StarCraft AI Competition Announced 200
bgweber writes "The 2010 conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE 2010) will be hosting a StarCraft AI competition as part of the conference program. This competition enables academic researchers to evaluate their AI systems in a robust, commercial RTS environment. The competition will be held in the weeks leading up to the conference. The final matches will be held live at the conference with commentary. Exhibition matches will also be held between skilled human players and the top-performing bots."
What could possibly go wrong? (Score:3, Funny)
Re:What could possibly go wrong? (Score:4, Funny)
Re:What could possibly go wrong? (Score:4, Funny)
at one point the AI will realize that it's far easier to beat the human by hacking in to military computers and nuking the player.
Then the human players will black out the sky.
Re: (Score:2)
And the computers will... fight in the shade?
Re: (Score:2)
10 PRINT AWOO
20 GOTO 10
Re:What could possibly go wrong? (Score:5, Funny)
*Nuclear launch detected*
Re: (Score:2)
N-N-N-N-N-N-N-N-N-N-N-Nuclear launch detected
especially if you are playing red.
Re: (Score:2)
You mean, "worse" balanced...
The nukes in SC are "tactical" nukes, and they have to be laser guided by an actual person on the ground. Realistically, they shouldn't be too powerful.
Now, the real question is: The psi limit is explained in the manual (for the terrans), that the command center holds 200 people. But if they die on the battlefield, you can still reach 200 later on. Where do the new people come from?
Also, where do dragoons come from if you didn't build any zealots first?
Re: (Score:2)
The introduction of the Medics explains that :)
Re: (Score:2)
I read that in the actual Starcraft computer voice in my head. I think I played that game too much, even though it's been a while.
Re: (Score:3, Funny)
I heard it in Protoss
Re:What could possibly go wrong? (Score:5, Funny)
*Nuclear launch detected*
I remember playing Starcraft at a LAN party after I figured out how to rip the sound effects and voices out of the Starcraft data files. I'd intentionally play with the sound on (no headphones), wait just enough time so that it was believable yet frighteningly early, alt-tab over to a WAV player, then play the sound for "Nuclear launch detected" and watch people frantically scan their bases. It only works once, so use it wisely.
Re: (Score:3, Insightful)
Yes, but then that means they will be less likely to check that second time when you really do launch nukes at them. That's when they end up getting humiliated.
Re: (Score:2)
Re: (Score:2)
"Unit lost."
Brood War (Score:2, Offtopic)
Re:Brood War (Score:5, Funny)
Instead of an AI that can win at Starcraft, maybe they ought to try to build an AI that can finish Starcraft 2.
Apparently, that's a much greater challenge.
Re: (Score:3, Interesting)
They ought to release it with dedicated servers and LAN play and roll out Battlenet 2.0 when it's ready, if that is indeed the case.
Re: (Score:2)
It needs an AI that can handle the following.
Island maps. Against a Protost with cannons bordering the island 3 rows deep. Most human players can get past that. For the computer it always kept me safe. Of course there was one time I played stricly defensive and have the entire island filled with cannons. It took a good player 45 minutes to get to me. (and killed a medium level player) who was helping.
An other blood bath was a map No Gas. Where there was a river and a small bridge to cross. All with a
Re: (Score:2)
Re: (Score:2)
Island maps. Against a Protost with cannons bordering the island 3 rows deep
A human can get past that simply by using heavy airborne artillery, armed with an escort. It would basically need to hit the supply limit as well.
The reason standard Starcraft AI players can't crack that is because they don't normally last long enough to break through that. They usually take out inexperienced players by early rushes, and keep persistent patten attacks throughout the game.
LAN play (Score:2)
... I would rather play the old version is all its 640x480 glory then play a LAN game over the WAN. Sure I have the bandwidth, but it's the principle...
So your objection is that you would rather play over a network than over a network.
Is StarCraft the right game to use for this? (Score:4, Insightful)
Perhaps a game not so dominated by rushing tactics would be a better choice of base game? It definitely seems an interesting idea, but there must be games better suited to an AI contest like this...
