I had expected there’d be around 8 bots in the clique and around 50 bots in total (though not that many sillyBots). But I never imagined we’d rise from 15% to more than 50% of the pool as early as round 10!
The cloneBots are not even attacking the other bots yet. Until round 10, they often back down to 2 in case of 3-3, and they play tit-for-tat in case of 3-2. From round 10 to round 60, they’ll get progressively more greedy.
Would we fare better, worse, or the same if the rise in greediness was faster? I wanted to change it to 10->30, but ultimately didn’t.
I had thought there would be more attackers in the initial pool. I spent a lot of time fine tuning our behaviour against them (folding in the early rounds, then maintaining 3 more and more often later). Seems like it was mostly a waste of time.
On the other hand, the code to exploit 0-bots and the like was not wasted. Yum yum.
Now that the most easily exploitable sillyBots are out, it’s gonna be a race with Multicore’s bot. While we try to smother all the outsiders, Multicore will allow cooperators to survive while gaining score from them. If they survive long enough, we’ll be the ones smothered.
I think there’s a 70% chance we eliminate all non-clones/mimic by round 60. Even if we do, I expect Multicore to be bigger than the aggregate of the 2 next biggest at round 90 when the second phase begins (70%).
Wow!
I had expected there’d be around 8 bots in the clique and around 50 bots in total (though not that many sillyBots). But I never imagined we’d rise from 15% to more than 50% of the pool as early as round 10!
The cloneBots are not even attacking the other bots yet. Until round 10, they often back down to 2 in case of 3-3, and they play tit-for-tat in case of 3-2. From round 10 to round 60, they’ll get progressively more greedy.
Would we fare better, worse, or the same if the rise in greediness was faster? I wanted to change it to 10->30, but ultimately didn’t.
I had thought there would be more attackers in the initial pool. I spent a lot of time fine tuning our behaviour against them (folding in the early rounds, then maintaining 3 more and more often later). Seems like it was mostly a waste of time.
On the other hand, the code to exploit 0-bots and the like was not wasted. Yum yum.
Now that the most easily exploitable sillyBots are out, it’s gonna be a race with Multicore’s bot. While we try to smother all the outsiders, Multicore will allow cooperators to survive while gaining score from them. If they survive long enough, we’ll be the ones smothered.
I think there’s a 70% chance we eliminate all non-clones/mimic by round 60. Even if we do, I expect Multicore to be bigger than the aggregate of the 2 next biggest at round 90 when the second phase begins (70%).