A method for achieving honest climate predictions

I’ve written (repeatedly) that the “projections” in the IPCC’s Third Assessment Report are pseudoscientific nonsense.  I haven’t seen the Fourth Assessment Report, but I’m so confident (or would that be depressed?) that they will be nonsense too, that I propose a better way.

The fundamental problem is that the IPCC has no incentive to tell the truth in its projections.  But they do have an incentive to lie...to exaggerate the amount of warming that’s likely to occur.  So that’s what they do...they lie.  I propose the following solution:

The U.S. government should set up a prize fund totaling $400 million, payable in 2031.  The prize fund would be open to any U.S. university with accredited science or engineering programs.  The fund would be awarded as $200 million for first place, $100 million for second, $50 million for third, $25 million for fourth, $12 million for fifth, $6 million for sixth, $3 million for seventh…and $1 million until we run out of money.

Prizes would be awarded for most closely predicting the following parameters:

1)    globally averaged surface temperature anomaly for 2029-2031, relative to 1990;
2)    globally averaged lower tropospheric temperature anomaly for 2029-2031, relative to 1990;
3)    Atlantic hurricane basin sea surface temperature anomaly for 2029-2031, relative to 1990;
4)    average insured U.S. hurricane losses for 2029-2031,
5)    global sea level rise for 2029-2031, relative to 1990;
6)    CO2 atmospheric concentration;
7)    anthropogenic (industrial) CO2 emissions;
8)    methane atmospheric concentration,
9)    anthropogenic methane emissions,
10)  anthropogenic black carbon emissions,

The ten parameters listed above would be weighted such that the first 5 parameters are twice as important as the bottom 5.  Let’s say 100 universities enter the contest, with #1 given to the best prediction for each parameter, and #100 given to the worst prediction for each parameter. 

Let’s take hypothetical University X (not Xavier!).:  Suppose its rankings on the first 5 predictions are:  #3, #4, #10, #20, #60.  Since all those are multiplied by 2, its score would be:  (3+4+10+20+60)*2 = 194.  Suppose its rankings for the bottom 5 predictions are #20, #6, #70, #10, #8.  Its score on the second 5 would be 114.  So the total score would be 194 + 114 = 308.  The university with the LOWEST score would get the $200 million first prize, the university with the next lowest score would get $100 million, and so on.

For a total investment of $400 million, the U.S. government would get far better predictions than it currently gets from the IPCC.  (In fact, it’s not possible to get poorer predictions than from the IPCC…but that’s another story.)

In fact, Exxon-Mobil (to choose just one company at random ;-)) could get a similarly good deal by offering a prize fund of only $4 million (i.e. the proposed U.S. government prize to universities, divided by 100).  But this would be payable to INDIVIDUALS (or their heirs).  Exxon could restrict the prize fund to 100 scientists of its choosing.  The top prize of $2 million in 2031 would certainly be enough incentive to make truthful predictions!

Comments

[Mark Bahner]

Hi Zeratulss,

About the only question you ask that I have a real opinion about is the one about implantable microchips in people's brains. I'm pretty sure that by the middle of the century, most people will have some sort of electronics directly feeding into their brains. I also think many people will have electronics to supplement their senses, such as sight and hearing.

But I don't expect any substantial movement in that direction for another decade or two.

What do you think?

Mark

[Zeratulss]

24% of Americans believe that the Internet is able for a time to replace them with a loved one. For obvious reasons, such sentiments particularly prevalent among residents of the United States alone. Both men and women can replace the beloved, beloved trips to the World Network. However, the willingness to such transactions vary among followers of different ideologies: conservatives frowned relate to this idea, and the "progressive-minded" on the contrary, Nerkarat it.

Study company Zogby International also showed that every fourth resident of the United States have their own representation in the web-site or internet-stranichka. Creating internet-dvoynikov most passionate about young people (18-24 years of age) - 78% of them have personal Web page. In doing so, 68% of those surveyed said that the World Wide Web, they do not appear in its original capacity, their virtual overnight seriously different from the real.

Only 11% of Americans would agree implantable microchip in his brain, which would provide them with direct contact with the Internet. But the situation is changing, in the case of children. Almost every fifth resident of the United States would agree to equip their child safety device which would allow him to track the movement in space on the Internet.

10% of U.S. stated that the Internet brings them to God. " In turn, 6% are convinced that because of the existence of the World Wide Web God away from them.

And how you feel? Sorry bad English.

[Patrick]

This should be made self-funding, like a stock marketplace or the futures market. Think of the Iowa Presidential markets, where bets are made and paid based on winner of presidential race. The market reflects a good estimate
of actual probability of a candidate winning, in a way that efficiently sorts out the information.

Consider a futures market that paid out at the price of the average measured temperature in a given year.
Consider another futures market that paid out $1 for each measured ppm CO2 value. So for example, if in 2010 it measured at Muana Loa at 380ppm, it would pay off $380. What would the 2020 future go for?

These markets would help record the 'smart money' bets on things like this.