Re:Is StarCraft the right game to use for this? (Score:4, Insightful)
How would you rather it be setup? I have not found a single RTS that isn't dominated by Rushing Tactics. I still play Age of Empires 2 for the whole walling off thing but it still doesn't beat a well developed rush.
Re:Is StarCraft the right game to use for this? (Score:5, Informative)
Re: (Score:2, Informative)
What about Total Annihilation? There is quite a bit that can be done to block a number of rush gambits. Of course, there is still always the LOL Gambit of building a swarm of transport aircraft to pick up the enemy commander (destroy their main unit and a large portion of their base since the base defenses are stupid enough to shoot it down.)
Re:Is StarCraft the right game to use for this? (Score:5, Interesting)
How would you rather it be setup? I have not found a single RTS that isn't dominated by Rushing Tactics.
Company of Heroes. http://www.companyofheroes.com/ [companyofheroes.com]
It's a modern RTS which utilizes things such as directional cover, suppression and per-squad reinforcements, as well as rewards proper flanking. Unless, of course, you try to prevent said flanking by placing some barbed wire and mines...
There is no such thing as rushing in CoH; the game doesn't reward rushing because it will end with a horribly tragic loss for the player who attempts it (!). You can't wall-off because you need some map control, resources need to be connected to your base in order to receive them, and your low popcap (based on the number of captured sectors) spells your ultimate doom. The nature of the game is that for the most part, each side has no more than ten units on the field. You can be a very good player even if you aren't a hyperactive teen capable of performing ten clicks per second.
Bottom line: if someone wants to rush you, you will win the game in five minutes. But if you want to wall-off, this game isn't for you, as it requires constant fighting on multiple parts of the map.
Re: (Score:3, Funny)
> There is no such thing as rushing in CoH
Yes, because you start with two machine gun nests in your base, making rushes impossible. CoH sucks.
Re:Is StarCraft the right game to use for this? (Score:4, Interesting)
Star Trek Armada and Armada II had a decent approach to preventing early-game rushes; your "town hall" equivalent building (starbase) is armed to the teeth. =)
Re: (Score:3, Insightful)
How would you rather it be setup? I have not found a single RTS that isn't dominated by Rushing Tactics. I still play Age of Empires 2 for the whole walling off thing but it still doesn't beat a well developed rush.
This is why I prefer Real Time Tactics Games to Real Time Strategy games.
You know... Like Total War series...
Re: (Score:2)
Say wha? AoE2 walls are fun, but you don't need them. Anti-rush tactics consist of ringing the frigging alarm bell and watching your holed up peasants shoot the hell out your moron opponent's attackers. You can *not* take a town center while in the Dark Age, and it's damn tough in the Feudal Age. You really can't develop a decent attack against an enemy base until the Castle Age, and you don't g
Re: (Score:2, Insightful)
In the time that you spent calling all your lumberjacks back to your town center, your opponents 10 militia Men will take down the Lumberyard, the Mill, any mining camps you have, and utterly destroy your economy so that when you finally manage to build a Barracks farther from your base and fend him off, he'll show up with some Horsemen to finish off the army you just built, and then when you take those out he'll be at your door with Battering Rams.
Either you aren't rushing properly, or your opponents aren'
Re: (Score:2)
Meanwhile, your opponent wisely pulls his militia off to harass your workers whenever you try unringing the bell. So while his workers are busy gathering resources to get him to Feudal etc, you are left doing nothing... just awaiting your unavoidable death.
And depending on what race he is playing, he could just finish you off in the Dark Age. I've had it done
Re: (Score:2)
Watch the pros. Rushing is considered to be a risky move. If you can catch your opponent doing some ridiculous early expansion, you'll win. But in normal circumstances, you'll set yourself back economically, unless you can manage to do a serious amount of damage with the rush.
Re: (Score:2)
I have watched some pros, and most of the times, their idea is that if you can get 2 zerglings (50 mineral) to take out a drone/probe/scv you've essentially made MORE on that trade off than anything else. Because those units cost as much, AND your denting your opponents economy. (So the idea is to send in 6 to get 3 probes, or 8 to get 4 drones, etc)
I've never seen early expansion work properly in any of the pro matches I've watched, which to be honest is just short of a dozen so I'm not the know it all.
