forum.starcraftai.com

[quote name='Taranok' date='29 August 2012 - 11:58 AM' timestamp='1346209105' post='11074']
I don't think AIUR ever scouts.
[/quote]

What? o_O
You should re-watch replays carefully.

[quote name='krasi0' date='01 September 2012 - 12:34 AM' timestamp='1346427255' post='11082']
BTW does anyone here know which of the bots don't use I/O so I add them to the ladder?
[/quote]

It's written on the tab "Entrants" of the google doc results: https://docs.google.com/spreadsheet/ccc ... xcUE#gid=8
However, Dave only mentioned us by email AIUR, SCAIL, BBQ and UAlbertaBot. Dunno why.

Yeah, got the list. I am now trying to add support for I/O on the ladder. It's kinda tricky though. Let's see if it will work out.

[quote name='Kae' date='02 September 2012 - 12:15 AM' timestamp='1346562915' post='11083']
[quote name='Taranok' date='29 August 2012 - 11:58 AM' timestamp='1346209105' post='11074']
I don't think AIUR ever scouts.
[/quote]

What? o_O
You should re-watch replays carefully.
[/quote]

Sorry, I wan't aware of how the replay vision works in BW.

How do bots pick 3rds? Usually they seem not in the normal position.

[quote name='Taranok' date='04 September 2012 - 06:17 AM' timestamp='1346707057' post='11087']
How do bots pick 3rds? Usually they seem not in the normal position.
[/quote]

I can tell you for AIUR: it takes 3rd, 4th and so on, if no expands have been taken for a while (wrt the current mood and a random parameter) OR when we launch an attack.

[quote name='Taranok' date='28 August 2012 - 06:58 PM' timestamp='1346209105' post='11074']
I was kinda disappointed in the UAlbertaBot vs versus Skynet games where UAlbertaBot just overreacted to a scouting probe.
[/quote]

This was a but I introduced when I added the combat simulator to the bot. It was supposed to send a single worker to defend against a scouting worker, however this cap got removed in error. I only had about 2 weeks to work on the bot, and only tested it against the default ai, so I never caught the bug. Hilariously though, it ended up being a good thing because it beat Aiur's cannon rush strategy, which I would have lost to otherwise

Also, I was well aware of the combat oscillation problem in the bot, but didn't have a clever fix for it so I left it in. I have since added a timer to the retreat code so that we can't reverse our decision more than once every 5 seconds.

[quote name='Kae' date='01 September 2012 - 09:15 PM' timestamp='1346562915' post='11083']
[quote name='Taranok' date='29 August 2012 - 11:58 AM' timestamp='1346209105' post='11074']
I don't think AIUR ever scouts.
[/quote]

What? o_O
You should re-watch replays carefully.

[quote name='krasi0' date='01 September 2012 - 12:34 AM' timestamp='1346427255' post='11082']
BTW does anyone here know which of the bots don't use I/O so I add them to the ladder?
[/quote]

It's written on the tab "Entrants" of the google doc results: https://docs.google.com/spreadsheet/ccc ... xcUE#gid=8
However, Dave only mentioned us by email AIUR, SCAIL, BBQ and UAlbertaBot. Dunno why.
[/quote]

I sent emails to those who had valid output in their IO directory.

[quote name='abcdefg' date='28 August 2012 - 03:46 PM' timestamp='1346197618' post='11072']
I was running ualbertabot and it looks impressive it has dynamic build orders it recalculates on the fly and does some kind of simulation on whether it can win or not which seems to be kind of accurate. Here is its big flaw that dooms it as I saw it figuring success and failure over and over again when a few marines were near a bunker.
It constantly retreats and then attacks and never ends up doing anything while it loses its units for no reason. Also it still doesnt gather all its zealots together before attacking but sends them in to suicide one or two at a time. I think if it just engaged better and made a more definite decision it would have done much better just from what I noticed in a game or two.
[/quote]

See my above post with regards to the oscillation.

I also stupidly forgot to add bunkers to the combat simulation. Their damage was assumed to be 0 and so they never factored into the simulation. Oops