Wednesday, September 25, 2013

Fifteen years later, Patrick Michaels finally makes a bet on climate change

Via Tamino, I learned of a not particularly large, $250 bet with Scott Supak over whether there will be statistically significant warming in 25 years starting in 1997 (details posted at Roy Spencer's blog).

This is an update to a bet offer that Michaels' newsletter made in 1998 for a 10 year period. James Annan learned of that offer in 2005 and tried to accept, but the new editor Chip Knappenberger pulled a Lindzen (defined here) and declined to keep that bet. Seems like a pretty good bet for Scott despite starting in the 1997-1998 El Nino, some uncertainty about defining statistical signficance, and despite using HADCRUT which as I understand it leaves out the rapidly-warming high Arctic.

I btw have my own series of bets with up to $9,000 on the line, starting with 2007 five-year average and ending with 2017, 2022, and 2027 five-year averages. So far it's not going well for me, but it's early days. Best case scenario is at the end of 2029, I've lost my shirt. Worst case scenario is I've won every bet. Almost-worst case scenario is that statistical or a real temporary lull cost me on the 10-year bet (which will slow down efforts to address the problem) while I win on the 20.

7 comments:

Anonymous said...

I'm not sure whether to wish Eli and Scott luck, or not...


I've tried myself to convince many deniers to wager with me, usually on the Arctic sea ice volume but once or twice on the next mean global temperature in an El Niño year, but they have all run like hares from my offers.

I feel a bit like Connor McLeod in the scene from Highlander when he says "nobody will fight me - they all run away".

I don't look as good in a kilt though.


Bernard J.

Scott Supak said...

"I'm not sure whether to wish Eli and Scott luck, or not..."

My wife has this same problem. She likes to remind me that I'm betting on something awful to happen. I tell her, it's going to happen, I'm just taking money from the people who are helping it happen.

Well, I'm not taking money. I'm transferring money from them, in this case, to the Climate Science Legal Defense fund.

That was the best thing about this. I kept my choice of charity quiet until he'd basically agreed to the bet. When he heard what my charity was, he stopped talking to me, so I had to chase him down in the comment section of Roy Spencer's blog.

Roy also has a bet I'm trying to decide if I want to take of offer a counter. I wish he would phrase things in better "bet" parlance, like Intrade, but maybe you guys can advise me on this (I'm not a climate scientist, I'm a gambler who sees this as a much better bet than the don't pass line, or poker.

Here's what Spencer offered:

"I’m also in discussions with Scott over betting on a trend that would be 1 standard deviation below the average model warming, which would be +0.162 deg. C/decade for 1997-2021, compared to the 90-model average of +0.226 deg. C/decade."

All advice is welcome. If I understand this correctly, he wants to bet that the average of all models will be less than 0.162 dC/decade from 97-21?

Someone said I should offer 0.15. But I'm curious what you guys would put the odds at of hsi number.

Thanks!

Scott Supak said...

BTW, "Stupak" is the name of a Vegas casino owner, or a conservative, anti-choice former Democratic Representative from Michigan. My name is Supak.

Anonymous said...

Since you asked: I put the odds at 60:40 in your favor for 1997-2021 at Spencer's number. That is a lot of warming, basically more than over the past 35 years, wood for trees index 0.137 per decade.

Rib Smokin' Bunny

Brian said...

Hi Scott - sorry and I'll fix your name spelling.

Spencer's offer is worse than my bet terms, and I'm losing my bets currently, kind of. Starting at 1997 seems dubious to me. Maybe let any random year in the 90s be your start year, and after you agree on the bet then choose a random process to settle on a year?

What seems like a fair bet to me is one that's midway between predictions. So what's Spencer's prediction? He seems to run away from that.

Atomsk's Sanakan said...

Hey Rabett. We've seen a lot of each other on Twitter, but I think this is the first time I'm commenting on your blog. Just wanted to let you and Mr. Supak know a few things:

1) Patrick Michaels is basically guaranteed to lose the bet at this point, and there's a good chance Spencer would have lost the bet, if he's decided to make it. At this point, for 1997 - 2019, HadCRUT4 has a statistically significant warming trend of ~0.18°C/decade:
http://archive.is/TnltZ#selection-1273.0-1291.9

2) I let potholer54 [Peter Hadfield] know about the bet, since he made a video on temperature trend predictions, along with debunking Michaels' spurious attacks on James Hansen's temperature trend projections:
http://archive.is/fJud6#selection-2227.0-2265.61

3) I have a Twitter sub-thread on this bet, as part of my thread on temperature trend predictions and bets. That thread also includes your 3 bets with David Evans, the first of which you're going to win at the end of this year. Same point for your attempted bet with Lindzen, which you would have been well on your way to winning, if Lindzen had had the nerve to accept it:
Supak vs. Michaels: https://twitter.com/AtomsksSanakan/status/1171055898933940226
Rabett vs. Evans: https://twitter.com/AtomsksSanakan/status/1073811906769682433
Rabett vs. Lindzen: https://twitter.com/AtomsksSanakan/status/1084138077600317440

Atomsk's Sanakan said...

Checked with Dr. Kevin Cowtan, and found that there's something off with my previous HadCRUT4 calculation from the WRIT website.

So I've checked Dr. Cowtan's website and the KNMI website to confirm a 1997 - 2019 HadCRUT4 trend of ~0.14 +/- 0.08 °C/decade (for 2-sigma). That trend is statistically significant at the 95% level, and thus still confirms that Scott Supak will almost certainly win the bet and Michaels will almost certainly lose:

http://www.ysbl.york.ac.uk/~cowtan/applets/trend/trend.html

https://climexp.knmi.nl/getindices.cgi?WMO=UKMOData/hadcrut4_ns_avg&STATION=HadCRUT4_global_temperature&TYPE=i&id=someone@somewhere