believe that state legislatures are likely to do a better job. The current activities of the legislatures in Wisconsin and several other states strongly supports this contention. The money pot invariably associated with a defined benefit plan just represents too tempting a target. So I think that governments should follow the lead of private industry. Private industry has almost entirely shifted over to defined contribution pension systems. I think it is both inevitable and likely that in the coming years governments will switch over to defined contribution systems too. The switchover needs to be watched carefully. It can be done properly so that the employee at least starts out with an equivalent benefit. But it can also turn into a license to steal if done badly.
But once done opportunities for mischief are greatly reduced. It now becomes a discussion between the employer and employee about total compensation. The employer is now indifferent as to how much of the compensation is in salary and how much is in the form of a pension contribution. He only cares about the total amount. And there is no longer an opportunity to raid the pension fund or to underfund the plan in the current year. Pensions would be off the table completely in Wisconsin if the state had a defined contribution plan. And that would be better for state employees and for the rest of us.
Saturday, March 5, 2011
Robot Cars
This is the third and final installment of my "robots in transportation" series. The first one was http://sigma5.blogspot.com/2010/12/space-final-frontier.html. It argued in favor of shutting down the manned part of the space program and going with robot space probes. The second installment: http://sigma5.blogspot.com/2011/01/robot-jet-fighters.html discussed unmanned airplanes. Here I discuss robot cars. But before I get into replacing drivers with robots I am going to discuss some other aspects of cars of the future.
Does the car have a future? There are some who would argue that it does not. There is a small but vocal contingent of people in my town that hate cars. They walk, bicycle, or take public transportation. They see cars as evil incarnate. In the most general sense they are arguing against personal transportation, unless it is people powered. So, taking things in reverse order, is people powered transportation practical? People have been walking since there were people. And animals have been walking for a lot longer. So in some sense walking is practical. But during the long period when walking was how most people got around most people never ventured more than 25 miles from where they were born. I for one, do not want to give up the option of venturing further abroad.
Walking also has another disadvantage. You can't carry much along with you. If you intend to walk more than about 10 miles in a day I would estimate that it would be impossible for most people to carry more than about 100 lbs. Many people couldn't handle even this amount. And if you want to go further you need to cut down on your load. If you want to carry more, and the limit would be less than 200 lbs, you would not be able to walk even 10 miles per day. Domestic animals have been around for about 10,000 years as a way to improve the distance/load calculus. People have also been inventing things like ships and wagons as another solution to the problem.
A more recent invention is the bicycle. It is another approach to beating the distance/load calculus. There is an annual "Seattle to Portland" bicycle race in my neck of the woods that demonstrates this. Most participants travel the roughly 200 mile distance in two days. Many of them do it in a day. So a bicycle allows you to travel four to eight times as far as pure foot power would permit. This is a definite improvement. But most people, given the option of trading a bicycle for an automobile, opt for the car. This is most obvious in China. We all have seen video of hordes of bicycles on the streets of Beijing a few years ago. But Chinese are deserting their bikes for cars in the millions. China is now the largest single car market in the world.
Public transportation, typically in the form of buses, but also in the form of light rail, is touted by many as the "correct" alternative to cars when feet or bicycles are not the answer. Why? Well, when you strip the argument down it is efficiency. Public transportation is more efficient and produces less pollution than cars. There is also the gridlock problem. Let's take each of these issues separately.
The theory is that public transportation is more efficient. But is it? If you take a bus and fill it full of people it will be cheaper per passenger mile than the equivalent number of cars, each with only one person in it. That's the way the efficiency argument is usually presented. But are the buses really full? Currently the answer is pretty much yes. But this is because the number of buses is far less than the number that would be needed to meet the demand in a car free environment. Bus systems all lose money and are limited in size to what the taxpayer will support. Tax payer support falls off rapidly as the load (number of people on the typical bus run) decreases. So the current subsidy is only enough to provide for a few pretty full buses. This works because there are lots of cars around to take care of most of the transportation need.
There is also a hidden cost to buses and other mass transportation solutions. That is lost time. One of the real benefits to a car is in it I can go long distances whenever I want. I take a lot of short in-city trips. Frequently I have some flexibility as to when I go so theoretically I could time the trip to fit the bus schedule. But much of the time this is not true. I got my hair cut today. It took me about 10 minutes each way. In a bus, if I timed it right, it might have taken me 20 minutes each way. So right away my travel time doubled. Next, it was to an appointment. So I really needed to travel to the appointment at a specific time. It is possible to go early but whatever time I would have waited between when I arrived and when my appointment started would have been lost time. Even on the most traveled bus routes an "every 20 minutes" schedule is about as good as it gets. So I would have lost another ten minutes in synchronizing with the bus schedule. And all this is true for my trip home too. So in this semi-ideal situation 20 minutes of travel time has ballooned up to 60 minutes. And this is for an in-city to in-city trip.
Buses (or light rail) don't go most suburban places. And they do not go every 20 minutes and they do not run all the time. Let's say we fixed that. Buses would now go everywhere and they run all the time on an "every 20 minutes" schedule. What do things look like now? First, we need a lot of buses, between 10 and 100 times as many buses as we now have. We would also have to use some kind of "hub and spoke" system. It is impractical to have buses running from everywhere to everywhere. So for a lot of trips you would take a local to a hub, a trip to a second hub, and finally a local to your actual destination. That means a 10 minute delay at the first hub, a 10 minute delay at your second hub, and arriving at your destination 10 minutes early. We add an hour to a typical longer round trip. And, in order to meet our "every 20 minutes" and our "goes everywhere" requirement we are going to be running most buses pretty empty a lot of the time and some buses completely empty some of the time. This scenario I have outlined may seem unrealistic but it is exactly what cars provide. I can get in my car whenever I want and go reasonably directly to wherever I want. My route and schedule is completely independent of anyone else's route and schedule. What is your time worth? There is an incredibly large time penalty to shifting most people from traveling by car to traveling by public transportation.
And, once you increase the density of public transportation sufficiently to reduce the time cost to a reasonable amount the efficiency goes out the door. And the efficiency of public transportation is none too good now. There is no public transportation system in existence that recovers all of its costs. If you increase the quality of service of a public transportation system to anything approaching that currently provided by cars it becomes fantastically expensive. Even in a place like New York City with its subway system that was built 100 years ago and its very high density, public transportation is heavily subsidized and they have lots of automobile traffic.
Next, let's consider pollution. Cars are a definite improvement over horses. The release far less pollution per mile. But is a car inherently a polluter? The answer is no! We know this because we can now buy electric cars. The single problem with electric cars is that the current battery technology sucks. The motors that turn the propellers of the aircraft carrier Enterprise are electric. Given this it should be clear that there is absolutely no problem making electric motors that will provide all the performance anyone could want. But current batteries can't store much power. So manufacturers put in wimpy motors to make the batteries last longer so people think electric cars are wimpy. Fix the battery problem and you fix the wimpy problem. Until the crappy battery problem is fixed electric cars are not for everybody.
Given cheap gas and a lack of powerful cheap batteries we will continue to have gas powered cars that pollute. The obvious solution, if we want to reduce the pollution problem, is to make gas more expensive. I drove a big old car while I was in college. It ran on "super". One day I bought super for 29.9 cents/gallon. I said to myself "I will never buy super cheaper in my life". I was right. That was a lot of years ago but it demonstrates what has been happening to gas prices over the last 40 years. Even with the price increases gas is still cheap. But let's assume it gets expensive or we decide for other reasons that we need to make cars much more efficient. What will we do?
The current answer is a hybrid. This is a combination of gas and electric. There are a number of ways to do hybrid and it is not clear how most hybrids work now. In a purely gas car you have an engine, a transmission, some mechanical connections like the differential and this all this spins shafts that turn wheels. This is a pretty inefficient process. There is another way to do this, the way aircraft carriers and diesel locomotives do it. There the primary motor (nuclear or diesel) that is connected to a generator. The electricity is fed to electric motors that spin the wheels. For powerful machines like locomotives and aircraft carriers, this is the most efficient way to do it. It seems to me that this should be the most efficient way to do cars. You would also throw some batteries and a more complex "control" system in between the generator and the motors on the wheels. This approach has many advantages. You can put more or less batteries in. You can add in the capability of charging the batteries from the electric grid. More batteries and charging allows you to run the car on electricity more of the time. The motor generator approach allows you to make the motor more efficient because it doesn't have to run at different speeds and loads. It is no longer connected to the wheels. You get rid of the transmission and other mechanical equipment that is inefficient at transferring power along the line to the wheels. This general approach should result in the most efficient car. So why don't all cars do it this way?
It turns out that if you do the same thing a lot of times for a long time you get very efficient at it. Auto makers have been making a lot of traditional cars for a long time. They have gotten very efficient at it. Going to the design I recommend means learning how to do a lot of new things very efficiently. It may be cheaper to do the theoretically less efficient thing (a car with a lot of old technology) because you are so darn good at doing the old thing. Eventually some auto makers will figure out how to be good at making the new kinds of cars. They will force the rest to figure it out too. It might take a while but probably less than 10 years.
Moving from our current "gas" car to hybrid cars is hard but doable. Most of our infrastructure; most of the manufacturing process, roads, gas stations, etc., will require little or no modification. Switching to all electric cars would require a much greater change. Batteries are the critical problem. I have been following battery technology for 30 years. The newest batteries are better than the old ones but we have merely moved from appalling to awful. We need to move all the way to good. Then there is our electricity grid. It is not set up to handle the load that moving our transportation system over to a high percentage of electric would require. Part of it is just more as in more generating capacity and more transmission lines. We know how to do these things. The part we don't know how to do is storage. Our electric grid is real time. It has a little inertia built into it but mostly it generates and distributes what is needed now. Any serious imbalance between supply and demand results in outages. And wind, solar, and some other sources are intermittent. It would be nice if we could store the excess from some periods to cover the shortages from other periods. Being able to store large amounts of power for a few days or even a few hours would make a tremendous difference. This is the wholesale version of the electric car battery problem. Again, I have been watching this area for over 30 years and not much progress has been made. Nuff said. Back to cars.
