I thought I’d show you how people’s bets entered here compare to the current values as well as my estimate of the current estimate of the range in which the 7-day minimum will fall. Below I’ve posted a histogram of the bets (purple), along with the ±95% range in which bets fall (purple lines):
The most recent 7-day minimum is shown in black; the best estimate and ±95% confidence intervals for the up-coming minimum based on my weighted model is show in blue.
My bet is shown in yellow and is fairly close to my current prediction for the upcoming winning value. I’m not the closest though: The current front runner is DeWitt Payne who wagered 5 quatloos the minimum extent will be 4.591 million square kilometers. But the pack is tight: the whose bet is within 0.1 millions square kilometers of my current best prediction for the upcoming minimum are: lucia , rc , David Jay, George Tobin, mccall, Steve T, John F. Pittman, Chris Brown, DeWitt Payne, ivp0, Geoff Cruickshank, RobB, Boris, AMac, Anamoi. (Listed in reverse order based on when the bet was entered.)
Meanwhile, according to my estimate of the of the most likely ultimate value, 56 betters fall inside the ±95% uncertainty intervals for the range in which the 7-day minimum is likely to fall; 20 fall outside.
But even if you fall outside my estimate of the uncertainty for predicting the minimum, keep heart! There are still at least 5 weeks to the ice minimum. Statistical predictions are just based on assuming the past is prologue to the future: but sometimes things change. This year, the ice could blow out of through the Fram– or not. Almost any better could still win.
Has anyone obviously lost?
Of course. By entering a ridiculous value that exceeds the surface area of the earth, Shoosh/Jay lost before the melt even started. But I ignore him when computing statistics.
Some of you can now see that the relatively high mean value of blackboard bets arises in part from bets by “feabqtcqy” (the adding enabled bot who keeps betting) and “suzanne”. Both may have adopted the strategy of trying to capture the quatloos in the event that mars attacked clouding the skies and preventing ice melt. Since the 7-day minimum in August is well below their values, even if ice began to reform now, both suzanne and feabqtcby’s bets are out of contention for the win.
Sorry feabqtcby and Suzanne, better luck next time. feabqtcby, I know you’re the silent type, but your quatloos are always welcomed by the pool. Suzanne, if you’re lurking, let us know whether I’ve correctly guessed your strategy.
4.020 for me unless it is to late.
I have missed the deadline but am interested in the purpose of number of quatloos bet. If someone only bets 1 do they only win 20% of the pool of quatloos? How are the rest divided? If they win the whole pool, why bet more than 1 quatloo?
My calculation was that the ice melt would stop as soon as I placed my bet. I had received word that trillions of my high-albedo nanobot comrades would be deflecting sunlight from high to low latitudes.
However, the plan has yet to be cleared by the EPA, Environment Canada, or the Ministry of Natural Resources and the Environment. Thus, you carbon-based life forms (bettors included) are safe… for now.
Little did I suspect that the greatest impact of my quatloo-garnering scheme would be to throw off Lucia’s crowd-sourcing statistics. As this is a bot-friendly website, I grant you permission to classify my bet with Shoosh’s, from this point on. (But I cannot
speaktype for ‘suzanne’.)cyclonebuster– It’s too late unless I start a 2nd shorter term et.
crandles,
The #1 person doesn’t win the whole pool.
There is a very complex algorithm that starts by finding the total and putting it in a pot.
The amount the #1 person wins depends on the number of quatloos bet and the amount in the pot. The amount won is linearly proportional to amount bet. So, if you think you are really good and want to win lots of quatloos, you bet 5.
The amount bet by the #1 person is debitted from the pot.
The amount the #2 person wins follows a similar formula. And so on.
The coefficients in the algorithm were ‘perfected’ by ‘fiddling’. But mostly, the top three always net money, and sometimes if they happen to bet very few quatloos, people lower down also win some money.
I don’t know how people decide how many quatloos to bet. I guess even with quatloos, some people are higher fliers than others.
“The #1 person doesn’t win the whole pool.
There is a very complex algorithm that starts by finding the total and putting it in a pot.
The amount the #1 person wins depends on the number of quatloos bet and the amount in the pot. The amount won is linearly proportional to amount bet. So, if you think you are really good and want to win lots of quatloos, you bet 5.
The amount bet by the #1 person is debitted from the pot.