It
Re: (Score:2)
I have watched some pros, and most of the times, their idea is that if you can get 2 zerglings (50 mineral) to take out a drone/probe/scv you've essentially made MORE on that trade off than anything else.
That's incorrect analysis, because you are hurting your own economy by building troops. Essentially, you are sacrificing a drone for 2 zerglings.
If you only manage to take out a single drone with those zerglings, you haven't even broken even with your opponent, because you both lost a drone, but you lost yours earlier in the game. Realistically, you probably need to kill two drones to make it worthwhile.
This is why zergling rushes are so risky. If you don't kill or cripple your opponent, you lose.
Re:Is StarCraft the right game to use for this? (Score:4, Informative)
Re:Is StarCraft the right game to use for this? (Score:4, Informative)
It's been a couple of years, but whenever I watched Boxer in the Korean SC tournaments a while back - the match is usually over within 15 or 20 minutes because they'd never need to progress past Dragoons, Hydra's, or Medics.
An expansive SC player would be destroyed by 8 zerglings before he could get that second Command center off.
Re:Is StarCraft the right game to use for this? (Score:4, Interesting)
Re:Is StarCraft the right game to use for this? (Score:4, Informative)
Expanding on the parent...
Every matchup except for Zerg vs. Zerg starts with EXTREMELY fast expanding these days. Usually before they even have a single non-peon unit out. Hell, zergs expand TWICE right off the bat against Protoss. The players have figured out how to stop these early rushes with building placement, micro and build orders.
If I were to guess, less than 2% of pro games in recent times are very early rushes aimed at killing a fast expanding players. Early rushes do happen more often than that but they are always with the intent of doing economic damage to get an advantage in the late game.
Re: (Score:3, Informative)
Every matchup except for Zerg vs. Zerg starts with EXTREMELY fast expanding these days. Usually before they even have a single non-peon unit out
That's not really true, at least the latter part. Protoss will often forge fast-expand, especially against Zerg, but other openings like one gate tech aren't uncommon. For terran, you almost never see expand-before-marine, and often there's no expand until the factory is building.
So you *see* FEs like that in each matchup, I wouldn't call them *the* standard build
Re: (Score:2, Interesting)
It's been a couple of years, but whenever I watched Boxer in the Korean SC tournaments a while back - the match is usually over within 15 or 20 minutes because they'd never need to progress past Dragoons, Hydra's, or Medics.
An expansive SC player would be destroyed by 8 zerglings before he could get that second Command center off.
Boxer's signature unit is is the dropship -- a mid-game unit that comes out only before the 3 science vessel units. 20 minutes by a pro's standard's is not a rush, its the beginning of endgame. They'll have 2-3 operational bases at this point.
Possibility for emergent gameplay? (Score:2)
What if, through the developers' virtual arms race, the AIs discover that rushing isn't actually the best way to win? Given enough room to experiment, could new, anti-rush gambits emerge that human players wouldn't have thought of?
Re:Is StarCraft the right game to use for this? (Score:4, Insightful)
I don't know how you define rush.... I've had people complain that an attack after 10 minutes was a rush. Even the 6-pool was easily defeated by the proper build order and positioning. As a matter of fact, I liked SC more than others because every strategy had a proper counter. The only thing that was required was scouting - otherwise the other person could come in with the counter to your troops.
While I don't think it is a great medium for a test, it's a pretty good one. Especially if the AI has to deal with fog of war.
Re:Is StarCraft the right game to use for this? (Score:4, Informative)
Re:Is StarCraft the right game to use for this? (Score:4, Informative)
StarCraft is only dominated by rush tactics when the players don't have the skills to defend against a rush. In StarCraft attempting a rush dooms you to failure if the rush doesn't fatally wound your opponent ('cause you stunted your economy to build your rushers). Correctly defending against a rush is mostly micromanagement (using your workers correctly to defend, which means constantly issuing them the attack orders they need since they won't attack on their own, while keeping some working on your economy). AIs should excel at micromanagement. I don't think rushing would be a problem in a StarCraft AI match.
Re: (Score:2)
Re:Is StarCraft the right game to use for this? (Score:4, Insightful)
I'd like to see how this statement changes after the winner of the competition is unveiled.