There is another "clean car" idea out there. That's fuel cell/hydrogen. The idea is to use a fuel cell to turn hydrogen into electricity. NASA has been doing this since the '60s so its something we know how to do. The problem is not in the fuel cell it is in the Hydrogen. If you are NASA sending a space probe to the back side of beyond, the many problems associated with dealing with Hydrogen are worth the hassle. But this is not true here on Earth. Hydrogen has two problems: making it and storing it. There is no Hydrogen loose around to collect. You have to make it. Hydrogen is a constituent of lots of things including water. But Hydrogen really likes to combine with stuff. So it does. To make it into a fuel we have to uncombine it. That takes lots of energy. Well, the whole point of Hydrogen is as a source of energy so this whole thing about consuming a lot of energy to create Hydrogen is just wrong. A simple way to make Hydrogen is to use electricity to separate out the Hydrogen in water. Other than the fantastic amounts of energy this requires it works pretty well. And that's the problem with the many alternatives. You end up having to use a lot of energy to make the Hydrogen. This is not good. Then when you have made the Hydrogen you have made something that loves to recombine with other stuff. The name for this "combine" process, in many cases, is explosion. So you need to be very careful how you handle things or you get explosions or perhaps just a very large very hot fire.
So Hydrogen is dangerous to store and you have to be very careful. That sounds like the "storage" problem but it's not. To store gas in a car you make this thing that is a couple of cubic feet in size out of sheet metal called a gas tank. It's not very big and it's not very heavy, even full of gas. But in this not very big not very heavy thing you can put enough gas with enough energy to move a big SUV 300 miles. To store the same amount of energy as Hydrogen under similar conditions you would need a tank many times the size of the SUV. So you have to do something. One thing is to compress it. But you have to compress it a lot. So you need a very strong tank and you need to transfer the Hydrogen from the gas station to the car under these very high pressures. This is dangerous and expensive. And probably the gas tank is now very heavy in order to be strong enough.
Another approach is a sponge. It turns out that there are certain materials that wick up Hydrogen. They do it so well that you can get a lot of cubic feet of Hydrogen into a few cubic feet of tank. And the pressure is not very high. It's some kind of chemistry magic but it works. Effectively you get a lot of compression without a lot of pressure. You need a kind of sponge material that will store a lot of Hydrogen in each cubic foot of sponge. There are some materials that do this but not many. And it's tricky. You have to get the Hydrogen to go into the sponge material at a reasonable rate at roughly room temperature and pressure. This is tough. And the Hydrogen needs to leak out of the sponge material without a lot of encouragement so that you can get the Hydrogen back to use in the fuel cell. And it has to be cheap enough to be practical to put into millions of car fuel tanks. So far no one has come up with a magic sponge material that has all these characteristics. I don't see Hydrogen fuel cell cars in any numbers any time soon.
So what are we going to see on the road in the next 20-40 years? I don't see anything replacing the car. So we will see lots of cars. I see a lot of hybrids, some electrics. And a lot of old style gas cars unless gas gets up to $40/gal. I also don't see cars looking a lot different than they do now. One reason to change the shape of cars is aerodynamics. The first aerodynamic car was introduced in the '30s. Wind resistance does not make much difference in the 30-60 MPH speed range. It makes even less difference under 30 MPH. So if you can make the power train more efficient or the car drastically lighter you will get a lot more bang for the buck. I'm sure auto designers will find new and different ways to bend the sheet metal but this will be due more to fashion trends than anything else. We will see more plastic, especially carbon fiber but that won't make much change in the look and feel of cars.
But I do predict a major change in one area. It gets back to congestion and the title of this piece: robot cars. A robot car as anything other than Science Fiction is a pretty recent development. The idea of a practical robot car that you could imagine sharing a street with regular cars is only about 10 years old. But the field is now moving rapidly. The first development to demonstrate this was the DARPA Grand Challenge series of rallies. DARPA, a DOD agency, issued its first challenge in this area for an event that took place in 2004. Driverless cars were to navigate on their own over a 150 mile route on regular roads on a closed course (no other traffic). The best car went 7.3 miles before coming to a stop. Not a very impressive showing.
But oh what a difference a year makes. Round two took place in 2005. The course was similar. But this time 6 out of 15 vehicles finished including one that was a 30,000 lb military vehicle. The winning vehicle was put together by a team from Stanford University. Two years later in 2007 round three was held. This time the course was only 60 miles long but it was over an "urban" route. Vehicles had to obey speed limits, stop for traffic signs, and avoid other "moving hazard" vehicles. Again, 6 teams finished.
So the DARPA challenges resulted in autonomous vehicles that could tell road from not-road, identify stop signs, avoid moving vehicles and perform other basic driving tasks. This was a tremendous accomplishment but could they drive on ordinary roads beside vehicles driven by ordinary people? Other than the last one, the DARPA challenges represented a more sophisticated version of what people had been experimenting for longer. Several demonstrations had been done earlier with robot vehicles that could navigate in a closed "toy" system. The last DARPA challenge introduced a more "real world" environment.
Things have been moving forward rapidly since. Google has been experimenting with a driverless car. Most of the testing has been in closed "toy" environments but not all. The Google car has driven on California freeways. It has even navigated Lombard Street, the famous twisty road in San Francisco. The Google car has not been actually driverless. There has been someone aboard who can take over if necessary. But it has rarely been necessary. The Google car has even been involved in an accident. It was rear ended while stopped at a stop sign. So I have every confidence that we will crack the "robot car" problem from a technical point of view in the next few years.
But the robot car question is also one of those "how do we get there from here" problems. If all cars were robot cars then we can imagine all cars being robot cars but they are not. I think it is idiotic to think that we will have dedicated robot car roads and other non-robot car roads so robot cars need to be able to work in a "real world" environment where there are lots of non-robot cars around. DARPA, Google, and whoever come next are busy proving that it is possible to build robot cars that can do this. So technology will not be the impediment. So we can see the end. But it is necessary to see the intermediate steps too.
I believe the foundation we can build the intermediate steps on is a Collision Avoidance System (CAS). Cars have had crude cruise control systems for many years. These are capable of maintaining a constant speed. But it was the driver's responsibility to do collision avoidance. And the systems were so dumb that you can't even lower the target speed. But we are now seeing much more sophisticate systems coming on line in new cars, especially luxury cars. A simple version of this ties into the cruise control system. It will detect that the vehicle in front is getting too close and alert the driver and disconnect the cruise control. Another improvement is a system that checks what's behind you when you are backing up and alerts you. Another system will parallel park your car. I believe the current version of the parallel park system operates blindly but I can see an upgrade that checks for obstacles and stops. Another possible component is a system that looks for vehicles in your blind spot.
These are the beginning steps. But sensors are getting cheaper and computing power is getting cheaper. Adding more sensors and tying them together to give you a smart cruise control, a "blind spot" (to the side and rear) detection system and other features gives the auto manufacturing companies something to build on. They can market them not only as a differentiator (my CAS has more features than your CAS) but at least some of the features will bring real value to the driver. If you can use your cruise control in heavier traffic and go faster with greater fuel economy, that's worth something. The features that warn you of vehicles in your blind spot, save you from backing into things, and do most of the work in parking your car would all be appealing to me. The equipment that enables all this is the equipment that can be enhanced to provide the robot car capability.
There is one feature that I see that is important to moving things along in the proper direction. GM has an "EN-V" program. The cars themselves are cute and toys, in my opinion. But they do have one feature I see as a good idea. The cars talk to each other. This allows one EN-V to not run into another EN-V. But the system is proprietary. I think it would be a big help if the automakers got together and built a standard for cars talking to each other. The difference in road knowledge possible when a car is on its own versus even the situation where a only few of the cars are exchanging information is tremendous. Imagine a simple situation where one car is following another car. Assume the cars are exchanging information and the rear car is following the front car in "cruise control" mode. The rear car could easily maintain a constant distance because it knows the speed of the front car. Now let's say that he front car needed to brake severely to avoid an obstacle. It could pass the information back to the rear car so that it too could slow and avoid a collision. And in a more mundane case, say the front car was about to exit the freeway. It could signal the following car so it could break off station keeping. There would be fewer situations where the rear car could provide information useful to the front car but there would be some.
Now imagine a situation where most cars were information sharing and had sophisticated robot car capability. Here you could transition to convoying. This would let cars stay closer together and go faster safely. This increases the effective capacity of our current road system thus reducing congestion. And with many sensors in many vehicles the chances of a surprise that might lead to an accident become very small. From here it becomes possible to transition to a true robot car environment. The result would be a cleaner, safer, and more efficient situation than what we have now. Why more efficient? As with bicycles, slowing down and then speeding up uses a lot of energy. If you can save the energy you save the cost of generating the energy. So we get the benefits the car haters desire without getting rid of the convenience benefit car lovers love. A win all around.
Does the car have a future? There are some who would argue that it does not. There is a small but vocal contingent of people in my town that hate cars. They walk, bicycle, or take public transportation. They see cars as evil incarnate. In the most general sense they are arguing against personal transportation, unless it is people powered. So, taking things in reverse order, is people powered transportation practical? People have been walking since there were people. And animals have been walking for a lot longer. So in some sense walking is practical. But during the long period when walking was how most people got around most people never ventured more than 25 miles from where they were born. I for one, do not want to give up the option of venturing further abroad.
Walking also has another disadvantage. You can't carry much along with you. If you intend to walk more than about 10 miles in a day I would estimate that it would be impossible for most people to carry more than about 100 lbs. Many people couldn't handle even this amount. And if you want to go further you need to cut down on your load. If you want to carry more, and the limit would be less than 200 lbs, you would not be able to walk even 10 miles per day. Domestic animals have been around for about 10,000 years as a way to improve the distance/load calculus. People have also been inventing things like ships and wagons as another solution to the problem.
A more recent invention is the bicycle. It is another approach to beating the distance/load calculus. There is an annual "Seattle to Portland" bicycle race in my neck of the woods that demonstrates this. Most participants travel the roughly 200 mile distance in two days. Many of them do it in a day. So a bicycle allows you to travel four to eight times as far as pure foot power would permit. This is a definite improvement. But most people, given the option of trading a bicycle for an automobile, opt for the car. This is most obvious in China. We all have seen video of hordes of bicycles on the streets of Beijing a few years ago. But Chinese are deserting their bikes for cars in the millions. China is now the largest single car market in the world.