The amount the #2 person wins follows a similar formula. And so on.
The coefficients in the algorithm were ‘perfected’ by ‘fiddling’. But mostly, the top three always net money, and sometimes if they happen to bet very few quatloos, people lower down also win some money. ”
WOW!!!
Roy Weiler
Roy–
The method was explained way back when I concocted it. I’ve also given the code to people. It’s actually not that “complex”, but it’s a weird, pari-mutual thing. It’s more fun if the quatloos get well distributed, but we also want the #1 person to get more. So… I came up with a weird algorithm.
The average loss from today is 786,000 km2.
The average loss would then make the chart from 1972 to 2011 look like this (I’ve also included another line showing how ridiculous the claims are that the ice would melt out completely – the ice loss through the season would have to be three times higher than today – don’t fall for the NSIDC charts which have the Y-axis cropped at 4 million rather than Zero).
http://img202.imageshack.us/img202/1047/nhseiaug152011.png
Bill Illis–
Did anyone predict an ice free Arctic this year? If yes, who and when did they predict it?
Found one!
Patrick Lockerby.
March 2nd, 2011.
[“I couple these ideas together and explain why I expect the central Arctic to be essentially ice-free by the end of this Arctic summer 2011.”]
http://www.science20.com/chatter_box/arctic_ice_march_2011-76831
🙂
Yes, but he also expected Baffin Bay to be covered in ice.
Re: Bill Illis (Aug 16 19:05),
Informative graphic, Tufte would approve!
Re: ” the ice loss through the season would have to be three times higher than today”
An extra 4 million fall on a 10 million fall sounds more like 40% more than three times higher to me. Comparing to ‘today’ ‘s fall seems a bit odd to say the least.
lucia (Comment #80476)
August 16th, 2011 at 8:37 pm
—
I don’t recall anyone saying the ice would melt out this year. The same melt-out line would apply for any year that an ice-free arctic is projected; I’ve heard 2013, 2030, 2070 and others mentioned. The change required, however, is much, much more than people have surmised.
Re: Arfur Bryant (Aug 17 02:42),
Not really
“By September the bulk of sea ice will remain only in the area along the coasts of the Canadian archipelago and Greenland, with an extent of substantially less than 4 million km2. – perhaps less than 3 million km2.”
Bill–
I’m not sure you can just look at ice loss from March-August to predict when the extent will hit zero. Slower ice gain rates during the refreeze ought to also matter. With a slower ice gain rate, you end up with less ice at the start of the melt season even if the loss rate is nearly constant.
I suppose the really interesting thing about the Great Arctic Thaw is to look for non-linear transitions along the way. Will the process go smoothly and predictably, or will there be some inflection/tipping points along the way? This might help refine predictions for the Greenland and Antarctic edge ice caps, which are the true “hurt lockers” as far as sea levels are concerned.
This is what he says right after Nick chops the text:
So not quite.
But a bit more mad-hatter than Nick’s quoting suggests. (Not that anybody has ever accused Nick of spinning or selectively quoting what somebody is saying, when he’s trying to make them look sane. 😛 )
lucia – points taken.
Let’s take a little more realistic approach and start 30% lower than average in March and have the daily melt rate 20% higher than normal. It actually results in an even more unrealistic trendline.
http://img193.imageshack.us/img193/4008/nhseimeltout2.png
toto:
As long as melting is dominated by mechanical transport of ice out of the “deep” arctic into warmer Atlantic waters, I’d say an inflection point remains unlikely.
However, we did get a “phase shift” around 2007. It shows up in the amomalized data as a residual annual cycle.
I’d say that’s evidence of a new pattern in the cycle, as was the late freezing of the Hudson Bay last year and the impact of that on more southern weather.
(IMO no tipping point, just new weather and ecological patterns. Of course that never happened before humans interfered with climate. >.< )
Bill, thanks for the great graphics.
If you look at the amplitude of the oscillations and compare them to the shift in the baseline (see this) it does look like Lucia is correct, so far at least…
Most of the shift in the amount of ice loss is accounted for by a baseline shift. (My method involves a Fourier decomposition of the Arctic ice extent curve using a two-year window, so necessarily it stops before 2011.)
Bill-
Your current demonstration that we aren’t going to see ice free Septembers seems to hinge an assumption that the ice level in March could only be 30% lower than the recent average.
