Would an AI-only mod get broken by patches? I've been playing "Entropy II v1.0" since the regular computers are kind of lacking, but the computers seem to break often...

There are a few changes in the dynamics of A.I.'s through different patches. However, these changes only make them less responsive, and nothing should "break."

Well, what happens is the computers will do really stupid things (protoss that builds only zealots the whole game, zerg that never attacks anyone, terran that makes a whole one barracks and three supply depots... and nothing else)...

Yeah, they'll do that. AI's apparently work the best on patches 1.10 and before.

Any particular script or matchup? And are you playing the same maps as you were prior? Entropy Ai of course has NO island blocks.. i.e. if the Ai can't get to you via the ground at games start, it will just do nothing. Entropy AI isn't perfect, and there are a few problems with some of the scripts. Not all of the scripts that make up Entropy II are equal. Some are definitely a bit more prone to fizzle out. For the most part it runs pretty well on Standard maps tho. Upload or send me a replay of a poor performance if you would too as I'd like to see what might be bogus.

This is only speculation and not based on concrete evidence/testing. Our own Ai vs. Ai tests show that while results may vary for isolated instances, the overall performance is not affected from version to version. Terran incomplete building bug is about the only real advantage to a pre 1.11 SC I can see.

AI's are more responsive. They expand more and attack more on old versions.

I'd like to see the replay or collection of replays that demonstrates this. I need someone to SHOW me this. Nothing has changed in aiscript.bin itself, and I don't think anything has changed in the hardcoded internal AI engine?

I wrote an AI once that was purely an expansion script to test out why my mod's AI was not expanded and it turned out the likeliness of an AI expanded increased in 1.10.

It was hard coded engine edits, You can go ahead and do some tests, like, Play 5 games? with like you Vs 1-3 comps, With an AI mod or Not, (I would do it with a good AI mod), And then play the same 5 games, when downgraded to 1.10, you Should notice a huge improvement in the AI.

Well I will agree that the Ai plays "differently" from version to version, but I wouldn't use terms like "huge improvement" nor "more responsive". I just watched the first 3 replays from our AI vs. Ai pack in both 1.10 and 1.16.1 using default SC AI. @ 45 mins a piece, this equates to nearly 4 hours of gameplay. The final tally shows that in some of the replays, the AI had a better overall score in 1.10, but nothing outrageous. Another replay demonstrated that 1.16.1 proved to be the better game .. i.e. higher scores more units produced etc. So if there is a better performance from 1.10 it is or marginal importance and somewhat conditional such as the pure expansion script A-of-st was testing and will vary from game to game.

Regardless, I will add a footnote to our custom Ai page as well as BWAI Gold that states performance may vary with each SC version, and that 1.10 is widely believed to offer the best performance.

So this covers "AI script only" mods. Now let's all go back to Stardraft and throw Firegraft in the trash! OR commission DofA to make a 1.10 Firegraft. OR force him to make the code public so users can modify it for 1.10. OR we just forget about this whole thing and move on. 1.10 has no minimap preview, has no player color chat, replay speed is broken, etc.

Let's just kindly ask Blizzard to fix the A.I.