Public transportation, typically in the form of buses, but also in the form of light rail, is touted by many as the "correct" alternative to cars when feet or bicycles are not the answer. Why? Well, when you strip the argument down it is efficiency. Public transportation is more efficient and produces less pollution than cars. There is also the gridlock problem. Let's take each of these issues separately.
The theory is that public transportation is more efficient. But is it? If you take a bus and fill it full of people it will be cheaper per passenger mile than the equivalent number of cars, each with only one person in it. That's the way the efficiency argument is usually presented. But are the buses really full? Currently the answer is pretty much yes. But this is because the number of buses is far less than the number that would be needed to meet the demand in a car free environment. Bus systems all lose money and are limited in size to what the taxpayer will support. Tax payer support falls off rapidly as the load (number of people on the typical bus run) decreases. So the current subsidy is only enough to provide for a few pretty full buses. This works because there are lots of cars around to take care of most of the transportation need.
There is also a hidden cost to buses and other mass transportation solutions. That is lost time. One of the real benefits to a car is in it I can go long distances whenever I want. I take a lot of short in-city trips. Frequently I have some flexibility as to when I go so theoretically I could time the trip to fit the bus schedule. But much of the time this is not true. I got my hair cut today. It took me about 10 minutes each way. In a bus, if I timed it right, it might have taken me 20 minutes each way. So right away my travel time doubled. Next, it was to an appointment. So I really needed to travel to the appointment at a specific time. It is possible to go early but whatever time I would have waited between when I arrived and when my appointment started would have been lost time. Even on the most traveled bus routes an "every 20 minutes" schedule is about as good as it gets. So I would have lost another ten minutes in synchronizing with the bus schedule. And all this is true for my trip home too. So in this semi-ideal situation 20 minutes of travel time has ballooned up to 60 minutes. And this is for an in-city to in-city trip.
Buses (or light rail) don't go most suburban places. And they do not go every 20 minutes and they do not run all the time. Let's say we fixed that. Buses would now go everywhere and they run all the time on an "every 20 minutes" schedule. What do things look like now? First, we need a lot of buses, between 10 and 100 times as many buses as we now have. We would also have to use some kind of "hub and spoke" system. It is impractical to have buses running from everywhere to everywhere. So for a lot of trips you would take a local to a hub, a trip to a second hub, and finally a local to your actual destination. That means a 10 minute delay at the first hub, a 10 minute delay at your second hub, and arriving at your destination 10 minutes early. We add an hour to a typical longer round trip. And, in order to meet our "every 20 minutes" and our "goes everywhere" requirement we are going to be running most buses pretty empty a lot of the time and some buses completely empty some of the time. This scenario I have outlined may seem unrealistic but it is exactly what cars provide. I can get in my car whenever I want and go reasonably directly to wherever I want. My route and schedule is completely independent of anyone else's route and schedule. What is your time worth? There is an incredibly large time penalty to shifting most people from traveling by car to traveling by public transportation.
And, once you increase the density of public transportation sufficiently to reduce the time cost to a reasonable amount the efficiency goes out the door. And the efficiency of public transportation is none too good now. There is no public transportation system in existence that recovers all of its costs. If you increase the quality of service of a public transportation system to anything approaching that currently provided by cars it becomes fantastically expensive. Even in a place like New York City with its subway system that was built 100 years ago and its very high density, public transportation is heavily subsidized and they have lots of automobile traffic.
Next, let's consider pollution. Cars are a definite improvement over horses. The release far less pollution per mile. But is a car inherently a polluter? The answer is no! We know this because we can now buy electric cars. The single problem with electric cars is that the current battery technology sucks. The motors that turn the propellers of the aircraft carrier Enterprise are electric. Given this it should be clear that there is absolutely no problem making electric motors that will provide all the performance anyone could want. But current batteries can't store much power. So manufacturers put in wimpy motors to make the batteries last longer so people think electric cars are wimpy. Fix the battery problem and you fix the wimpy problem. Until the crappy battery problem is fixed electric cars are not for everybody.
Given cheap gas and a lack of powerful cheap batteries we will continue to have gas powered cars that pollute. The obvious solution, if we want to reduce the pollution problem, is to make gas more expensive. I drove a big old car while I was in college. It ran on "super". One day I bought super for 29.9 cents/gallon. I said to myself "I will never buy super cheaper in my life". I was right. That was a lot of years ago but it demonstrates what has been happening to gas prices over the last 40 years. Even with the price increases gas is still cheap. But let's assume it gets expensive or we decide for other reasons that we need to make cars much more efficient. What will we do?
The current answer is a hybrid. This is a combination of gas and electric. There are a number of ways to do hybrid and it is not clear how most hybrids work now. In a purely gas car you have an engine, a transmission, some mechanical connections like the differential and this all this spins shafts that turn wheels. This is a pretty inefficient process. There is another way to do this, the way aircraft carriers and diesel locomotives do it. There the primary motor (nuclear or diesel) that is connected to a generator. The electricity is fed to electric motors that spin the wheels. For powerful machines like locomotives and aircraft carriers, this is the most efficient way to do it. It seems to me that this should be the most efficient way to do cars. You would also throw some batteries and a more complex "control" system in between the generator and the motors on the wheels. This approach has many advantages. You can put more or less batteries in. You can add in the capability of charging the batteries from the electric grid. More batteries and charging allows you to run the car on electricity more of the time. The motor generator approach allows you to make the motor more efficient because it doesn't have to run at different speeds and loads. It is no longer connected to the wheels. You get rid of the transmission and other mechanical equipment that is inefficient at transferring power along the line to the wheels. This general approach should result in the most efficient car. So why don't all cars do it this way?
It turns out that if you do the same thing a lot of times for a long time you get very efficient at it. Auto makers have been making a lot of traditional cars for a long time. They have gotten very efficient at it. Going to the design I recommend means learning how to do a lot of new things very efficiently. It may be cheaper to do the theoretically less efficient thing (a car with a lot of old technology) because you are so darn good at doing the old thing. Eventually some auto makers will figure out how to be good at making the new kinds of cars. They will force the rest to figure it out too. It might take a while but probably less than 10 years.
Moving from our current "gas" car to hybrid cars is hard but doable. Most of our infrastructure; most of the manufacturing process, roads, gas stations, etc., will require little or no modification. Switching to all electric cars would require a much greater change. Batteries are the critical problem. I have been following battery technology for 30 years. The newest batteries are better than the old ones but we have merely moved from appalling to awful. We need to move all the way to good. Then there is our electricity grid. It is not set up to handle the load that moving our transportation system over to a high percentage of electric would require. Part of it is just more as in more generating capacity and more transmission lines. We know how to do these things. The part we don't know how to do is storage. Our electric grid is real time. It has a little inertia built into it but mostly it generates and distributes what is needed now. Any serious imbalance between supply and demand results in outages. And wind, solar, and some other sources are intermittent. It would be nice if we could store the excess from some periods to cover the shortages from other periods. Being able to store large amounts of power for a few days or even a few hours would make a tremendous difference. This is the wholesale version of the electric car battery problem. Again, I have been watching this area for over 30 years and not much progress has been made. Nuff said. Back to cars.
There is another "clean car" idea out there. That's fuel cell/hydrogen. The idea is to use a fuel cell to turn hydrogen into electricity. NASA has been doing this since the '60s so its something we know how to do. The problem is not in the fuel cell it is in the Hydrogen. If you are NASA sending a space probe to the back side of beyond, the many problems associated with dealing with Hydrogen are worth the hassle. But this is not true here on Earth. Hydrogen has two problems: making it and storing it. There is no Hydrogen loose around to collect. You have to make it. Hydrogen is a constituent of lots of things including water. But Hydrogen really likes to combine with stuff. So it does. To make it into a fuel we have to uncombine it. That takes lots of energy. Well, the whole point of Hydrogen is as a source of energy so this whole thing about consuming a lot of energy to create Hydrogen is just wrong. A simple way to make Hydrogen is to use electricity to separate out the Hydrogen in water. Other than the fantastic amounts of energy this requires it works pretty well. And that's the problem with the many alternatives. You end up having to use a lot of energy to make the Hydrogen. This is not good. Then when you have made the Hydrogen you have made something that loves to recombine with other stuff. The name for this "combine" process, in many cases, is explosion. So you need to be very careful how you handle things or you get explosions or perhaps just a very large very hot fire.
So Hydrogen is dangerous to store and you have to be very careful. That sounds like the "storage" problem but it's not. To store gas in a car you make this thing that is a couple of cubic feet in size out of sheet metal called a gas tank. It's not very big and it's not very heavy, even full of gas. But in this not very big not very heavy thing you can put enough gas with enough energy to move a big SUV 300 miles. To store the same amount of energy as Hydrogen under similar conditions you would need a tank many times the size of the SUV. So you have to do something. One thing is to compress it. But you have to compress it a lot. So you need a very strong tank and you need to transfer the Hydrogen from the gas station to the car under these very high pressures. This is dangerous and expensive. And probably the gas tank is now very heavy in order to be strong enough.
Another approach is a sponge. It turns out that there are certain materials that wick up Hydrogen. They do it so well that you can get a lot of cubic feet of Hydrogen into a few cubic feet of tank. And the pressure is not very high. It's some kind of chemistry magic but it works. Effectively you get a lot of compression without a lot of pressure. You need a kind of sponge material that will store a lot of Hydrogen in each cubic foot of sponge. There are some materials that do this but not many. And it's tricky. You have to get the Hydrogen to go into the sponge material at a reasonable rate at roughly room temperature and pressure. This is tough. And the Hydrogen needs to leak out of the sponge material without a lot of encouragement so that you can get the Hydrogen back to use in the fuel cell. And it has to be cheap enough to be practical to put into millions of car fuel tanks. So far no one has come up with a magic sponge material that has all these characteristics. I don't see Hydrogen fuel cell cars in any numbers any time soon.