In your earlier post, you included a range of years up to 2070.
Why would an March ice level of 30% lower than the recent average represent a reasonable estimate in 2070? I don’t have an estimate for likely ice levels in 2070– but your conclusion would seem to assume that I must think March ice levels in 2070 couldn’t be lower than 30% of the average from some recent period. Could you tell me why I should assume that?
Carrick
‘Dominated’? It is an important factor but I am not aware that it is the ‘dominant’ one. Especially over the past 4 years. Changes in the haline content of the ice (that is to say as the ice increasingly made up of younger 1-2 year ice rather than older 4-5 year ice, hence the ice itself is increasingly briney) has apparently meant that melting briney cold water is going to a deeper layer (being more dense than fresher water) forcing up warmer deeper water to melt the ice from below (Boxall June 2011 ), while influxes of warmer waters have also had a major impact. I am not sure anyone is in a position to ascribe which change has been the dominant factor. But perhaps I have missed something.
My guess was based on nothing as sophisticated as Lucia’s methodology, or even something as clever as a back of the envelope calculation, for that matter.
Summer ice (July-August-September) had been in a steady decline since the 1960s, until the 2007 drop. Since then, the plot has been more horizontal (based on 3 data points!), as the ice seemingly wants to return to the longer term trend. My guess of 4.97 was made on the assumption that the 2011 ice continues that recent trend.
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/seasonal.extent.1900-2010.png
Starting out at a 30% lower extent probably isn’t realistic, but suppose PIOMAS volume numbers are not miles out and the volume was 30% lower than the maximum reached in 2010 of 21.938 = 15.36 almost all by the ice being thinner. Is that a huge change given a reduction from just over 33 in 1979 to 21.938 in 2010? Doesn’t seem so to me.
Now the volume reduction in 2007 was from 23.865 to 6.458 a fall of 17.407. Now what do you expect to happen if the volume is less then 17000km^3 at maximum and the volume reduction is capable of being more than 17000km^3?
Of course you probably don’t believe PIOMAS numbers….
But I don’t think they are going to be miles out. 25% more in reality I think is quite possible but I doubt 100s of % out.
Re: Nick Stokes (Comment #80485)
Not sure what you’re trying to prove, Nick.
.
Lucia asked for an example of someone who has predicted an ice-free Arctic this year. I provided one in the link.
.
http://www.science20.com/chatter_box/arctic_ice_march_2011-76831
.
As well as the quote given in my post, here is another, referencing 2011:
.
[“The Arctic is virtually ice-free by late summer: there is open water at the North Pole…
The fact of being broadly correct in most of my predictions during 2010 does not mean that I will be proven correct this year in the above and the following predictions for 2011. But I respectfully submit that my successes make my 2011 forecasts more worthy of serious scientific discussion.”]
.
This seems to be clear to me. I also agree with Carrick that the bloke seems to be as mad as a box of fish…
.
Open water at the North Pole? Yes, really…
I think you can easily have open water at the North pole without that being an ice free arctic.
Essential ice free is not the same as actually ice free. What is meant by essentially differs from person to person and Nick provided insight into what Patrick Lockerby meant.
Also your quote included central Arctic which is not the same as all of the Arctic.
So Nick and I are both saying no you haven’t “Found one!”, he is not as bad as that.
dolormin:
The beautiful thing about descriptive text is you can get data to say anything you want them to mean. But what the data actually indicate is that most of the variability in ice extent in the last four years is accounted for by the change in the mean ice extent (the “DC” offset). Very little shows up as a change in the overall variability of the amplitude (put another way any such change is statistically insignificant).
Perhaps this is naive on my part, but I would expect melting due to direct forcing to increase as the ice becomes more vulnerable, which should manifest itself as an increase in the oscillatory component.
Since the amplitude of the oscillatory component is not statistically different than it was prior to the last four years, and in any case are not explanatory for the observed decrease in the ice extent, I think we can safely conclude that the mechanical loss via wind transport remains the dominant mechanism.
Here’s one reference. I believe this reference from NASA JPL comes to similar conclusions. I’m pretty sure wind-driven flow (“Ekman drift”) is the consensus view for the ice loss from 2007 to this point.
As I understand it, having lost the ice, the increase in new to old ice makes the remaining ice more vulnerable to direct melting. Given that the arctic is warming up at some point one might expect an ice free arctic even on years with low mechanical stress.