The main difficulty is the liquidity of such a prediction-only long-range market. Its not the 2020 predictions, but the 2040 and beyond predictions that are most at variance and most at risk.
There is a website called 'longbets' where people can make predictions and bet on them.

"Again, I need to do some more thinking over Memorial Day, but right now I'd say there's probably at least a 50/50 chance that ALL energy in 2100 will come from sources like photovoltaics and fusion."

You could for example, make a 'long bet' that says "By 2030, we will have developed commercial PV at a price point of under 5 cents/kwh, competitive with other forms of electricity." And you could see the pros/cons on it.

[Mark Bahner]

"Interesting idea. I see one obvious flaw, which is how "climate policy" is included in predictions.

Eg, I might predict that CO2 concentrations in 2030 will be 400 ppm, much lower than the IPCC. But maybe I believe that it will be only 400 ppm because I believe that Europe will extend Kyoto, that the US will enact Lieberman-McCain, and that China and India will join a climate treaty in 2020."

A couple comments:

1) I don't think ANYONE thinks there is a 50/50 chance of the concentration in 2030 being as low as 400 ppm (since it's already over 380 ppm). :-)

2) The problem that you hypothesize to exist could be addressed by having people submit one "on the record" set of predictions for 2030 that includes even actions as the result of climate treaties, and one "off the record" prediction that doesn't include any additional climate treaties.

"Whereas the IPCC SRES process is trying to make best estimates of future emission scenarios in the absence of any explicit climate policy."

No offense, but the IPCC SRES process is NOT trying to make "best estimates of future emission scenarios". No group of scientists could possibly be that incompetent.

It's not just the CO2 emissions and atmospheric concentrations that are too high. In fact, the CO2 emissions and atmospheric concentrations are MORE defensible than the methane emissions and atmospheric concentrations. Or the black carbon emissions and atmospheric concentrations. Or (on the cooling side) the SO2 emissions and atmospheric concentrations.

ALL four of those (CO2, methane, black carbon, and SO2) are too high. NO group of scientists who are legitimately making "best estimates" could be wrong on all four.

But even more irrefutable evidence that the IPCC has not done its "best estimates" is that the IPCC Third Assessment Report and Fourth Assessment Report projections are unfalsifiable, because they make no attempt to estimate the probability of the various scenarios. Even high school and first year college students learn that predictions must be falsifiable in order to be scientific.

"And indeed, it makes it hard to use the best estimate via your method for climate policy purposes."

No, if people are allowed to make a parallel "off the record" prediction for no climate additional climate policies, the policy makers could use the "off the record" predictions for amusement. (Policy makers need to have fun, too.)

"...btw, I do think that the SRES process was highly flawed, but I think their estimates are much closer to what I expect of a "world-with-no-climate-policy" than your estimates in which you predict a world with 50% fewer emissions by 2100 than today..."

I think that's because you haven't thought about or studied the matter nearly as much as I have. I plan to make a new set of predictions when (if!) I have some free time over Memorial Day. It's likely that my new predictions will be for even lower CO2 emissions come 2100 (though I'll probably boost my predictions to 2030 or so).

The reason I'll predict even LOWER emissions come 2100 is that I've become increasingly convinced that the world of 2100 will be almost unimaginably wealthier than in 2007. (I'm predicting essentially a factor of 1000 wealthier.) That will mean that technological progress will be almost unimaginably great.

Do you think any coal will be used in 2100? I think that's unlikely.

Oil? Again, I think it's unlikely that any automobiles will use oil come 2100.

Natural gas? Perhaps, but even if natural gas WAS used, it would be fairly easy to capture the CO2.

Look at the price of solar cells. They're currently about 25 cents a kWh, versus 3-5 cents for coal-fired electric power. But in 30 years or less, the cost of a solar cell will probably be equal coal-fired power. And in 100 years, there will be no place on the planet that allows strip mining of coal.

Again, I need to do some more thinking over Memorial Day, but right now I'd say there's probably at least a 50/50 chance that ALL energy in 2100 will come from sources like photovoltaics and fusion.

P.S. If you had $30 million in the bank, what would you use for electricity and heating of your home? I probably wouldn't use photovoltaics (don't look very nice) but I'd definitely have a fuel cell in my basement to generate both electricity and hot water/heat. I wouldn't care that such a setup might cost $20,000 and still not generate electricity more inexpensively than what I could get from the grid. I'd do it because it would be cool. (I'm an environmental engineer, with a mechanical engineering background.)

And I expect that in 20-30 years, photovoltaic shingles will look almost exactly like non-photovoltaic shingles. (As well as costing about the same as power from the grid.)

Interesting idea. I see one obvious flaw, which is how "climate policy" is included in predictions.

Eg, I might predict that CO2 concentrations in 2030 will be 400 ppm, much lower than the IPCC. But maybe I believe that it will be only 400 ppm because I believe that Europe will extend Kyoto, that the US will enact Lieberman-McCain, and that China and India will join a climate treaty in 2020.

Whereas the IPCC SRES process is trying to make best estimates of future emission scenarios in the absence of any explicit climate policy.

This makes it and apple & oranges comparison. And indeed, it makes it hard to use the best estimate via your method for climate policy purposes.

