The prize is relatively easy to make unwinnable,and do so fairly.
A data set containing noise of a particular type can be generated.
That data will by chance have trends in it.
Make sure the actual trends you add are so small they cant be detected reliably.
Muck around some more, and make sure the odds of doing it by chance are such that the $10 entry fee means it a very good bet you will have more than 100K in entry fees before someone would guess and your done.
To make the contest dishonest that can also be done if you want to and done in a way that is pretty much undetectable. (Well kinda)
Create an algorithm, a PseudoNRG (see the web there's recipes) then pick seed using a real RNG .Generate your test set. Check if its quite nasty as you got lucky and the trends tended to be added to random data sets with opposite trends. If it didn't, run it again with new seed. Stop when you have problem that cant be solved.
But as I said there is no need to do that. There are indeed limits to what size trends can be detected.
If you don't have apriori knowledge about what the earths temperature evolution might look like, and no sanity limits on what time scales, there might just be chaotic butterfly like effects. Then there is indeed no amount of evidence that will ever be sufficient to reach statistical conclusion about climate.
To keep his $100K, he just denies there are any limits on the size of the timescales for random noise due to factors we dont know about.
he just presupposes the existence of an ever larger magic variation by unknown causes, until all evidence fails to show AGGC forever.
That we in fact have such good handle on why so many climatic events happened.
https://www.youtube.com/watch?v=RffPSrRpq_g is something he will ignore.
That we have direct physical evidence that CO2 blocks IR radiation, and that outgoing radiation has in fact changed over time. Is something he will ignore.
That if you do the physics of line by line analysis of the emission spectra and what happens when you add Co2 to the atmosphere, then you get a climate signal response similar to what happened.
All that^^^^ for instance is actually in the IPPC reports yet he claims there is no basis for expecting a roughly linear increase in temps since about 1980, due to the changes that occurred in CO2.
And while that is a low order approximation, the GCM provide comprehensive solid physical basis for the expectation of just what shape the Climate change will be.
One of the details that has been worked out in recent years si we know the models are bad at predicting just when EL Nino will change phases. The models do change phases periodically, and in a way that looks like the real pattern of El ninos and La ninas, but being chaotic processes with strong feed backs, it does not predict just when.
If you take climate model modelling all the basic Physics, and artificially force it to get the SST in one region of the pacific correct, then the rest of the model correctly simulates the rest of the climate, and then suddenly the model not only predicts the warming trend but now also matches the whole lot of the wiggles.
That demonstrates that the bit the models don't get right is the timing of El nino.
There is in fact extraordinary amounts of evidence justifying the shapes of the models used to fit Climate data.
Pretty sure arguing with him will be about as ineffectual as arguing with some of the nay sayers I have seen on this forum. So if you felt you must I couldn't suggest you shouldn't.
Just don't be surprised if the semi obvious happens.
Bookmarks