Kate and Phil263, can you give me a point estimate to go with your uncertainty ranges? I need point estimates for the basic stats. Regarding discussion vs. just-prediction posts, discussion is welcome too, explaining why you think your number is reasonable. This thing will be most interesting, though, if everyone's predictions reflect what they seriously believe is going to happen with the ice, not shifted by tactical thinking.

As might be expected, the ASI median estimate came in substantially lower than the SEARCH SIO median. I'll write a post comparing the two distributions over the next couple of days, setting up an analysis that should continue through the season.

4 century breaks for CT area this month, while DMI 30% is declining more slowly.

Neven, I think the spam filter just gobbled my wrap-up post.

Some might imagine a battle of the blogs here, but that's not where I'd like to take it. As described in the OP we've started an experiment, with most of the data yet to come. And the analysis, and the writing.

I'm making a point of keeping my own thumb off the scales here! But I will publish some analysis soon to give y'all some feedback. That will include comparison with the distribution of predictions by SIO science teams, and a preview of the analytical strategy.

BTW, one well-known sea ice researcher suggested to me that if we averaged our Sea Ice blog value with that from WUWT, it would be a mean between pessimists and optimists, and closer to most of the science teams.

Just to be clear, I'm not planning to send our results to SIO this month, viewing it as a trial period -- unless you folks tell me otherwise. Next month's post will be more clear about Sep mean in the title, and I think we should send those results to SIO. I'll also soon post an analysis of the June results, which have about 113 individual estimates (median still 3.0).

One interesting thing about public opinion on this Arctic/weather question is that it varies with weather on the day you ask the question.

Dr. Francis made a crucial suggestion regarding our new paper on what the public *believes* about an Arctic/weather connection. That paper is now out for review, but you can see the survey question and basic response percentages here: http://www.carseyinstitute.unh.edu/cera/exploring-public-views-science-environment

Kamaljit_raj, should I put you down for 3.9, 4.8 or neither? To include in the stats I need a point estimate, although discussion can range far from that here. The median so far (111 by my count) is 3.0, with late-arriving estimates tending to be higher than those from early May. The SEARCH SIO deadline for June estimates is this Friday, June 7. Since this is our first iteration of this crowd-source experiment I had not planned to send the results directly to SEARCH. A few bugs need working out -- confusion about minimum day vs. September mean, and the problem of drawing most of the estimates a month before the official deadline. I'll try to improve both points for the next iteration/July deadline. Part of the interest will be in watching how estimates evolve with the season. If you haven't sent yours in yet and would like to, there are still a couple more days.

Our raw data consist of exactly what you see here, the OP plus all the comments. Analysis starts with the numerical estimates. I don't know how that will turn out, but the comments will be a first resort to interpret. For me this whole undertaking has an exploratory, EDA-like flavor. Take a robust/visual approach and see what we learn.

To R. Gates and others who might wonder about outliers -- Internet polls are, you might say, a dime a dozen. I don't know if this one will be any righter than others, but I can promise it will be thoughtfully analyzed. (And not just by me, it's crowd-sourced, after all.) Some statistical summaries such as mean, standard deviation, correlation, OLS regression (the classical family of moment or least-squares statistics) are notoriously not resistant to outliers, meaning that one or a few wild values can move them any distance. Other summaries, however, such as median or robust regression, have higher resistance to outliers. These are good choices if you don't want your statistics to be controlled by a few wild values, but also don't want to arbitrarily exclude certain values from the data. A typo could shift the highest guess in our poll from 6 million to 600 million, and the median would not move at all. In my (inadvertently) dual posts above I cited the median for this reason, although currently the median and mean are pretty close so it doesn't make much difference. Also, I showed the whole distribution as a stem-and-leaf display, a method invented by the very clever John Tukey as part of his Exploratory Data Analysis (EDA) approach -- which emphasizes graphical displays that make no assumptions about the data, and can highlight unexpected features such as outliers. Noticing outliers turns them from statistical pitfall into information, which could be interesting in its own right. Anyway, at this point I'm not worried about outliers, apart from correcting obvious mistakes (mine or yours) if such occur. The analysis presented will include a fair central value (such as median), along with a graphical look at how estimates varied around that -- out to the extremes.