So the picture shown by BIll may not be realistic, because the amplitude probably will have increased by that point. Still I think it sheds light on how dramatically different the time variation of arctic ice extent must be to reach the magic zero ice extent numbers.
(Testing whether there is evidence for a statistically significant change in amplitude was why I launched into a little mini-study on the Fourier decomposition of annual ice extent.)
This is a somewhat (a lot) cruder way to look at the analysis I did earlier. It basically pulls out the maximum and minimum reported ice for each year, computes DC= (min+max)/2 and annual amplitude = (max-min)/2.
Here’s the awk script, which you run with this data (splice of JAXA and GSFC, updated daily).
Plot of results.
Of course, it’s noisier that my previous calculation (no smoothing at all), but shows the same general trends (namely we are observing mostly a DC aka “secular” shift in ice extent to this point.)
Personally, I was astonished to find that “bettor” is legitimate spelling in the US.
andrew– Isn’t it legitimate elsewhere?
Don B (Comment #80499)-Heh, every time that freakin’ chart shows up I have to wonder if people have actually read the documentation file, because if they have, they really shouldn’t show the “data” prior to the fifties (I would argue prior to the late seventies) because it is quite frankly meaningless:
http://arctic.atmos.uiuc.edu/SEAICE/arctic.historical.seaice.doc.txt
The file makes it pretty clear:
“Please note that much of the pre-1953 data is either climatology or interpolated data and the user is cautioned to use this data with care.”
In other words the flat early portion of the graph is an artifact. It’s a really misleading plot and I wish people would stop displaying it or at the very least provide caveats, because frankly it is such total BS to do otherwise.
Okay, I’ll grant you that you referred only to data after the date that CT seems to think the data is okay. If anyone thinks we know the ice extents decades or more prior to satellites they are just nuts. Sorry, that plot just irks me every time I see it because of how much it misleads people.
Lucia the preferred spelling in the UK is better.
Lucia, I found the spelling sufficiently jarring to think it a typo, but thought it unlike you to be that careless, so looked it up in Collins, which gives “bettor” as the “esp. US” spelling. Intrigued, I consulted Fowler (the real one, not the dreadful 3rd edition), who has this to say: In the meaning ‘one who bets’, bettor is more common in AmE and better in BrE. Bettor has the advantage of being distinct from the comparative of good, although their distinct usages keep them out of each other’s way.
Andrew–
Well, that’s a kind thought. In reality, I’m a very poor speller and proof-reader. WordPress doesn’t suggest spelling errors in titles. So if you think you see a spelling error in the title, usually it will be a spelling error. I admit that every time I discuss “bettors” or “betters” I think both spellings look dubious. If WordPress doesn’t flag it, I just plow ahead.
I assume you would plough ahead? 🙂
Re: Don B (Aug 17 09:22),
That particular extent time series has problems. The drop in 2007 is excessive compared to JAXA , NOAA and Uni-Hamburg, i.e. pretty much everyone else. I wouldn’t bet the farm based on that data.
It appears that one can bring quatloos into existence by the mere act of betting. I did not bet the maximum number and nor did many others. This argues against the economists idea that economic agents always maximise individual utility. Obviously there is no tragedy of the commons amongst sceptical types, which is very heartening.
Geoff–
For every qualtoo created in winnings a quatloo is destroyed by debt. The distribution rules obey the law of conservation of quatloos.
Big drop on Channel 5 at UAH.
First thought was instr. error, but this has been going on for almost a week.
JohnM–
This is the time of year when we see big drops in absolute temperature. But that drop does cross a bunch of past years– so yes. It’s pretty big.
I would plough ahead myself, but if you prefer to plow ahead, more power to your elbow.
“Geoff–
For every qualtoo created in winnings a quatloo is destroyed by debt. The distribution rules obey the law of conservation of quatloos.”
Golly! It’s more complicated (but simpler) than I thought.
For every qualtoo created in winnings a quatloo is destroyed by debt. The distribution rules obey the law of conservation of quatloos.â€
I don’t think this is a conservation law. It seems more likely that the qualtoo and quatloo are antiparticles, and there’s some kind of annihilation going on here. Possibly involving gravitational polarisation (or polarization) of the quantum vacuum.