So what are we going to see on the road in the next 20-40 years? I don't see anything replacing the car. So we will see lots of cars. I see a lot of hybrids, some electrics. And a lot of old style gas cars unless gas gets up to $40/gal. I also don't see cars looking a lot different than they do now. One reason to change the shape of cars is aerodynamics. The first aerodynamic car was introduced in the '30s. Wind resistance does not make much difference in the 30-60 MPH speed range. It makes even less difference under 30 MPH. So if you can make the power train more efficient or the car drastically lighter you will get a lot more bang for the buck. I'm sure auto designers will find new and different ways to bend the sheet metal but this will be due more to fashion trends than anything else. We will see more plastic, especially carbon fiber but that won't make much change in the look and feel of cars.
But I do predict a major change in one area. It gets back to congestion and the title of this piece: robot cars. A robot car as anything other than Science Fiction is a pretty recent development. The idea of a practical robot car that you could imagine sharing a street with regular cars is only about 10 years old. But the field is now moving rapidly. The first development to demonstrate this was the DARPA Grand Challenge series of rallies. DARPA, a DOD agency, issued its first challenge in this area for an event that took place in 2004. Driverless cars were to navigate on their own over a 150 mile route on regular roads on a closed course (no other traffic). The best car went 7.3 miles before coming to a stop. Not a very impressive showing.
But oh what a difference a year makes. Round two took place in 2005. The course was similar. But this time 6 out of 15 vehicles finished including one that was a 30,000 lb military vehicle. The winning vehicle was put together by a team from Stanford University. Two years later in 2007 round three was held. This time the course was only 60 miles long but it was over an "urban" route. Vehicles had to obey speed limits, stop for traffic signs, and avoid other "moving hazard" vehicles. Again, 6 teams finished.
So the DARPA challenges resulted in autonomous vehicles that could tell road from not-road, identify stop signs, avoid moving vehicles and perform other basic driving tasks. This was a tremendous accomplishment but could they drive on ordinary roads beside vehicles driven by ordinary people? Other than the last one, the DARPA challenges represented a more sophisticated version of what people had been experimenting for longer. Several demonstrations had been done earlier with robot vehicles that could navigate in a closed "toy" system. The last DARPA challenge introduced a more "real world" environment.
Things have been moving forward rapidly since. Google has been experimenting with a driverless car. Most of the testing has been in closed "toy" environments but not all. The Google car has driven on California freeways. It has even navigated Lombard Street, the famous twisty road in San Francisco. The Google car has not been actually driverless. There has been someone aboard who can take over if necessary. But it has rarely been necessary. The Google car has even been involved in an accident. It was rear ended while stopped at a stop sign. So I have every confidence that we will crack the "robot car" problem from a technical point of view in the next few years.
But the robot car question is also one of those "how do we get there from here" problems. If all cars were robot cars then we can imagine all cars being robot cars but they are not. I think it is idiotic to think that we will have dedicated robot car roads and other non-robot car roads so robot cars need to be able to work in a "real world" environment where there are lots of non-robot cars around. DARPA, Google, and whoever come next are busy proving that it is possible to build robot cars that can do this. So technology will not be the impediment. So we can see the end. But it is necessary to see the intermediate steps too.
I believe the foundation we can build the intermediate steps on is a Collision Avoidance System (CAS). Cars have had crude cruise control systems for many years. These are capable of maintaining a constant speed. But it was the driver's responsibility to do collision avoidance. And the systems were so dumb that you can't even lower the target speed. But we are now seeing much more sophisticate systems coming on line in new cars, especially luxury cars. A simple version of this ties into the cruise control system. It will detect that the vehicle in front is getting too close and alert the driver and disconnect the cruise control. Another improvement is a system that checks what's behind you when you are backing up and alerts you. Another system will parallel park your car. I believe the current version of the parallel park system operates blindly but I can see an upgrade that checks for obstacles and stops. Another possible component is a system that looks for vehicles in your blind spot.
These are the beginning steps. But sensors are getting cheaper and computing power is getting cheaper. Adding more sensors and tying them together to give you a smart cruise control, a "blind spot" (to the side and rear) detection system and other features gives the auto manufacturing companies something to build on. They can market them not only as a differentiator (my CAS has more features than your CAS) but at least some of the features will bring real value to the driver. If you can use your cruise control in heavier traffic and go faster with greater fuel economy, that's worth something. The features that warn you of vehicles in your blind spot, save you from backing into things, and do most of the work in parking your car would all be appealing to me. The equipment that enables all this is the equipment that can be enhanced to provide the robot car capability.
There is one feature that I see that is important to moving things along in the proper direction. GM has an "EN-V" program. The cars themselves are cute and toys, in my opinion. But they do have one feature I see as a good idea. The cars talk to each other. This allows one EN-V to not run into another EN-V. But the system is proprietary. I think it would be a big help if the automakers got together and built a standard for cars talking to each other. The difference in road knowledge possible when a car is on its own versus even the situation where a only few of the cars are exchanging information is tremendous. Imagine a simple situation where one car is following another car. Assume the cars are exchanging information and the rear car is following the front car in "cruise control" mode. The rear car could easily maintain a constant distance because it knows the speed of the front car. Now let's say that he front car needed to brake severely to avoid an obstacle. It could pass the information back to the rear car so that it too could slow and avoid a collision. And in a more mundane case, say the front car was about to exit the freeway. It could signal the following car so it could break off station keeping. There would be fewer situations where the rear car could provide information useful to the front car but there would be some.
Now imagine a situation where most cars were information sharing and had sophisticated robot car capability. Here you could transition to convoying. This would let cars stay closer together and go faster safely. This increases the effective capacity of our current road system thus reducing congestion. And with many sensors in many vehicles the chances of a surprise that might lead to an accident become very small. From here it becomes possible to transition to a true robot car environment. The result would be a cleaner, safer, and more efficient situation than what we have now. Why more efficient? As with bicycles, slowing down and then speeding up uses a lot of energy. If you can save the energy you save the cost of generating the energy. So we get the benefits the car haters desire without getting rid of the convenience benefit car lovers love. A win all around.
Saturday, February 19, 2011
Artificial Intelligence
In 1950 Alan Turing, a noted mathematician and cryptographer, published a paper in which he described the "Turing Test". In its modern form imagine texting to a stranger. You can send any kind of text message you want. The stranger can reply or not as and how he chooses. After some time (days, perhaps a week) you are asked to answer a simple question: Is the stranger a person or a machine? The Turing Test is designed to answer an important question: Is it possible to create an intelligent machine? If a machine can pass the Touring Test by successfully impersonating a human then it is possible to create an intelligent machine, according to Turing.
Since then Computer Scientists and many others have been fascinated by the Turing Test. Individuals and groups have set up actual Touring Tests but it turns out to be tougher than you would think to get right. What would happen, for instance, if the stranger was a human who tried to imitate a machine? Is this really fair? Also, certain limitations are usually necessary because machines that can look, sound, and act like humans only exist in the realm of fiction. Hence, the text message scenario.
But the concept behind the Touring Test has continued to fascinate people because Turing had a profound idea. If a machine can do things that intelligent humans do then it must be intelligent. This approach seems like a very intuitive and natural approach to figuring out what we mean by "intelligent". And we have just witnessed a very public event. The TV show "Jeopardy" hosted a three day two game exhibition match between a computer (actually a network of IBM computers) and two champion winners (Ken Jennings, winner of 74 matches in a row, and Brad Rutter, the all time money champ). We know a lot of people are fascinated because the match bumped Jeopardy's ratings up 30%, according to Nielson.
Technically, it wasn't a Touring Test because we all knew that Watson (the name IBM gave to their computer system) was a machine. But the question lots of people including yours truly were asking themselves was whether Watson was able to play like a human. The answer we got, if we just go by results. was that Watson was better during these two games than his human competition. And, given the quality of the players he was up against, we can say that he was far better than the typical Jeopardy contestant and, therefore, far far better than the rest of us.
As I have pointed out, staging a Touring Test, even a fake one like the Jeopardy exhibition, is a lot tougher than it looks, if one of your objectives is fairness. IBM staged a number of demonstration matches in order to convince the Jeopardy producers that the tournament would be a good idea. In these demonstration matches it was obvious that Watson could be led astray by constructing "questions" properly. (I know about the "response in the form of a question" gimmick. It works well on the show but I am not going to bother in this article). On the other hand, electronics are far faster at reflex activities like pressing a button. So a lot of work was put into providing what both sides saw as a level playing field. "Normal" Jeopardy questions on one side versus making Watson actually push a button on the other side. And both sides wanted to create an entertaining result so both sides wanted the human participants to have a chance and for Watson to not look like a comedy punch line. And they succeeded. It was fun to watch.
So how did it all come out? Well, Watson showed some weak spots but generally won handily. The final score was Watson - $77,147, Jennings - $24,000, and Rutter - $21,600. Watson was also way ahead at the end of the first match. And Watson was pretty good at figuring out whether he knew the answer or not. We got to see Watson's top three possible answers for most questions. The top answer was coded green for confident, red for not confident, and yellow for somewhere in between. Most of the time Watson's answer was green and another player rang in first when Watson coded his best answer red. Watson also rang in first about two thirds of the time. But the tale is told by the answers Watson got wrong, especially the ones he got really wrong.
We can decide Watson is a machine by the sheer speed and breadth of his performance. But if the IBM people did not think he was quick and knew lots of stuff the exhibition would never have taken place. So that's not enough. We all know that modern computers can organize a lot of data. But, while computers are very good at dealing with "structured" data, say where you have a table with rows and columns, computers are poor at dealing with unstructured data. Put simply, computers can't read.
Oh, they can scan text. Then they can identify all the letters and assemble the letters into words by taking advantage of spaces and other punctuation. But it is very hard for computers to take the next step and understand what the words mean. Most of the truly massive amount of data that was loaded into Watson was in the form of long sequences of text. All of Wikipedia was loaded into Watson. Wikipedia consists of over 2 million articles. And each article consists mostly of standard text because that's what people are good at using. The Internet Movie Database was also loaded in, along with a truly astounding number of other references. A lot of the IMDB data is structured. You have the movie name at the top. Then you have a section that lists each actor and role, one to a line, and so on. If you take a hard look at IMDB you will find out that it is not that easy but, for IMDB it seems like you at least have a chance of sorting much of it out. But for Wikipedia and most of what was loaded into Watson it is a lot harder.
