A.I. too overpowered

Just stop the bubbling about APM.

An AI can only take any advantage from higher APM, if the players was APM capped already. While this is true for Starcraft in so many ways, it isn't so much with PA.

For once, there are only few microsploits possible, and about half of them are fixable by adjusting the exploited game mechanics, and the other half is supposed to be executed by the individual unit / platoon AI by default anyway.

Micro is something a proper AI enemy doesn't need to know about, since this isn't supposed to be exposed to any player in PA. The AI only needs to how to put the effects of the implicit micro into relation to the bare unit stats which are used for the computation / lookup.

The actual strength of the AI is determined by the quality of the internal model, regardless of whether the AI uses historical data or live computation, because it can only win if it makes the right assumptions about the full (due to FoW unknown) gamestate and the implications of high level strategy. Breaking down the high level strategy into actual commands in subroutines is still not trivial (as you can see at the retreat issue in SoaSE), but there's also not much gain.

Even for base construction / multitasking, the AI hasn't that much of a headstart. Currently, the player is mostly limited by the controls, yet thats an issue which can be resolved by the means of area commands, first order commands, build templates and smart formations / platoon AI. The more abstract the controls become, the easier it becomes to schedule a full intention ahead rather commanding individual units.

It's basically the same the AI does, as it actually also only knows a few abstract routines, which are then broken down into basic commands upon execution.



Apart from the first few minutes, the game's economy is rather forgiving when it comes to missing a timer, thanks to the streaming economy, even if the AI played perfect, it could only perceive a minor advantage. That is, if it made the correct high level decision which tech tree to follow. Calculating perfect build order is just the same as every human does.

Intelligence is still very much broken, too many attack vectors with no meaningful prewarn system, no persistent information, so the AI would be in favor, since is better at registering and classifying potential threats which would go unnoticed to any human player. The basic flaws however affect both the AI and humans alike, and the remaining issues (mostly lack of persistent informations) are quite easy to solve. In this case it could even be useful for the human to get access to the threat detection system of the AI, since it supposedly already reduced the amount of data by merging contacts into threats.

Battles are currently critical, far too much micro due to dodge mechanics being encouraged by both off scale damage to hitpoint ratio, as well as off scale turn rate to movement speed ratio. Oh, and the not remotely solved deathball issue.
In this case, neither the AI nor the player is in favor, since both player and AI are supposed to have access to the same unit AI, which again removes the need of explicit attendance, so for both AI and human player the decision is reduced to choose the composition and location of the troops, but target selection, advancing, falling back and alike decisions are made by the unit itself.


So in the end, the performance of neither the AI nor the human does need to be determined by the corresponding APM values. While you still have a lower cap below which you won't be able to carry out your full strategy, you don't win much by increasing the APM even further - because all the micro level stuff could as well be executed by an simple assistance system which gives at least the same performance.

In PA, there is only one limitation to this: The lower APM cap scales with the size of the playable gamefield. So if playing multiplanetary, FFA, then a single human might no longer be able to keep up, not even with the basic assistance systems.
But then again: There's coop mode for that. The AI is capable of scaling up with ease, it doesn't need a coop partner.

Remember, this is not Starcraft. Client side mods or smart units are not considered cheating - since the primary goal isn't to deal with the limitations of the game client, hardly doing more than executing a premade build, but rather to execute a responsive, high level strategy.
(Well, I must admit that it is possible to play in a strategic way in Starcraft, but unless you are a bot, you are just not going to have the time to focus on it. Especially not when you can cheat yourself a much bigger advantage with micro exploits, and the maps disallow complex maneuvers.)

The key to battling AI, in its current state, is find them quick and harass them a lot. If you give them time to build they will mop the floor with you.

While they don't cheat they are still able to do far more actions/minute and they always have complete awareness of what they have scouted out.

Oops, me thinks Sorian has picked up a challenge. LOL