While I agree most "antiban" is useless, there are a number of techniques which actually have potential to reduce bans (these do not require "neural network machine learning fuckery"). Most antiban in public scripts is there because the users demand it, not because it is effective in reducing the chance of bans. Much of these traditional techniques may actually increase the chance of a ban by increasing the amount of routine and observable inputs the bot is performing. Over time, these "antiban" techniques could actually provide more evidence of botting.