I might find a sentence "Adam begat Cain". This tells a person that you have a parent child relationship where Adam is the father and Cain is the son. And to completely nail it down, you have to know that Adam and Cain are both male names. But what about "A boy named Sue", the popular Johnny Cash song. While Sue is normally a female name, in this case Sue is a male. A friend had a dog named Sam, short for Samantha. Sam is usually the name of a male person. And I could come up with even tougher examples where it is hard for a machine to make sense of things. There is ambiguity. There are contradictions. People are pretty good at functioning in this kind of messy environment but computers aren't.
So how did Watson do? Watson came up with the correct answer in a truly astounding number of areas. So whatever the IBM people did did to collect and organize Watson's data, it worked pretty well. Most of the time Watson came up with up with a green answer that was correct. In one case Watson came up with a yellow rated answer of "Serbia" when the correct answer was "Slovenia". I wouldn't have known which answer was the correct one. So I score Watson high for getting close and knowing when he wasn't sure. I don't know what process the IBM people to assemble the database. The advantage they had was that this could all be done "offline" before the exhibition started. And it might bee that they cheated by using statistical techniques like "this word is frequently found near that word". But if they were able to do the linguistic analysis necessary to get from "Adam begat Cain" to "Adam is the father of Cain", "Cain is the son of Adam", etc., in other words do linguistic and other analysis to turn strings of text into usable information, that would be a truly useful feat.
Jeopardy also poses quite a challenge in the structure of the clues. They are not standard English. There are frequently puns and other tricks. These are hard for people type contestants to deal with, especially with the time constraint. But they are much more difficult for a computer to deal with. And in this case, you can't use statistical tricks. It won't yield enough information because Jeopardy clues are very tightly packed and are frequently constructed so that the components are ambiguous and you have to combine all the components to narrow things down to one answer.
On several occasions Watson went astray by not figuring out an attribute that the correct answer needed to possess. For instance, in one case Watson gave a green rated answer of "Picasso", which was wrong. The clue was asking for a painting style not the name of a painter. The correct answer was "Modern Art". This would have been completely obvious to a human. In another case Watson was unable to correctly process the category. It was a tricky one, keys found on a computer keyboard. For instance, one answer that Watson did not ring in on was "F1". But in another case Watson rang in and supplied a green rated answer of "Chemise". There is no "Chemise" key on a computer keyboard but there is a "Shift" key. The clue had to do with clothing styles. I might or might not have come up with the correct answer but I would definitely have known that "Chemise" was wrong. Had Watson gotten the category, the questions would have been a piece of cake for him. Watson also answered "Dorothy Parker", an author, when what was required was "Elements of Style", the title of a book. I believe this was on a Daily Double and Watson did correctly rate his answer as a red.
So tricky categories and clues were a disadvantage to Watson. But an area he should have had a decided advantage was with ringing in. One would expect that Watson would let red answers go but would always ring in first when he had a Green answer. But in 11 cases Watson with a green answer was beat by one of the human players. I don't know what the story was here. Ken Jennings did say somewhere that it is possible to do an "anticipatory" ring in. You try to figure out when Alex is going to finish the "question" and ring in an instant after he should finish. If you ring in early your button is locked out for a while so there is a high penalty for ringing in early. Successful contestants try to figure out the answer while Alex is still talking. Watson got the questions in text form as soon as it was displayed on the board and employed the same strategy. So, for green answers, Watson should always ring in the same way.
Without knowing a lot more of the details (see - I told you this is tricky) it is impossible to give Watson a grade on the Jeopardy version of the Touring Test. But, ignoring the obvious, like the presence of an Avatar rather than a real person, and assuming the whole undertaking was "legit", for the most part Watson did very well. He was able to make the proper sense of most of the clues most of the time. And, besides the Avatar there was another dead give away. In one case Watson gave the same answer as another contestant who had gotten it wrong. Watson was deaf and they did not feed any of the audio to him. Again, ignoring the obvious (Avatar, deafness) that fact that Watson made at least one bonehead mistake means that technically he failed. But it's still a stunning achievement.
So, beyond the Touring Test aspect, what does this all mean? I was in school in the early '70s and it was one of those times when Artificial Intelligence (AI) research was on the rise. A number of "proof of concept" projects had generated a lot of buzz inside the Computer Science community. The story was "just let us crank these up a bit and see what we can really do". But none of these projects was able to progress past "proof of concept" into something more general and more powerful. After watching for a couple of years I decided that real progress in AI was a long way away and that AI was really hard to do. Unfortunately, my observations turned out to be spot on. There hasn't been a lot of headline progress on AI since. But Watson proves that real progress has been made.
The issue I discussed above of turning text into data was completely beyond the state of the art in the '70s for anything but toy environments. If, as it seems, the Watson project has been able to process astounding amounts of raw text data and turn it into real information that can be used by computers, that is tremendous and real progress. There is another aspect of the Watson project that I want to discuss next. That's machine learning.
The original approach to AI was to put in a bunch of rules. Things like "normal temperature for a human is 97 - 99 degrees Fahrenheit", that sort of stuff. You put in a bunch of rules and the computer used them to answer your question. The theory was that if you had enough rules and they were good rules, you could get a good result. That approach eventually petered out. The modern approach to the same problem is to put some kind of general structure and analytic capability into the system. They you give the system a bunch of right and wrong examples Including whether each example is right or wrong, called a "training set". In our case they fed in a bunch of right and wrong "answers and questions". Then the idea is that the system does some kind of analysis of the "training set" and figures out its own rules. Computer Science people have been playing around with this approach for many years now. And its the approach the IBM people used. Based on their results, I would have to say that the state of the art in machine learning is now pretty good. And that's good news too.
So where do we go from here? Years ago a "physician assistant" was developed. Modern medicine is (and was) unbelievably complicated. The idea was to provide some computer assistance to help medical people diagnose tough cases. The program ultimately went nowhere. But this project seems like a perfect fit for Watson's capabilities. Pump in a lot of medical data, most typically found in text form. This medical data will be chock full of ambiguities and contradictions, just as the Jeopardy database was. Watson has tremendous English language capabilities and "medical language" should be no harder to master than "Jeopardy language". Finally this "training set" approach to machine learning should work as well for medicine as it did for Jeopardy. So it sounds like this would be a good application for Watson technology.
This has already occurred to IBM. They are starting work on a medicine version of Watson. Another area they have identified is the law. Again you have vast amounts of text, in this case legalese, that is ambiguous and contradictory. But again the same types of abilities that made Jeopardy Watson successful look like a good fit. Certainly if one or both of these projects are successful we can expect IBM (and eventually others) to come up with other applications.
Finally, can we look forward to a rematch? I don't think so. I have seen some of the test runs IBM did before the Jeopardy producers decided Watson was ready for prime time. The goofs were much more frequent and much more embarrassing. Yet the progress from there to what we saw in prime time only a few months later was truly astounding. Given even a few more months I am sure the Watson team could fix the few goofs we saw and many more to the point that human contestants would go from having little chance to having no chance at all. The only reason I can think of that I might have misjudged the situation would be if the books were cooked is some non obvious way. Since I don't think the books were cooked I think this Jeopardy challenge will be a one time event.
Since then Computer Scientists and many others have been fascinated by the Turing Test. Individuals and groups have set up actual Touring Tests but it turns out to be tougher than you would think to get right. What would happen, for instance, if the stranger was a human who tried to imitate a machine? Is this really fair? Also, certain limitations are usually necessary because machines that can look, sound, and act like humans only exist in the realm of fiction. Hence, the text message scenario.
But the concept behind the Touring Test has continued to fascinate people because Turing had a profound idea. If a machine can do things that intelligent humans do then it must be intelligent. This approach seems like a very intuitive and natural approach to figuring out what we mean by "intelligent". And we have just witnessed a very public event. The TV show "Jeopardy" hosted a three day two game exhibition match between a computer (actually a network of IBM computers) and two champion winners (Ken Jennings, winner of 74 matches in a row, and Brad Rutter, the all time money champ). We know a lot of people are fascinated because the match bumped Jeopardy's ratings up 30%, according to Nielson.
Technically, it wasn't a Touring Test because we all knew that Watson (the name IBM gave to their computer system) was a machine. But the question lots of people including yours truly were asking themselves was whether Watson was able to play like a human. The answer we got, if we just go by results. was that Watson was better during these two games than his human competition. And, given the quality of the players he was up against, we can say that he was far better than the typical Jeopardy contestant and, therefore, far far better than the rest of us.
As I have pointed out, staging a Touring Test, even a fake one like the Jeopardy exhibition, is a lot tougher than it looks, if one of your objectives is fairness. IBM staged a number of demonstration matches in order to convince the Jeopardy producers that the tournament would be a good idea. In these demonstration matches it was obvious that Watson could be led astray by constructing "questions" properly. (I know about the "response in the form of a question" gimmick. It works well on the show but I am not going to bother in this article). On the other hand, electronics are far faster at reflex activities like pressing a button. So a lot of work was put into providing what both sides saw as a level playing field. "Normal" Jeopardy questions on one side versus making Watson actually push a button on the other side. And both sides wanted to create an entertaining result so both sides wanted the human participants to have a chance and for Watson to not look like a comedy punch line. And they succeeded. It was fun to watch.
So how did it all come out? Well, Watson showed some weak spots but generally won handily. The final score was Watson - $77,147, Jennings - $24,000, and Rutter - $21,600. Watson was also way ahead at the end of the first match. And Watson was pretty good at figuring out whether he knew the answer or not. We got to see Watson's top three possible answers for most questions. The top answer was coded green for confident, red for not confident, and yellow for somewhere in between. Most of the time Watson's answer was green and another player rang in first when Watson coded his best answer red. Watson also rang in first about two thirds of the time. But the tale is told by the answers Watson got wrong, especially the ones he got really wrong.