Hmmm, an earlier post might have been spam-filtered? In it I mentioned that we have (by my rough count) 95 estimates so far, with a median of 3.0. Here's the distribution: Stem-and-leaf plot for extent (June SIO estimate by comment in ASI blog) extent rounded to nearest multiple of .1 plot in units of .1 0* | 000 0. | 1* | 013 1. | 578 2* | 00000012222444 2. | 555555778888888888999999 3* | 0000000111112222333334 3. | 5567778888888999 4* | 000011334 4. | 5* | 0

Lodger makes a good point above that my title might have confused some people about the monthly vs. 1-day minimum. I'll be more obvious next time. The SEARCH Sea Ice Outlook focuses on September monthly mean, so that should be our target for this crowd-source estimate too. I stated that somewhere in the original post but it not prominently enough. There's a learning curve for us all.

The SEARCH SIO folks are interested in hearing our results, if we choose to send something in. I count 60 responses so far, this experiment is generating real data.

@Rob Dekker (and anyone else) ... if you haven't voted yet, no hurry, the SEARCH deadline is June 7. I'd prefer not to revise estimates you've already sent in, just because that will confuse me, but in an extreme case let me know.

A-Team: "I wonder what drives this criminal evasiveness in so many climate scientists. To be frank, I think it is all about covering their butts." Hey, I love your visualizations but object strongly to this characterization of NSIDC and other scientists who aren't certain the end is nigh yet. They all see things changing faster than the models, are working to understand why.

Since the discussion is ranging widely ... I did a brief radio interview about this line of research yesterday. It's not specifically about the polar questions but explains more generally what we've found about knowledge/belief connections. http://www.nhpr.org/post/demographics-disagreement-climate-change

OldLeatherneck: "While it is encouraging to see that 45% of New Hampshire Republicans see the demise of Arctic Sea Ice as a serious threat, I don't think that those same numbers would hold true in a southern State." You're probably right that our New Hampshire survey would have come out differently in Texas, but New Hampshire seems to be a surprisingly good proxy for the US as a whole. The article Neven cites describes some other climate questions where US and New Hampshire responses are fairly close. The statewide New Hampshire surveys (which interview random samples of 500 people four times each year) provide a cost-effective testbed for new science perception and knowledge questions. For example, We've been repeating a basic question about climate beliefs every few months since 2010, now almost 7,000 interviews forming a unique tracking poll.

We have a new public-opinion survey paper in progress, data collection finished this week. Sixty percent of the large statewide sample (n = 1,688) think that Arctic warming will have "major effects" on the weather where they live. Interesting breakdowns by personal characteristics and beliefs. We're checking for daily weather effects on how people answer.

Also new in GRL and seasonally relevant: Recent changes in pan-Arctic melt onset from satellite passive microwave measurements L. Wang1, C. Derksen, R. Brown, T. Markus A new satellite passive microwave (PMW) melt onset retrieval algorithm based on temporal variations in the differences of the brightness temperature between 19 and 37 GHz is shown to be as effective as radar (e.g., QuikScat) measurements. The PMW technique shows improved melt estimates that are more closely linked to observed snow-off dates than previous studies. An integrated pan-Arctic (north of 50°N) melt onset date (MOD) dataset is produced by combining estimates on land and sea ice for the entire satellite PMW record. During the 1979–2011 period, significant trends of 2~3 days (decade)−1 to earlier MOD are mainly concentrated over the Eurasian land sector of the Arctic, consistent with changes in spring snow cover extent observed with visible satellite data. The variability and change in melt onset are largely driven by spring surface air temperature, with insignificant influence from low-frequency modes of atmospheric circulation. http://onlinelibrary.wiley.com/doi/10.1002/grl.50098/abstract

DMI, which often leads CT by a day or two, on 3/22 is about 300k below its high point from 3/17.