archive by month
Skip to content

new Steamhammer plan: win

In AIST S1, Steamhammer got to play only 5 games and lost 4 of them. It was a sad showing, though the field was strong. CIG is coming soon, and Steamhammer is registered. Then AIIDE is after that. Also, as I have mentioned, Steamhammer is unbalanced in its play skills, much stronger in strategy and macro than in tactics and micro. Looking at results in the past few months, I’ve concluded recently that it is so unbalanced that improving strategy no longer helps at all. When I fix a strategy weakness, even one which demonstrably loses games, the benefit is so small that there is no detectable change in skill, either objectively in elo or subjectively in my judgment. The imbalance is so extreme that nearly all losses, even in games affected by strategy problems, are due to mistakes in tactics and micro.

That means that if I work on the opponent model and opening selection as I planned, I won’t be able to tell if am doing a good job. No matter what I try, it will hardly affect results. It also means Steamhammer is at risk of getting whomped in CIG and AIIDE.

I have made a bunch of bug fixes recently (thanks Bruce!), so here’s my new plan. I’ll release 1.4.4 shortly with source, and to save time I’ll skip releasing the source of 1.4.3. (It takes more effort than you might imagine.) 1.4.4 includes bug fixes (one of which hasn’t been mentioned here), plus a start on an influence map for tactics, plus a zerg strategy fix that may save 1 game in 200 and a zerg tactics change that is more likely to save 1 game in 15 or 20. I think it’s all stuff that people should have.

Then the next public release will be after the CIG deadline next month (CIG releases source itself), and the following one may be after the AIIDE submission deadline in October. I will buckle down and make the changes that improve play the most in that time.

Trackbacks

No Trackbacks

Comments

Antiga / Iruian on :

The difference between Steamhammers and say zzkbot or purpleswarms 4 pool (probably the best one on ladder) is pretty large in actual performance. (2:15-2:20 depending on map) w/ a 1 drone scout to ensure perfect arrival time on 4 player maps. Not having a real 4 pool enabled is also costing the bot a ton of free ladder points and wins with the new learning you have implemented imo. If you want to win... win by any means necessary which means 4 pooling when required. I bet there is 10-15% winrate to be had in that area or more.

Jay Scott on :

A logical idea. With learning, the weights should be more even, and I may not have taken it far enough. I rejiggered the numbers so that all the fast pools will be somewhat more frequent, and the different flavors are around equally likely, with adjustments for map size.

Tully Elliston on :

> Finally going to improve SH micro & tactics

https://orig00.deviantart.net/44bd/f/2008/116/c/1/hell_its_about_time_by_biohazard_apocalypse.jpg

Marian on :

So I have watched a few of recent SH games and there is one thing I have noticed that might improve your rating quite a bit.
Overdroning.
Ideally you want really little drones in ZvZ low drone count ZvT (but with high gas mining) and average drone count ZvP.
Of course you can make more especially in ZvP but I consider that quite a advanced skill to do right.
Actually my bot uses some min-max drone formula something like this: minmax(16, mineralCount*1.5+gasCount*3, 70)
where the constants differ for each matchup.

Jay Scott on :

Agreed, economy control is lacking and it severely affects the army size. Economy choices are usually not too bad in ZvZ, quite weak in ZvT and ZvP.

Dan on :

I believe you have some dynamic selection of how many Drones you can afford to make based on the opponent's army. How does your implementation work?

Jay Scott on :

Steamhammer has multiple mechanisms that independently adjust the drone count. There is an upper limit based on the mineral patches and geysers taken. When under immediate threat of attack, it makes army over drones, but still makes drones at a low rate so it doesn’t get stuck when it is contained. There is also “compensation” for static defense: Steamhammer’s static defense makes it safer, so it can safely squeeze in more drones, and the opponent’s static defense makes the opponent less dangerous, ditto. Otherwise drones are made at a fixed average rate depending on the matchup, which is generally a poor idea. The system tends to make too few drones early (though it depends) and too many in the middle game.

jtolmar on :

If my calculations are right, it takes 7 minutes for an extra drone to mine enough resources to pay for itself and a hatchery, morph into a hatchery, and produce its replacement larva. So unless the extra mutalisk or zerglings cause/prevent a significant economic loss within 7 minutes, or the game ends within 7 minutes, the extra drone is strictly worth it. Which lines up pretty well with the way the matchups are played - the extra zerglings really will swing the matchup within 7 minutes in ZvZ, and are much less likely to in the other matchups.

But since bots are kind of bad at killing each other, I'm not as sure over-droning is as big of a weakness as it would be against humans. If all matchups are less lethal and swingy, they all become less like (properly played) ZvZ, and economy becomes more important in all of them.

Marian on :

There are many things to consider.
For example you want to hit protoss before he has high tech units or static defences set up on his expansions.
Also you want to build up units before enemy timing hits e.g. +1 zealot rush.
If your bot can properly scout and guess what you opponent is doing and when the timings are coming then yes you can properly switch from making 100% drones to 100% combat units.
Otherwise you should compromise by making less drones in favor of combat units.

Add Comment

E-Mail addresses will not be displayed and will only be used for E-Mail notifications.

To prevent automated Bots from commentspamming, please enter the string you see in the image below in the appropriate input box. Your comment will only be submitted if the strings match. Please ensure that your browser supports and accepts cookies, or your comment cannot be verified correctly.
CAPTCHA

Form options

Submitted comments will be subject to moderation before being displayed.