We can decide Watson is a machine by the sheer speed and breadth of his performance. But if the IBM people did not think he was quick and knew lots of stuff the exhibition would never have taken place. So that's not enough. We all know that modern computers can organize a lot of data. But, while computers are very good at dealing with "structured" data, say where you have a table with rows and columns, computers are poor at dealing with unstructured data. Put simply, computers can't read.
Oh, they can scan text. Then they can identify all the letters and assemble the letters into words by taking advantage of spaces and other punctuation. But it is very hard for computers to take the next step and understand what the words mean. Most of the truly massive amount of data that was loaded into Watson was in the form of long sequences of text. All of Wikipedia was loaded into Watson. Wikipedia consists of over 2 million articles. And each article consists mostly of standard text because that's what people are good at using. The Internet Movie Database was also loaded in, along with a truly astounding number of other references. A lot of the IMDB data is structured. You have the movie name at the top. Then you have a section that lists each actor and role, one to a line, and so on. If you take a hard look at IMDB you will find out that it is not that easy but, for IMDB it seems like you at least have a chance of sorting much of it out. But for Wikipedia and most of what was loaded into Watson it is a lot harder.
I might find a sentence "Adam begat Cain". This tells a person that you have a parent child relationship where Adam is the father and Cain is the son. And to completely nail it down, you have to know that Adam and Cain are both male names. But what about "A boy named Sue", the popular Johnny Cash song. While Sue is normally a female name, in this case Sue is a male. A friend had a dog named Sam, short for Samantha. Sam is usually the name of a male person. And I could come up with even tougher examples where it is hard for a machine to make sense of things. There is ambiguity. There are contradictions. People are pretty good at functioning in this kind of messy environment but computers aren't.
So how did Watson do? Watson came up with the correct answer in a truly astounding number of areas. So whatever the IBM people did did to collect and organize Watson's data, it worked pretty well. Most of the time Watson came up with up with a green answer that was correct. In one case Watson came up with a yellow rated answer of "Serbia" when the correct answer was "Slovenia". I wouldn't have known which answer was the correct one. So I score Watson high for getting close and knowing when he wasn't sure. I don't know what process the IBM people to assemble the database. The advantage they had was that this could all be done "offline" before the exhibition started. And it might bee that they cheated by using statistical techniques like "this word is frequently found near that word". But if they were able to do the linguistic analysis necessary to get from "Adam begat Cain" to "Adam is the father of Cain", "Cain is the son of Adam", etc., in other words do linguistic and other analysis to turn strings of text into usable information, that would be a truly useful feat.
Jeopardy also poses quite a challenge in the structure of the clues. They are not standard English. There are frequently puns and other tricks. These are hard for people type contestants to deal with, especially with the time constraint. But they are much more difficult for a computer to deal with. And in this case, you can't use statistical tricks. It won't yield enough information because Jeopardy clues are very tightly packed and are frequently constructed so that the components are ambiguous and you have to combine all the components to narrow things down to one answer.
On several occasions Watson went astray by not figuring out an attribute that the correct answer needed to possess. For instance, in one case Watson gave a green rated answer of "Picasso", which was wrong. The clue was asking for a painting style not the name of a painter. The correct answer was "Modern Art". This would have been completely obvious to a human. In another case Watson was unable to correctly process the category. It was a tricky one, keys found on a computer keyboard. For instance, one answer that Watson did not ring in on was "F1". But in another case Watson rang in and supplied a green rated answer of "Chemise". There is no "Chemise" key on a computer keyboard but there is a "Shift" key. The clue had to do with clothing styles. I might or might not have come up with the correct answer but I would definitely have known that "Chemise" was wrong. Had Watson gotten the category, the questions would have been a piece of cake for him. Watson also answered "Dorothy Parker", an author, when what was required was "Elements of Style", the title of a book. I believe this was on a Daily Double and Watson did correctly rate his answer as a red.
So tricky categories and clues were a disadvantage to Watson. But an area he should have had a decided advantage was with ringing in. One would expect that Watson would let red answers go but would always ring in first when he had a Green answer. But in 11 cases Watson with a green answer was beat by one of the human players. I don't know what the story was here. Ken Jennings did say somewhere that it is possible to do an "anticipatory" ring in. You try to figure out when Alex is going to finish the "question" and ring in an instant after he should finish. If you ring in early your button is locked out for a while so there is a high penalty for ringing in early. Successful contestants try to figure out the answer while Alex is still talking. Watson got the questions in text form as soon as it was displayed on the board and employed the same strategy. So, for green answers, Watson should always ring in the same way.
Without knowing a lot more of the details (see - I told you this is tricky) it is impossible to give Watson a grade on the Jeopardy version of the Touring Test. But, ignoring the obvious, like the presence of an Avatar rather than a real person, and assuming the whole undertaking was "legit", for the most part Watson did very well. He was able to make the proper sense of most of the clues most of the time. And, besides the Avatar there was another dead give away. In one case Watson gave the same answer as another contestant who had gotten it wrong. Watson was deaf and they did not feed any of the audio to him. Again, ignoring the obvious (Avatar, deafness) that fact that Watson made at least one bonehead mistake means that technically he failed. But it's still a stunning achievement.
So, beyond the Touring Test aspect, what does this all mean? I was in school in the early '70s and it was one of those times when Artificial Intelligence (AI) research was on the rise. A number of "proof of concept" projects had generated a lot of buzz inside the Computer Science community. The story was "just let us crank these up a bit and see what we can really do". But none of these projects was able to progress past "proof of concept" into something more general and more powerful. After watching for a couple of years I decided that real progress in AI was a long way away and that AI was really hard to do. Unfortunately, my observations turned out to be spot on. There hasn't been a lot of headline progress on AI since. But Watson proves that real progress has been made.
The issue I discussed above of turning text into data was completely beyond the state of the art in the '70s for anything but toy environments. If, as it seems, the Watson project has been able to process astounding amounts of raw text data and turn it into real information that can be used by computers, that is tremendous and real progress. There is another aspect of the Watson project that I want to discuss next. That's machine learning.
The original approach to AI was to put in a bunch of rules. Things like "normal temperature for a human is 97 - 99 degrees Fahrenheit", that sort of stuff. You put in a bunch of rules and the computer used them to answer your question. The theory was that if you had enough rules and they were good rules, you could get a good result. That approach eventually petered out. The modern approach to the same problem is to put some kind of general structure and analytic capability into the system. They you give the system a bunch of right and wrong examples Including whether each example is right or wrong, called a "training set". In our case they fed in a bunch of right and wrong "answers and questions". Then the idea is that the system does some kind of analysis of the "training set" and figures out its own rules. Computer Science people have been playing around with this approach for many years now. And its the approach the IBM people used. Based on their results, I would have to say that the state of the art in machine learning is now pretty good. And that's good news too.
So where do we go from here? Years ago a "physician assistant" was developed. Modern medicine is (and was) unbelievably complicated. The idea was to provide some computer assistance to help medical people diagnose tough cases. The program ultimately went nowhere. But this project seems like a perfect fit for Watson's capabilities. Pump in a lot of medical data, most typically found in text form. This medical data will be chock full of ambiguities and contradictions, just as the Jeopardy database was. Watson has tremendous English language capabilities and "medical language" should be no harder to master than "Jeopardy language". Finally this "training set" approach to machine learning should work as well for medicine as it did for Jeopardy. So it sounds like this would be a good application for Watson technology.
This has already occurred to IBM. They are starting work on a medicine version of Watson. Another area they have identified is the law. Again you have vast amounts of text, in this case legalese, that is ambiguous and contradictory. But again the same types of abilities that made Jeopardy Watson successful look like a good fit. Certainly if one or both of these projects are successful we can expect IBM (and eventually others) to come up with other applications.
Finally, can we look forward to a rematch? I don't think so. I have seen some of the test runs IBM did before the Jeopardy producers decided Watson was ready for prime time. The goofs were much more frequent and much more embarrassing. Yet the progress from there to what we saw in prime time only a few months later was truly astounding. Given even a few more months I am sure the Watson team could fix the few goofs we saw and many more to the point that human contestants would go from having little chance to having no chance at all. The only reason I can think of that I might have misjudged the situation would be if the books were cooked is some non obvious way. Since I don't think the books were cooked I think this Jeopardy challenge will be a one time event.
Wednesday, February 16, 2011
Defeating Creationism
An article “Defeating Creationism in the Courtroom, But Not in the Classroom” was published in the January 28 issue of Science, the official magazine of the American Association for the Advancement of Science. I characterize it as the latest in a long series of “what we are currently doing is not working” stories. The authors suggest as a solution a minor tweak in how we educate Biology teachers. Does anyone really believe that this is going to turn the tide? I don’t. It is wholly inadequate. I advocate a much more drastic change. Creationists want to “teach the controversy”. I think we should go even further and teach Biblical Creationism (e.g. Genesis) in the science classroom. Am I a Creationist? No! Am I nuts? Also, no. Hear me out.
The standard argument (cleaned up and clarified) propounded by Creationists is:
There are only two possible explanations Evolution and Biblical Creationism.
Since Evolution is flawed it must be false and Biblical Creationism must be true.
My point is that Biblical Creationism is never examined. Most Creationists are ignorant about Evolution. But if you dig a little deeper you will find out that they are also ignorant about Biblical Creationism. Almost everybody assumes that everybody knows what Biblical Creationism is. But almost everybody and especially almost all Creationists are wrong. In fact, everyone in general and Creationists in particular do NOT know what Biblical Creationism is. Sure they know that it is the first part of Genesis in the Bible. And they know a few phrases like “let there be light” but that’s about it. So everyone assumes that Biblical Creationism is a well understood theory. And Creationists in particular believe there is little reason to doubt its efficacy. The truth is actually the opposite.
