The crowd-wisdom challenge [Results]

This is the follow-up post on The crowd-wisdom challenge post I’ve posted on March 7th.

The real results just came in.. Out of the 145 people who have guessed during the last 7 days. Linas Medžiūnas and someone named ikajava were the closest to the correct answer which was….. drums… 35,905 seeds! Their guess was 35,000.

The experiment didn’t go as well as I hoped though ;). The Median guess was 8500 and the Average was 158,319. The margin of error from the median is 76.3%. An interesting note. If I would remove 5.5% of the high and low guesses I’ll get an average of 37,513 which is 4.5% error margin. If I’ll only remove the high 5.5% I’ll get an average of 35,375 which is 1.47% error margin.. Not bad.

So, is crowd-wisdom really a thing? Depends on how you digest your numbers 😉.

In the histogram below you can see the distribution of the guesses. As expected from the median, most guesses were around the 8500 or log(3.93). I did the X axis based on the log of the guess so both high and low guesses would make it into the graph. They Y axis show the number of people who’ve guessed in this range.

How did I count so many seeds?

The truth is I didn’t ;). I counted two large groups. One of 1060 seeds and one of 1430 seeds. I weighed the first group and got 8.0g. Then I weighed the second group and got 10.8g. So If I wanted to know how many were there in the second group I would multiply 1060 with 10.8 and divide it by 8.0 which gets me 1431 instead of the real number 1430. I’ve decided this was close enough and then carried on with weighing the rest of the seeds.