UpdateRe-dated I’ve re-dated to Aug 1 from Jul 6 so people can easily check their bets and request revisions if they think their entry was a obvious blunder and also to let you all see how your bets compare to other’s. I’ve done this in the past. Obvious blunders include negative signs, off by a factor of 10 or more etc. For example: ivp0 entered 459296 and I’m tempted to assume this is merely a typo and change that to 4.5296, but for now I’ll wait for him to confirm that was an error.
It looks like Dr. Jay/Shoosh entered a ridiculously high bet of 999999. This is Shoosh, so I suspect he meant to enter that. Given the rules of engagement, unless Mars attacks between now and Aug 8 and the attack involves freezing the entire ocean and simultaneously increasing the surface area of the earth, he will have lost within the first 7 days. If Shoosh/Jay wants that changed, I’d appreciate him confirming the error before August 7.)
Original below.
———— ————— ———— —————
It’s summer. As usual, as we watch the NH ice melt we ask ourselves: Will this years minimum be lower than last years? Will the 2007 sea ice minimum be broken? Will we be able to bet some quatloos?
The answer to the last question is: Yes. Here are the rules:
- We use Jaxa Values posted here.
- On October 3, 2011 (or slightly after), I’ll compute the 7 day over the most recent value in August and September. In case of drop outs, I still compute over 7 entries; this could span more than 7 days. This is the only way I know to deal with the possibility that Jaxa might fail to report for 7 days in a row.) Whoever gets closest to this value wins the quatloos. I anticipate this will be the minimum weekly average and it will probably be the minimum for the year.
- Whoever gets closest to the minimum 7 day average wins the quatloos.
- I am betting 4.881 x 10^6 km2 basing my bet on the sophisticated method of “reading the post discussing last years preliminary value” and “adding a trivially small amount”. The reason for the latter is merely to permit someone to bet 4.88 if they want that value. 🙂
From JAXA
[sockulator(../musings/wp-content/uploads/2011/UAHBets5.php?Metric=NH Ice Extent?Units=x 106 km2?cutOffMonth=8?cutOffDay=1?cutOffYear=2011?DateMetric=NH Ice Weekly Ave Minimum?)sockulator]
The cutoff day is entered as 8/1/2011, which really means you need to bet by midnight 7/31… somewhere….
Tip to not screw up! :
- When you are betting, numbers between 0 and 8 are more or less reasonable. Today, the most recent extent value is 8566563, or 8.566563 * 10^6. (Yes. Annabell Torres/ Andrea/ Albert/ Laura Gonzales / Marie Deschamps/ Rebecca /Fulton, you are permitted to believe Mars will attack, cut off the sun, ice will begin to re-freeze so there will be more ice in August than today. But it’s unlikely. If Mars does attack, and the minimum ends up around 10, betting near 8 will probably win you the quatloos.
- After you successfully enter your bet, the betting form will be replaced by a message similar to this:
“Thanks lucia. I added your prediction of 4.881 x 106 km2. You wagered 5 Quatloos.”
It happens to appear slightly below the graph above and just before these tips. This will vanish when you re-visit the blog. - If you screwed up your bet, refresh the page. The betting form will reappear. Just re-enter your new bet correctly.
- Should you enter 1,000,000 and fail to notice when the bets are revealed I am willing to change the order of magnitude of your bet on request. I won’t change bets of 5.1 to 4.9 on request because neither of those are obvious blunders.
- Email is used for disambiguation of those with identical first names and to keep track of total winnings over multiple games (which I have actually posted once.) For the former, you can use a made up email. For the latter, you want to use the same email over and over.
- You are required to complete a simple addition problem including decimals to prevent bots from betting. Addition-enabled bots have managed to ‘solve’ simple arithmetic problems like 3+5=8, but adding decimal places seems to have thwarted them. These bots generally bet exactly 0. If you really, really bet zero, and you are not a frequent commenter, you might want to let us know in comments so I can distinguish you from a bot!