So my proposal is to teach Biblical Creationism and then demolish it. It is easily demolished but little effort has been invested in demolishing it for the last couple of hundred years. I am a poor researcher and terribly lazy. So I suggest that someone who is a better researcher go back and look for the problems that led earlier scientists to find Genesis problematic, search for alternatives, and eventually adopt Evolution. In the mean time let me show you just how easy it is to find problems with Biblical Creationism. Genesis 1, verse 5 says “And the evening and the morning were the first day”. There is similar language for the other days. This language implies the existence of a “light and dark” day and night cycle. For this to happen you need the Sun. But God does not create the Sun until day 4 (“And God made two great lights, the greater one to rule the day” – Genesis 1, verse 16). So it is impossible to have days and nights until the end of day 4. Another problem with Genesis has to do with when mankind gets created. The first possible answer is on day 6 (“And God said let us make man” – Genesis 1, verse 26). But the whole “Adam and Eve” story is later in Genesis after God has rested on the seventh day. So when is it: on day six or after day seven? These two examples show the kinds of problems that can easily be found with "Genesis".
It is not necessary to demolish Biblical Creationism in the classroom, although I think it would be a good idea if it could be done. But it is necessary to demolish it loudly and publicly where Creationists can’t ignore it. I think the most important reason Evolution has been losing and Creationism has been winning in the forum of public opinion is that there has been no public effort to damage Biblical Creationism as a viable theory. So the battle becomes between a flawed (if you believe the Creationists) theory of Evolution and a flawless (it must be flawless, no one has attacked it directly) theory of Biblical Creationism. It would be nice to think that “everyone knows that Biblical Creationism is seriously flawed” but I think most people in fact don’t know.
A final thought. I think it is important to develop a rendition of modern scientific thought on the Genesis story in Genesis language. Here’s what I mean by “Genesis language”. If God made the world as Creationists believe, I think we would all agree he was a very smart guy. Nevertheless, he had a problem when composing Genesis, (the Bible is after all “the word of God”). He was talking to a group of people with a limited frame of reference for understanding the world. He had to stick with language and concepts that his audience could understand. So he would be forced to provide a simplified version of what happened. But even with this limitation he could have done a much better job of getting things right. Presenting an accurate story of what we now know happened would necessarily be simplistic. But it is not hard to avoid many of the problems found in Genesis while sticking to concepts people of that time and place could understand. If the scientific community develops a correct Genesis-like story to place beside the one that is actually in the Bible it will demolish many of the arguments Biblical Creationists will put forward to rebut scientific criticisms of Biblical Creationism.
Thursday, January 27, 2011
State of the Union
A few days ago President Obama gave his annual "State of the Union" message. It was well received by the public and by me. This is on top of his excellent speech after the Assassination attempt on Rep. Giffords in Tuscon. And, as I have indicated elsewhere (http://sigma5.blogspot.com/2010/12/negotiation-101.html), President Obama is on something of a roll. This is all good news. But it is too soon to tell whether the President and his team have learned the right lessons. I am still skeptical.
The President laid out a very modest agenda in his State of the Union speech. All things being equal I would have preferred a lot more. But all things are not equal so I am happy with the modest proposals he laid out. But I have been here before. President Obama is noted for his speaking ability. He has given many notable and some great speeches. The problem has been the follow through.
The narrative line from the DC pundits is that he concentrated too much on getting things done and not enough on communication. There is a lot of truth to this. Of course, the Republicans are past masters at throwing up noise ("Death Panels", "He is a Socialist who was born in another country", "you lie") to drown out the narrative. There is good evidence in these early days of the post-election period that the Obama administration has learned the "don't lose track of the narrative line" lesson. And in his speech he did an effective job at connecting "jobs, jobs, jobs" to his economic ideas of more R&D and education, something he had been less than successful at before. So on the messaging front there is reason for hope.
But on the "negotiation" front, we'll just have to wait and see if he has made the proper course correction. It shouldn't take too long. The "repeal Obamacare" front is a distraction. The GOP is currently putting a lot of energy into it but polling suggests that the public as a whole wants to move on. So I expect that this front will consume way more energy than it deserves but that doesn't stop it from being a side show. The main front is on the economy in general and the budget in particular.
Here I don't think the Republicans have figured out just how little time they have. There are three major items. In order of when they must be addressed they are: (1) The budget for the remainder of this fiscal year (April to September). (2) Raising the debt ceiling. (3) The budget for the next fiscal year (fiscal 2012: October 2011 - September 2012). Taking them in this sequence:
FY11: The Federal government is currently operating on a "continuing resolution", funding everything at FY10 levels. This is a real mess. Some old programs that have ended are funded whereas new programs that didn't exist in FY10 are unfunded. Then there are some funding level tweaks up or down for continuing activities that should be made but can't due to the rules of how "continuing resolution" funding works. This situation is entirely the result of GOP obstructionism in the Senate. Never the less, the Republicans have about a month to get a real budget for the second half of FY11 in place.
In this time the GOP controlled House needs to create and pass a bill. Then the "Split but with the Dems still having a slight majority" Senate must approve it too. Then, if there is any difference between the two bills (the most likely scenario), it must go to a joint House - Senate "reconciliation" committee and both houses need to pass the reconciled version. Then the President needs to sign it. And hopefully this is all done by March 1 so that the governmental departments get a few weeks to figure out how to implement it. It is unlikely in the extreme that the GOP will be able to meet this schedule. And I note in passing that these kinds of last minute budgeting shenanigans result in a lot of "waste, fraud, and inefficiency" the GOP pretends it is opposed to. We could probably increase the efficiency of and reduce the waste and fraud in the Federal budget by 2-5%, hundreds of Billions of Dollars, just by approving a budget in a timely manner.
On paper the GOP has taken a good first step. They have made Rep. Ryan the House "budget Czar" (hypocrisy alert - there's just no other way to describe the situation accurately). Theoretically Ryan can whip something out quickly and, on paper, he can run roughshod over the mandarins that control the various committees that usually deal with each part of the budget. We will see how this works out in practise.
The GOP has been very careful to promise dramatic budget reduction without being specific. Theoretically there are specifics, some of them from the esteemed Mr. Ryan. But these specifics have been glossed over for the most part. Mr. Ryan, for instance, did not trot out any of these specifics in his official GOP rebuttal to Obama's State of the Union. He very carefully stuck to "principles" and other generalities. But, wait! It's even worse. Little noted in the press was the fact that many Republicans accused Democrats of "cutting Medicare" during the last election campaign. This was based on the extremely modest cost saving provisions in Obamacare. It worked. The GOP did very well with seniors who thought these Republicans were the defenders of Medicare and Democrats were not. So these GOP politicians are now on record in the minds of these seniors as protectors of Medicare. The GOP is NOT the protector of Medicare and one of Mr. Ryan's proposals is to gut Medicare. And so it goes.
By being fiscally irresponsible while saying the opposite Republicans have put themselves in a bind. They have gotten away with this contradiction for years. They ran up huge deficits, most recently under Bush during the 6 years when they had complete control - they controlled the House, the Senate, and the White House. They have gotten away with this by adroitly deploying their brilliant PR machine. The media was asleep at the switch as were most voters. So no one noticed. And the Democrats were inept at both doing anything about it (understandable given their powerlessness) nor with getting the message out effectively (kudos to the GOP PR machine and the feckless press).
By now it should be obvious what the Obama administration and the Democrats need to do. Number one on the agenda is to start working together. I blame most of the past problem on the Obama administration. They spent all their time and energy on the GOP and a few Blue Dogs and ignored mainstream Democrats, especially House Democrats. The White House needs to work with and support a coordinated program. Given their choices, congressional Democrats will support the White House. But it would be tremendously helpful to their moralle and to their effectiveness if they knew the White House had their backs. The White House can then stay above the fray while Democrats in congress do the heavy lifting. The White House must be particularly careful to let the GOP go first on the budget for the remainder of FY11.
The Constitution, so near and dear to Tea Partiers, is on their side. Money bills must, by constitutional fiat, originate in the House. Beyond that, the White House can claim justifiably that they already laid out their plans in the budget they submitted almost a year ago. And they worked with the House and Senate on compromise legislation both in September and December. The GOP rejected all of this so it's time for the GOP to lay out in public the specifics of what they propose. I would further propose that, given how hard it is to get anything through the Senate, the White House refuse to comment on what the House GOP comes up with until they have passed a complete bill suitable for Senate action. And, while the White House is being conspicuously silent, House Democrats would be unleashed to lay into whatever the GOP in the House comes up with.
There are two general approaches the GOP representatives can go with. They can not try to cut spending in which case House Democrats lay into them for not cutting the budget as promised. Or they can cut drastically and House Democrats can lay into them for reneging on their promises (if they are foolish enough to cut into Medicare) or they can channel the anger of the supporters of whatever parts of the budget the GOP chooses to cut. And, if the GOP chooses to hew to some middle ground then Dems go at them from both directions. Critical to making this work is forcing the GOP to go first. Will the White House figure this out and actually do it? That's the question.
Debt Ceiling: The Debt Ceiling will have to be raised in the next few months. This can probably be put off until the FY11 budget is passed, assuming the GOP manages to stick to the schedule I have outlined above. There are a number of Tea Party types that are willing to shut the government down by not raising the limit. And, again this is a "fiscal" matter so legislation needs to originate in the House. That puts House GOP members in the hot seat, if Obama and the Democrats play their cards right. I am pessimistic they will. But I know what I would start doing if I was the Obama administration.
I would start publicly planning for a government shutdown. I would start quiet but make sure the story leaks. Then I would point to various statements by various Republicans and say "we are only trying to be prudent here". Then the next thing I would do is have House Dems start talking about not voting for any Debt Ceiling increase unless it has broad GOP support. "They are in charge. It's their responsibility to govern." would be the message. The message should also include documentation of the several times GOP members have voted en mass against raising the debt ceiling recently. Once this is done it should be just a matter of time before the GOP's business allies tell their lap dogs to get in line and make a debt ceiling increase happen.
Finally, there is the FY12 budget. The budget the Obama administration sends up in a few weeks should reflect what Obama said in the State of the Union. It should have increased funding for education and R&D. It should have cuts in the military. It should meet his promise of keeping discretionary spending flat, something that will be hard to do. (And "flat" means at the same level as his original FY11 budget proposal). He should not do the work of the Republicans by cutting except where I have indicated above. Of course, he could cut programs that are near and dear to Republicans but that is too much to expect. Let the Republicans take the heat from constituencies for cutting some one's pet spending. The key to the strategy is to get Republicans on record for unpopular cuts. And, if they are unwilling to step up and support unpopular cuts then to take them to task for that. Now a lot of the "bad cop" work can be done by congressional Democrats while the White House stays above the fray. But for this to work congressional Democrats need to know that the White House has their back. For the past two years it has been obvious to everyone that the White House did NOT have their back. That must change.
