An early RadioLab episode indicates that one of the ways in which emergence reveals itself is through contests similar to the sundial poll, in which a large group of people are asked to estimate a given quantity. According to the program, a group of people that are asked to predict the weight of an animal or the amount of jellybeans in a jar are more accurate collectively – via the mean estimate – than any individual by him or herself.
Having now determined the date and time at which the sundial reached the “never” location, September 28 at 4:17 p.m., I decided to test whether some element of emergence would present itself through the variety of guesses posted by some 70 students.
In order to do so, I gathered only those entries that offered a specific date and time (57) and initially averaged individually the months, dates, hours, and minutes posited by students. Months were averaged based on their position in the calendar year (i.e., January = 1, February = 2, etc.), and the dates, minutes, and hours were counted simply as their given numbers (i.e., 4 p.m. = 4). The average guess based on these calculations: October 13 at 4:20 p.m.
The date, of course, was incorrect (and would have been more incorrect had I tallied the entries for January and February as 13 and 14, as opposed to 1 and 2, because most estimates understood January as coming after December and not well before it as the numbers used would suggest). However, the mean time of day was estimated within three minutes of the actual time.
I had been trying to understand for the past two hours the substantial difference between the accuracy of the time estimation and the high inaccuracy of the date estimation. Emergence, it seemed, had and had not presented itself, which perhaps discredited the accuracy of the time estimation. There was a part of me that felt as though the recreation of the jellybean jar contest was only possible in estimating the time and not the date here, simply because the full range of times were available to choose from (i.e., 12 am to 12 pm), and likely only post-September times may have been deemed possible for the date because of the location of the sundial at the beginning of the contest.
However, what I instead found on further investigation was that only by the calculations above for time could emergence be said to appear. When I realized that in fact two a.m. times were included in the guesses, I recalculated the hours based on military time (i.e., 4 p.m. = 16), and this proved to decrease the estimation’s accuracy by about half an hour. Hoping then my main fault was in my broader means of calculation, I reduced everything to minutes to find an average based on this assumption: beginning with New Year’s Day at 12:00 a.m. as the zero minute mark, I calculated how many minutes passed until the guesses (i.e., November 26 at 3:45 p.m. = 513,585 minutes, etc.) and averaged those amounts. Sure to have finally found the solution and to have discovered the intelligence inherent in emergence, I was instead disappointed to find that the accuracy further decreased. The average was now October 14 at 5:19 a.m., and when only time was taken into account, the average guess was for 3:20 p.m.
This, of course, raises a number of questions: Is the sample size too small? Were there too many biases imbedded in the contest that restricted an accurate mean date and time (i.e., known presence of the sundial, submissions of comments without any thought to meet class requirements)? Or is the emergent property of which the RadioLab program spoke incorrect? Can emergence really show itself here, if at all?
Although I am unconvinced of anything, I will finish the post with a final thought: that perhaps the decreased accuracy of the time estimation occurred with more accurate means of estimating time because we all think more generally, qualitatively, than always in military time or in the amount of minutes that have passed since New Year’s Day. Did we make the estimates of the time based on hour and then minute, as my initial calculations unintentionally recreated? Perhaps. Or, perhaps the key to finding the most accurate prediction lies in something else entirely: we all just need to be Charles.