That O only took off once other variants were eliminated suggests a rock-paper-scissors relationship. But I suspect O only lost early on because parts of it’s ruleset are too vulnerable to the wider bot population. So which rules was O following most when it lost/ against early opponents, and which rules did it use to beat I and C4?
O is tit-for-tat with 3 defections at the end (TFT3D). It has extra rules to help with weird players and randoms. Those don’t matter much once the weird guys are gone and it becomes a game between the tit-for-tats so ignore those. There
I is TFT2D and roughly speaking so is C4. All the next players who are slightly behind the TFT2Ds are approximations of TFT1D (with random killer noise that we can again ignore once the real fight starts) then there are plenty of TFT0Ds as well.
Consider the following match ups:
TFT3D vs TFT2D: TFT3D earns 300. TFT2D earns 295.
TFT3D vs TFT1D: TFT3D earns 300. TFT1D earns 295.
TFT2D vs TFT1D: TFT2D earns 301. TFT1D earns 296.
Before that, while the pure tit-for-tats (TFT0D) are alive we of course also have * TFT0D vs TFT1D, TFT2D and TFT3D: TFT0D earns 297, 296, 295 respectively while the greater cheaters earn the 302, 301 and 300 that we expect for when they fight any nicer TFT. This helps explain why O started slightly behind even some TFT1Ds at the start but soon becomes irrelevant.
The pool starts up TFT1D >> TFT2D > TFT1D based on the number of variants submitted. All three of them feed early off weird strategies and TFT0Ds. The TFT2Ds then take off, feeding off the large population TFT1Ds. The TFT3D can feed off TFT1D too but not quite as effectively as the TFT2Ds. Once the TFT1Ds (and TFT0Ds) are eliminated from the population the TFT3Ds can slowly begin to catch up and overtake their more optimistic counterpart.
In any population made up of many TFT(n)Ds a TFT(n+1)D will thrive. Unfortunately for him, however, if n is too great and there are enough significantly more cooperative TFTs around it will be eliminated before it gets a chance to shine. There is no paper-scissors-rock relationship TFT1D can score more than TFT3D but never by beating anything that can beat anything that can beat anything (etc) that can beat a TFT3D. A more optimistic TFT only beats a less optimistic TFT if it has a sufficient clique of similarly or more optimistic TFTs and the rival is just too pessimistic and has nobody to feed off effectively.
O is tit-for-tat with 3 defections at the end (TFT3D). It has extra rules to help with weird players and randoms. Those don’t matter much once the weird guys are gone and it becomes a game between the tit-for-tats so ignore those. There
I is TFT2D and roughly speaking so is C4. All the next players who are slightly behind the TFT2Ds are approximations of TFT1D (with random killer noise that we can again ignore once the real fight starts) then there are plenty of TFT0Ds as well.
Consider the following match ups:
TFT3D vs TFT2D: TFT3D earns 300. TFT2D earns 295.
TFT3D vs TFT1D: TFT3D earns 300. TFT1D earns 295.
TFT2D vs TFT1D: TFT2D earns 301. TFT1D earns 296.
Before that, while the pure tit-for-tats (TFT0D) are alive we of course also have * TFT0D vs TFT1D, TFT2D and TFT3D: TFT0D earns 297, 296, 295 respectively while the greater cheaters earn the 302, 301 and 300 that we expect for when they fight any nicer TFT. This helps explain why O started slightly behind even some TFT1Ds at the start but soon becomes irrelevant.
The pool starts up TFT1D >> TFT2D > TFT1D based on the number of variants submitted. All three of them feed early off weird strategies and TFT0Ds. The TFT2Ds then take off, feeding off the large population TFT1Ds. The TFT3D can feed off TFT1D too but not quite as effectively as the TFT2Ds. Once the TFT1Ds (and TFT0Ds) are eliminated from the population the TFT3Ds can slowly begin to catch up and overtake their more optimistic counterpart.
In any population made up of many TFT(n)Ds a TFT(n+1)D will thrive. Unfortunately for him, however, if n is too great and there are enough significantly more cooperative TFTs around it will be eliminated before it gets a chance to shine. There is no paper-scissors-rock relationship TFT1D can score more than TFT3D but never by beating anything that can beat anything that can beat anything (etc) that can beat a TFT3D. A more optimistic TFT only beats a less optimistic TFT if it has a sufficient clique of similarly or more optimistic TFTs and the rival is just too pessimistic and has nobody to feed off effectively.
Thanks for the analysis and the TFTnD nomenclature, both of which I have expanded upon in a post.