The President laid out a very modest agenda in his State of the Union speech. All things being equal I would have preferred a lot more. But all things are not equal so I am happy with the modest proposals he laid out. But I have been here before. President Obama is noted for his speaking ability. He has given many notable and some great speeches. The problem has been the follow through.
The narrative line from the DC pundits is that he concentrated too much on getting things done and not enough on communication. There is a lot of truth to this. Of course, the Republicans are past masters at throwing up noise ("Death Panels", "He is a Socialist who was born in another country", "you lie") to drown out the narrative. There is good evidence in these early days of the post-election period that the Obama administration has learned the "don't lose track of the narrative line" lesson. And in his speech he did an effective job at connecting "jobs, jobs, jobs" to his economic ideas of more R&D and education, something he had been less than successful at before. So on the messaging front there is reason for hope.
But on the "negotiation" front, we'll just have to wait and see if he has made the proper course correction. It shouldn't take too long. The "repeal Obamacare" front is a distraction. The GOP is currently putting a lot of energy into it but polling suggests that the public as a whole wants to move on. So I expect that this front will consume way more energy than it deserves but that doesn't stop it from being a side show. The main front is on the economy in general and the budget in particular.
Here I don't think the Republicans have figured out just how little time they have. There are three major items. In order of when they must be addressed they are: (1) The budget for the remainder of this fiscal year (April to September). (2) Raising the debt ceiling. (3) The budget for the next fiscal year (fiscal 2012: October 2011 - September 2012). Taking them in this sequence:
FY11: The Federal government is currently operating on a "continuing resolution", funding everything at FY10 levels. This is a real mess. Some old programs that have ended are funded whereas new programs that didn't exist in FY10 are unfunded. Then there are some funding level tweaks up or down for continuing activities that should be made but can't due to the rules of how "continuing resolution" funding works. This situation is entirely the result of GOP obstructionism in the Senate. Never the less, the Republicans have about a month to get a real budget for the second half of FY11 in place.
In this time the GOP controlled House needs to create and pass a bill. Then the "Split but with the Dems still having a slight majority" Senate must approve it too. Then, if there is any difference between the two bills (the most likely scenario), it must go to a joint House - Senate "reconciliation" committee and both houses need to pass the reconciled version. Then the President needs to sign it. And hopefully this is all done by March 1 so that the governmental departments get a few weeks to figure out how to implement it. It is unlikely in the extreme that the GOP will be able to meet this schedule. And I note in passing that these kinds of last minute budgeting shenanigans result in a lot of "waste, fraud, and inefficiency" the GOP pretends it is opposed to. We could probably increase the efficiency of and reduce the waste and fraud in the Federal budget by 2-5%, hundreds of Billions of Dollars, just by approving a budget in a timely manner.
On paper the GOP has taken a good first step. They have made Rep. Ryan the House "budget Czar" (hypocrisy alert - there's just no other way to describe the situation accurately). Theoretically Ryan can whip something out quickly and, on paper, he can run roughshod over the mandarins that control the various committees that usually deal with each part of the budget. We will see how this works out in practise.
The GOP has been very careful to promise dramatic budget reduction without being specific. Theoretically there are specifics, some of them from the esteemed Mr. Ryan. But these specifics have been glossed over for the most part. Mr. Ryan, for instance, did not trot out any of these specifics in his official GOP rebuttal to Obama's State of the Union. He very carefully stuck to "principles" and other generalities. But, wait! It's even worse. Little noted in the press was the fact that many Republicans accused Democrats of "cutting Medicare" during the last election campaign. This was based on the extremely modest cost saving provisions in Obamacare. It worked. The GOP did very well with seniors who thought these Republicans were the defenders of Medicare and Democrats were not. So these GOP politicians are now on record in the minds of these seniors as protectors of Medicare. The GOP is NOT the protector of Medicare and one of Mr. Ryan's proposals is to gut Medicare. And so it goes.
By being fiscally irresponsible while saying the opposite Republicans have put themselves in a bind. They have gotten away with this contradiction for years. They ran up huge deficits, most recently under Bush during the 6 years when they had complete control - they controlled the House, the Senate, and the White House. They have gotten away with this by adroitly deploying their brilliant PR machine. The media was asleep at the switch as were most voters. So no one noticed. And the Democrats were inept at both doing anything about it (understandable given their powerlessness) nor with getting the message out effectively (kudos to the GOP PR machine and the feckless press).
By now it should be obvious what the Obama administration and the Democrats need to do. Number one on the agenda is to start working together. I blame most of the past problem on the Obama administration. They spent all their time and energy on the GOP and a few Blue Dogs and ignored mainstream Democrats, especially House Democrats. The White House needs to work with and support a coordinated program. Given their choices, congressional Democrats will support the White House. But it would be tremendously helpful to their moralle and to their effectiveness if they knew the White House had their backs. The White House can then stay above the fray while Democrats in congress do the heavy lifting. The White House must be particularly careful to let the GOP go first on the budget for the remainder of FY11.
The Constitution, so near and dear to Tea Partiers, is on their side. Money bills must, by constitutional fiat, originate in the House. Beyond that, the White House can claim justifiably that they already laid out their plans in the budget they submitted almost a year ago. And they worked with the House and Senate on compromise legislation both in September and December. The GOP rejected all of this so it's time for the GOP to lay out in public the specifics of what they propose. I would further propose that, given how hard it is to get anything through the Senate, the White House refuse to comment on what the House GOP comes up with until they have passed a complete bill suitable for Senate action. And, while the White House is being conspicuously silent, House Democrats would be unleashed to lay into whatever the GOP in the House comes up with.
There are two general approaches the GOP representatives can go with. They can not try to cut spending in which case House Democrats lay into them for not cutting the budget as promised. Or they can cut drastically and House Democrats can lay into them for reneging on their promises (if they are foolish enough to cut into Medicare) or they can channel the anger of the supporters of whatever parts of the budget the GOP chooses to cut. And, if the GOP chooses to hew to some middle ground then Dems go at them from both directions. Critical to making this work is forcing the GOP to go first. Will the White House figure this out and actually do it? That's the question.
Debt Ceiling: The Debt Ceiling will have to be raised in the next few months. This can probably be put off until the FY11 budget is passed, assuming the GOP manages to stick to the schedule I have outlined above. There are a number of Tea Party types that are willing to shut the government down by not raising the limit. And, again this is a "fiscal" matter so legislation needs to originate in the House. That puts House GOP members in the hot seat, if Obama and the Democrats play their cards right. I am pessimistic they will. But I know what I would start doing if I was the Obama administration.
I would start publicly planning for a government shutdown. I would start quiet but make sure the story leaks. Then I would point to various statements by various Republicans and say "we are only trying to be prudent here". Then the next thing I would do is have House Dems start talking about not voting for any Debt Ceiling increase unless it has broad GOP support. "They are in charge. It's their responsibility to govern." would be the message. The message should also include documentation of the several times GOP members have voted en mass against raising the debt ceiling recently. Once this is done it should be just a matter of time before the GOP's business allies tell their lap dogs to get in line and make a debt ceiling increase happen.
Finally, there is the FY12 budget. The budget the Obama administration sends up in a few weeks should reflect what Obama said in the State of the Union. It should have increased funding for education and R&D. It should have cuts in the military. It should meet his promise of keeping discretionary spending flat, something that will be hard to do. (And "flat" means at the same level as his original FY11 budget proposal). He should not do the work of the Republicans by cutting except where I have indicated above. Of course, he could cut programs that are near and dear to Republicans but that is too much to expect. Let the Republicans take the heat from constituencies for cutting some one's pet spending. The key to the strategy is to get Republicans on record for unpopular cuts. And, if they are unwilling to step up and support unpopular cuts then to take them to task for that. Now a lot of the "bad cop" work can be done by congressional Democrats while the White House stays above the fray. But for this to work congressional Democrats need to know that the White House has their back. For the past two years it has been obvious to everyone that the White House did NOT have their back. That must change.
Thursday, January 20, 2011
Truthiness - Even Older than I thought
In http://sigma5.blogspot.com/2010/10/dumbth-truthiness-and-steve-allen.html I argued that Truthiness is really the same thing as Steve Allen's word "dumbth". I thought I was doing pretty good to trace the idea that far back until I read "The Greatest Show on Earth" by Richard Dawkins. It's a great book and you should read it. But what I am specifically referring to is information I found in his "from the war of nature, from famine and death" section in chapter 13. The revelation is on page 402 of my copy.
Dawkins refers to a logical fallacy called "Argumentum ad Consequentiam", "Appeal to Consequences" in English. The concept of a logical fallacy (unreliable argument) dates all the way back to the ancient Greeks. But according to a paper by Douglas Walton called "Historical Origins of Argumentum ad Consequentiam" published in 1999 in the magazine Argumentation (13 (3) 251-264) this logical fallacy is of more recent vintage. The phrase was first used in print by James McCosh in 1879. McCosh was President of the College of New Jersey (now Princeton) at the time. But this still makes the idea over 200 years old.
An "Argumentum ad Consequentiam" argument goes like this:
Dawkins refers to a logical fallacy called "Argumentum ad Consequentiam", "Appeal to Consequences" in English. The concept of a logical fallacy (unreliable argument) dates all the way back to the ancient Greeks. But according to a paper by Douglas Walton called "Historical Origins of Argumentum ad Consequentiam" published in 1999 in the magazine Argumentation (13 (3) 251-264) this logical fallacy is of more recent vintage. The phrase was first used in print by James McCosh in 1879. McCosh was President of the College of New Jersey (now Princeton) at the time. But this still makes the idea over 200 years old.
An "Argumentum ad Consequentiam" argument goes like this:
- If A is true then something good will happen.
- Therefore, A is true.
- If A is true then something bad will happen.
- Therefore, A is false.
- I think it would be a good thing if A were true.
- Therefore, A is true.
Subscribe to:
Posts (Atom)