I’m going to come at this from a ganker perspective…
If you give me X amount of time before Concord shows up… I can do the math and figure what I need to do the job, how much it will cost, and if it is worth it. Predictability = Profit.
Don’t keep doing the same for botters. Sure, ban in waves, but vary them, make them not know how long it will be. Right now it is months, so they look at it and go, I can make X… X is worth it because it will be months before anything happens; they can plan ahead. But if this time X is 2 weeks, and next time it is 12 weeks or 8, now they don’t know, and be very public about it when the bans go out. It isn’t like they can suddenly spawn a fully useful account in an instant, it all takes time and/or money (or hacked accounts) and this is an investment on the botters part… set an environment where they are unlikely to get good ROI.
And yes, the bot code makers will adapt, but they probably already have iterations of their programs mostly ready to go. All based on how they figure you will catch them this time. It is unlikely the problem will ever go away, but at least, can we add more uncertainty in the plans?