Create your own AI (a parametric contest)

[kyrub]

Staying local means that only a small part of the code is expected to be changed by volunteers. You provide explanation of which part and what is does now. I suggest you start with 1 part only and go from easy to complex.

Hm. Maybe. It can be tested better, surely.
(I have some other concern / idea here [changeable AI game objectives], influencing local parts, but it can really wait for now.)

Re: opening IDA database
I meant publishing the disassembled exe with my notes online, like I did with MoM, which ILSe used to make his Tweaker. (At this point it means half of the notes are in czech, more than 50% of the code is not analyzed at all, graphics, sound and many other technical things. Lot of wrong understanding, I was rookie at the time.)

Re: AI in assembler
Major improvements? Thanks for generosity, but I disagree, 1.40m as far as AI goes is almost nothing, it is unfinished business. The much better version had to be dumped at the time, because of a bug that infected all my work. It was only afterwards that I started to make extensive internal documentation to avoid this. MoO assembler AI can be much better. Hopefully, we find some energy to achieve it, I certainly do not have the time and energy and mind to work on it alone now. (I think that through "parameters" alone, the AI will instantly become much bigger threat.)

Re: Vanshilar
It's great you'd like to try it! Welcome to the team.
1. I will explain how AI works in 2nd post of this thread, when we are ready to go.
2. Shortcomings... play the game! And: test your code changes. (I actually think that identifying the shortcomings may be a part of the contest. Another way is to purely start out of current algorithm and think of best way to use it... That's how I do it.)

[WhiteMage]

Major improvements? Thanks for generosity, but I disagree, 1.40m as far as AI goes is almost nothing, it is unfinished business. The much better version had to be dumped at the time, because of a bug that infected all my work. It was only afterwards that I started to make extensive internal documentation to avoid this. MoO assembler AI can be much better.

I mean major overall improvement, which includes a small portion on AI.

I think that through "parameters" alone, the AI will instantly become much bigger threat.

Yes, I think so too.

start out of current algorithm and think of best way to use it

Yes. You pick algorithm and we think and try to improve it.

[Zieman]

Correct me if I understood this wrong:
You want us to modify a set of parameters the AI uses in its decision making, and report what effect those changes had on the AI?
Using your latest patch?

IMHO we should start with a constrained set of parameters, or in other words: set of parameters that affect a distinct aspect of the AI.

[kyrub]

Correct me if I understood this wrong:
You want us to modify a set of parameters the AI uses in its decision making, and report what effect those changes had on the AI?
Using your latest patch?

Yes. But before reporting, you should test your modifications yourself and improve them.
After a given time, we collect the results, explain the ideas behind our changes - and we agree on tests showing how different concepts work under specific conditions (game stage / race / war-peace situation etc.).

IMHO we should start with a constrained set of parameters, or in other words: set of parameters that affect a distinct aspect of the AI.

Another vote for this, noted. That's reasonable.
We will start with the AI Research area, the one that I used as an example. The data to the contest, along with some utilities will be provided in next few days. I will make a special set of exe with one extra hardcoded alternative, for you to decide. Probable start time (when everything is ready) is next monday.