The essence of rationalist skepticism is the ability to allow evidence to change one’s mind, even when you were not particularly hoping to do so. The Litany of Tarski is a mental excercise intended to help cultivate the proper attitude towards truth, and on this particular occasion it runs like this:
If the block bot is contributing to suspensions on Twitter
I desire to believe that the block bot is contributing to suspensions
If the the block bot is NOT contributing to these suspensions
I desire NOT to believe that the block bot is contributing to suspensions
Now, I’ve been arguing for some time that mass-blocking on Twitter is most probably influencing Twitter’s suspension algorithms resulting in permanent suspensions for those targeted by the Atheism Plus Block Bot, and what is more I’ve been pressuring the bot’s creator (James Billingham) to fully disclose his dead blocks data. This he has not quite done, but he may have provided enough information to settle the question of whether the bot has been significantly impacting Twitter’s suspension algorithms relative to users on the block list:
His argument is essentially that since the rate of increase of dead blocks has been essentially constant in the weeks before and after the app was suspended by Twitter around the beginning of this month, we may reasonably assume that the accumulation of dead blocks on the block list is due to the background rate of accounts being suspended without any significant contribution of the block bot itself. Had the bot been a significant factor, we should have expected the rate of increase of dead blocks to drop at the time of suspension. Of course, this is assuming that the baseline rate of increase of the overall block list has been held fairly constant over the last six or seven weeks, and James does not provide this information. I’d also be interested in an explanation of the sudden spike from 62 to 70 dead blocks around the 13th of this month, not to mention those bizarre off-the-chart spikes.
All that said, I have to commend James for going to the trouble to share these data. They provide some prima facie evidence that the bot is not significantly driving Twitter’s suspension algorithms, and I am (…pauses to recite Litany of Tarski…) happy to accept that if it turns out to be true. The bot may not be the single most effective way to carve out a totally safe space on Twitter, but it certainly would be nice to know that it is not actually freezing peaches.
Of course, the real test will be when they turn the bot back on. We can all look forward to that, in the hopes that James will update his chart and close the book on this issue.
