Following the recent embarrassing trouncing of my once-bold-seeming prediction by that darn reality, I hesitate to prognosticate again.
Can we predict the future of The Thread? Here I try out a new approach, one involving multiple predictions and a bracketing of possibilities. This is what the experts do, I guess, when predicting future populations and climates and shit like that.
So the data analyzed are shown in the "reality" link above. I fit polynomial models to three different data sets, choosing the model to display in each case by the wholly arbitrary and rigorously indefensible strategy of adding terms until the r2 stopped increasing in the third decimal place. I then forecasted the models into the future as far as February 24, the one-year anniversary of the Thread.
The models are:
Light blue: Fourth-order polynomial fit to the entire data set, from day 1 of The Thread.
Bright green: Fourth-order polynomial fit to the data I consider representing the Thread's "modern era," post-dating the obvious correction of mid-September 2009.
Red: Third-order polynomial fit to the most recent data, post-anastomosation.
Oh, how many? These models predict that the comment count on Feb. 24 (the 1-y anniversary) will reach somewhere between 25,000 and 40,000. So, a lot. Personally, I hope it turns out to be the red one, because then we can use the term "hockey stick."
The increasing predictions as data are removed from the model simply emphasizes the recent acceleration of commenting rates. Something has to give pretty soon. An asymptote would break the Internets.
UPDATE 2/1/2010: I was goaded into essaying a more precise (and therefore more likely to be WOTI) prediction here: 50,000 comments at 3 am on 2/28/2010. That's the hockey-stick prediction, so anything less than that and I will claim success.