(btw, I do think that the SRES process was highly flawed, but I think their estimates are much closer to what I expect of a "world-with-no-climate-policy" than your estimates in which you predict a world with 50% fewer emissions by 2100 than today...)

[Mark Bahner]

Hi Bee,

"I would suggest that we also add several forecasts through time starting in 2010. I would then open the prize to all comers who submit models that can be reviewed."

I worry about opening it to all comers. If it's open to all comers, it will be open to nut cases who were just trying to game the overall results. It's like Internet opinion polls, where Rush Limbaugh (or Al Franken) tells his listeners to respond to the poll.

In fact, I've been thinking that prize fund could even be close to self-funding, by charging an entry fee. It would be like a raffle. Charge each entrant $500,000...but with a chance (if they finish #1) of winning $200 million. (Or $50 million, if the prize fund is switched to $100 million every 10 years from 2030 to 2100.)

"I think this idea can really be refined to address many of the problems associated with it but it is a BRILLIANT starting point."

Thanks. I'm open to all suggestions.

Mark

[Mark Bahner]

"Your idea is great. Plus each prediction should present its annual time series predicted value leading to the 2029-2031 target window. The contest sponsor then would post actual, empirical measures compared to each annual predicted value so that we all could evaluste the likely out come by trand analysis, before 2033 (when the final empirical measures would be available)."

Thanks.

Yes, I've been thinking about some modifications of this plan. There could be a prize fund that gives away a POTENTIAL $100 million every decade from 2030 to 2100 (i.e., 2030, 2040, 2050, 2060, etc.). That would be a POTENTIAL $800 million.

The predictions could be for every 10 years up to 2100. Then, starting in 2030, the original predictors could change their predictions...but they'd get a reduced reward.

For example, if the #1 most accurate prediction for 2040 was made in 2007, they'd get the full $50 million. (Remember that the winner for my $400 million prize pot got $200 million, or half of the pot.) But if the #1 most accurate prediction was made in 2030, the winner would only get 1/4th of the 50 million first place prize.

Or something like that.

You're absolutely right that it would be critical that prize fund sponsor would post empirical results. The results determination method would have to be beyond reproach, such that everyone agreed that the winner truly did win, according to the established winner determination method.

Thanks,
Mark

[Mark Bahner]

"I'm sorry, but I don't see the point of this at all. Presumably the predictions need to be made by a certain point in time (say, December 31, 2007) -- and then the prizes will be paid out on December 31, 2031. The cost would be whatever the cost of $400 million face in zero-coupon 2031 bonds is: let's say $100 million. What do you get for your $100 million? A set of 100 predictions."

Well, so far the IPCC process has probably spent over a BILLION dollars, and they have not made a single honest prediction. Only dishonest "projections."

This method would actually produce honest predictions. If dishonest projections are worth what has probably been over a billion dollars, 100 honest predictions should be worth at least 1/10th that much.

"I suppose at that point you can take the average prediction and compare it to the IPCC prediction."

No, one could compare the average of the 100 predictions to the IPCC "projections" long before 2031. One could do that immediately. I'm virtually certain that the 100 honest predictions would result in average temperatures much lower than the IPCC "projections."

"Let's say that the average prediction turns out to be more bearish than the IPCC prediction. Would you then say that the US govt should jump onto the IPCC bandwagon?"

There should be no "IPCC bandwagon." The IPCC is supposed to be providing objective scientific analyses. In the unlikely event that the average of the 100 predictions is more "bearish" than the IPCC projections, one would know that the IPCC projections are not dishonest.

Like I wrote, the IPCC has almost certainly spent more than $1 billion of the world's taxpayers' money over the 17 years it's been in the business of making pseudoscientific "projections." It is simply not possible for U.S. taxpayers to do any worse.

[bee]

I think this idea can really be refined to address many of the problems associated with it but it is a BRILLIANT starting point.

[bee]

A Brilliant Idea!!! Your idea is a 10!!!

Let's use the market!!!! Brilliant!!!

I would suggest that we also add several forecasts through time starting in 2010.

I would then open the prize to all comers who submit models that can be reviewed

Your idea is great. Plus each prediction should present its annual time series predicted value leading to the 2029-2031 target window. The contest sponsor then would post actual, empirical measures compared to each annual predicted value so that we all could evaluste the likely out come by trand analysis, before 2033 (when the final empirical measures would be available).

[Felix]

I'm sorry, but I don't see the point of this at all. Presumably the predictions need to be made by a certain point in time (say, December 31, 2007) -- and then the prizes will be paid out on December 31, 2031. The cost would be whatever the cost of $400 million face in zero-coupon 2031 bonds is: let's say $100 million. What do you get for your $100 million? A set of 100 predictions. I suppose at that point you can take the average prediction and compare it to the IPCC prediction. There may or may not be a significant difference between the average of the 100 universities' predictions and what the IPCC comes up with. But even if there is a significant difference, how is knowing that worth $100 million? Let's say that the average prediction turns out to be more bearish than the IPCC prediction. Would you then say that the US govt should jump onto the IPCC bandwagon?