- If you are worried your bet does not appear and want me to enter it after the bets are made public, reveal your bet in comments. That will provide evidence that justifies my entering your bet manually after the fact.
Lucia,
I think they mean this:
http://judithcurry.com/2011/07/05/the-ipccs-alteration-of-forster-gregorys-model-independent-climate-sensitivity-results/#more-3924
It’s an interesting story, but needs to be taken with many caveats.
Kim–
Please recall I have chastised you for refusing to give links and act in otherwise obscure ways. I am interpreting this as a violation and I am
* premoderating you on every comment.
* putting you in the “wait” filter so that you will be unable to even enter moderated comments during your time out.
Comments that are as cryptic as 78687 will be sent to trash never showing.
Dr. Jay. Send me the study and all supplemental material by email and I’ll spend time reading it. Use the “contact me” button.
Well, the first caveat is that the authors of the paper were also part of the IPCC team that “altered” their results for the IPCC report, and so the “alterations” were done with at least their implicit acquiescence.
The second caveat is that the authors in question followed up their paper two years later with another one that in fact implies a larger climate sensitivity, so there is no reason to hold on to the results of the earlier paper other than that one may like them better.
Oh– I’m also going to do something uncharacteristic and prune some Nic/Jay/Andrew_KY comments to tighten up the thread. I’m leaving julio’s even though he may seem to be talking to himself.
That’s OK, I talk to myself a lot.
But really, you could delete all my posts as well (including this one). It does not bother me.
I wish I had something meaningful to say about sea ice, but I’m afraid I don’t…
Julio–You can bet even if you have nothing meaningful to say.
All right, I just did 🙂 (My first bet here–I hope it’s not the beginning of a life-shattering gambling habit!)
If you win, you’ll become addicted! 🙂
your description of the last 7 days is kinda confusing. do you mean the last 7 days, starting at the end of sept and counting backwards
Stephen–
There will be a string of 61 days. The strings will contain numbers like:
8, 12, NA, 14, 94, 46, 79, 102,……, 67,89,
I will throw away the NAs. After I throw away the NA’s I will compute the average over each set of 7 numbers– overlapping. So, I will get a 7 ‘day’ average over 7 days that report a number.
Ordinarily, the 7 ‘day’ average will be over 7 continuous days. But if Jaxa reporst an NA, it could be like this:
These 8 reports: 8, 12, NA, 14, 94, 46, 79, 102,…
contain an “NA”. So, I toss that. Then I average these 7 numbers:
8, 12, 14, 94, 46, 79, 102,
It’s averaged over 7 days, They just aren’t actually adjacent because the “NA”, day created a space.
This way I don’t have to do anything like interpolate, freeze the previous number or worry about what happens if JAXA fails to report 7 days in a row in late September.
lucia:
1) Shouldn’t you have some sophisticated algorithm for infilling missing data in accordance with whatever you think you already know about trends? This is a climate science thing, after all.
2) I think that a 7-day period of missing JAXA data ought to mean that the house loses and it’s quatloos for everybody.
OT
Roy Spencer shows the June temperature up +0.314C.
I just bet myself, I’m betting on a massive sea ice recovery. In fact, we might find there is so much sea ice that we will be forced to send out crews on boats with flamethrowers to get rid of all the sea ice.
Bob–
Thanks. I’m just back from the gym, I’ll get a post up this afternoon. (I hope I didn’t forget to let all of you bet! I’ve set up robo-posts, and I’m getting an uneasy feeling…)
Re: lucia (Jul 7 06:16),
Now I’m even more confused.
which 7 day period am I supposed to predict?
if you have 61 data points and you compute the overlapping 7 day means
you end up with a bunch of 7 day means.
which one do I predict? the lowest one? the mean of all means? the last one.
telling me that you’ll compute the overlapping means
(1-7)(2-8)(3-9)
Doesnt tell me which one I have to predict? all of them? the lowest?
