Doing Science! Open Thread Experiment Results

Early in the month I announced that I was doing an experiment: I was going to start two Open Threads in January (one on the 1st, and the other on the 15th) and compare the number of comments on these threads to those of other months. My hypothesis was that having two Open Threads would raise the overall number of comments.

The reason for this experiment was recent discussions regarding how useful threads such as these were quickly buried. Well, the experiment is over now, and here are the results:

I did a search for Open Threads, and entered all the monthly ones I could find into an Excel spreadsheet. I made them into a graph, and I discovered an anomaly. There was an 8-month timespan from February 2010-September 2010, in which the comment counts were extremely high (up to 2112). Many of these threads had 2, 3, or 4 parts, because they were getting filled up.

I wasn’t around LW back then, and I don’t feel like reading through them all, so I don’t know why this time period was so active. My current hypothesis (with P=.75) is that anomalous time period was before the Discussion section was created. I’m sure I could look it up to see if I’m right, but I bet one of the long-term LWers already knows if this is true or not, so I’ll crowd-source the info. (Comment below if you know that I am correct or incorrect in my hypothesis.)

Now for the data:

The January 1-15, 2012 thread had: 122 comments
The January 16-31, 2012 thread had: 236 comments

For a grand total of: 358 comments in Jan 2012

The average Open Thread had: 448.6 comments
The median Open Thread had: 204 comments
The average OT of the past 14 mo’s: 126.5 comments


So overall, the January thread had LESS than the average monthly thread, but more than the median.

IF however we look at the past 14 months (which was the end of the anomaly), then the January 2012 Open Thread had almost THREE TIMES the average.

My original hypothesis had probabilities assigned to various increases in comment rate, but I was way off because I didn’t at all think it would shrink (if we include the anomaly) or that it would be 300% bigger (if we don’t)

Here’s a handy-dandy chart, because everything is better with pictures in!