?
Lucia, if there is time to change the rules, I’d go for daily minimum. Simply because it’s the most fun. Coming with the final result 2-3 weeks later is so 19th century.
It also makes betting easier, because people can easily compare with other years in the IJIS data set. Or are you going to tell us what the weekly minimum was in the period 2005-2010?
steven mosher– The minimum of all values. That is: The lowest one. I see where the problem is. The rule is: “Whoever gets closest to the minimum 7 day average wins the quatloos.”
Neven, I don’t want to change the rule to the daily minimum. I don’t see how it’s any more fun than the 7 day minimum.
I want the 7 day minimum — as in the title. I could easily tell you the 7 day minimum for 2005-2010. I assume anyone who can read the JAXA file and find the minimum single value for 2005 can also load them into excel or use their calculator to compute the running 7 day average. I also don’t see how we’ll know the minimum day much sooner than the minimum 7 day average.
Look at last year:
09,07,2010,5027188
09,08,2010,4989375
09,09,2010,4972656
09,10,2010,4952813
09,11,2010,4986406
09,12,2010,5005000
09,13,2010,5008750
09,14,2010,4998594
09,15,2010,4948438
09,16,2010,4890938
09,17,2010,4842031
09,18,2010,4813594
09,19,2010,4841563
09,20,2010,4878281
09,21,2010,4915313
09,22,2010,4983594
The record contains a relative minimum, the extent increases and then it decreases again. Maybe someone “knew” the value on 9/10 was not the minimum, before 9/18 and maybe they also a “knew” 9/18 would turn out to be the minimum when the 9/19 value was published. But I don’t know how they knew.
Also, I don’t know why you think my dates are calling the final result 2-3 weeks later. Look at 2007:
09,13,2007,4323750
09,14,2007,4291250
09,15,2007,4267813
09,16,2007,4267656
09,17,2007,4268750
09,18,2007,4281406
09,19,2007,4296250
09,20,2007,4310313
09,21,2007,4284531
09,22,2007,4276719
09,23,2007,4267344
09,24,2007,4254531
09,25,2007,4265000
09,26,2007,4297813
09,27,2007,4372188
09,28,2007,4441719
09,29,2007,4499688
09,30,2007,4592969
Maybe you think you can know for sure the 2011 minimum will happen before 9/15, but I don’t know how you know that. I need rules where we know who the winner is when I announce.
Re: lucia (Jul 7 13:42),
“The rule is: “Whoever gets closest to the minimum 7 day average wins the quatloos.””
PERFECT!
now I can write the program
I entered a bet but I think I’m going to be rejected on account of the fact that the second box did not appear to accept sufficient characters to handle my ridiculously long email address.
Geoff–
Did you get a message telling you your bet was entered?
There has been quite a down shift recently in the melt rate according to Jaxa’s data (the last two weeks). The daily numbers are still way below average but it is quite a shift in the trend.
The only time periods which have been this far off the normal trend would include 1996 (April to June, ended up at the highest Sept minimum), 1980 (mid-August to Sept, ended up at the highest daily minimum) and then 2007 (July to late-Sept, ended up at the lowest minimum).
http://img850.imageshack.us/img850/1165/dailyseiaug0111.png
Bill– Yes. I computed the 1 week loss rate and it’s currently at the lowest level for this time of year in the JAXA record. I have to pull in the extra data discussed on another thread– but it’s definitely low!
Geoff Cruickshank
Tell me what you intended to bet. I’ll add you.
So I went through my database back to 1972.
Over the past 21 days (since the 2011 trend started to change), only 1973 had a lower melt rate over the same period than 2011 – it is 33% lower than the average.
Not sure what’s going on. It could be a sensor issue.
Bill–
The melt rate over 7 days is at a record low since 1973. It could well be a sensor issue– or not. It’s not so low that I’d say “just has to be instrument error”, but it is low.