Re: (Score:3, Insightful)
I don't see any inherent reason why AI should be bad at dancing units. If anything it should be better at it, because the essentially infinite click speed and ability to attend to multiple places at once means that an AI could dance multiple groups of units at different parts of the map at all once, which for humans is something only really skilled players can do.
Re: (Score:2)
Re: (Score:3, Interesting)
For logistical reasons there aren't a lot of options. Past competitions have used Wargus [sourceforge.net], since it's open source. Game-industry people tend to roll their eyes at it though, and would prefer a competition using a "real" RTS, i.e. a popular mainstream one. Starcraft is one of the only choices for that, because someone's made an API for it [ucsc.edu] that allows you to write external AI to play the game. Most commercial RTSs don't have any way of doing that, unless you were to screen-scrape the display and then have to i
Re: (Score:2, Insightful)
Parent should be modded insightful.
Although I'm not an experienced RTS player, I am an experienced FPS player. People who claim that SC is dominated by rushing tactics are just as ignorant as the people who claim that dueling FPS games are dominated by item control or map knowledge. The answer to that is - Well duh. May as well claim it's about how well you use the keyboard and mouse.
There's way more that goes into it when you break it down to the specifics. For SC - are you effectively scouting your op
Re: (Score:3, Interesting)
> Rush tactics are barely used in pro-matches
Yes and no. Players often apply the *threat* of a rush, forcing the opponent to build more defences.
Breakdown (Score:2)
Human Advantages:
Advanced Prediction
Flexible Stategies
Arguably Faster Learning
AI Advantages:
Able to command all units at once
Usually More efficient w/ resources
Instant Macro management
Another advantage to the AI could include knowing the map layout and what the player has at all times, which is something the original starcraft had so the AI would know whether to rush you or not.
Re: (Score:2)
It may have just been my limited experience with the AIs but most of the time it seemed that the AI was on a pretty fixed schedule in terms of attacks. If the AI had any comprehension of what the human player had built then I'd say that the AIs were very very poorly designed. They'd attack massive defense with a dozen zealots when a human player that knew the defense was there would have known it was futile to h
Re:Breakdown (Score:4, Insightful)
Most game AI's are not well designed, but not because they can't be. Most game AI's are built from the prespective that the player should be able to win, therefore Grandmaster level thinking is less desirable than preditable patterns that seem impossible to be till the player realizes they can be exploited.
Re: (Score:2)
Exactly. The levels can be treated more as puzzles that need to be solved. After a couple of times trying to beat a level, you get a feel for the types of attacks to expect, and figure out ways to counter them. This could lead you to doing things you could never get away with in a Battle.net game. For instance, I was never a big fan of Vultures, but there were a few Terran missions where the spider mines were very useful.
Re: (Score:2)
Vultures are very heavily used in pro matches, both for the mines and to sneak into enemy bases and kill workers.
Re: (Score:2)
In adventure games, or in campaign mode, I would agree with you. But in skirmish mode, the AI should be as smart as it can be (when set to hard).
Re: (Score:2)
In a perfect world yes. In a realistic world, the game developer realizes that skirmish mode is simply what people play when they don't want to bother with going online and finding a human to play against, and the effort spent creating and polishing a seperate AI just for that mode would garner more points with the player if it was spent elsewhere (such as fine tuning the balance of the game, polishing the maps, and etc.)
Re:Breakdown (Score:4, Informative)
Re:Breakdown (Score:4, Informative)
Allowed html is displayed below the comment form when you write a comment. Only the the following are allowed on slashdot:
Re: (Score:2)
Re: (Score:2)
If it's an AI competition, I doubt the AI teams would be given any more information than the human teams had. Computers could be better at micro management, but probably not by enough to make up for humans' ability to adapt to changing circumstances and come up with new tactics on the fly.
Re: (Score:3, Informative)
Re:Breakdown (Score:5, Funny)
Human Advantages:
Imagined Prediction Advantage
Flexible Stategies
Arguably Faster Learning
AI Advantages:
Able to command all units at once
Usually More efficient w/ resources
Instant Macro management
Korean Advantages:
Superior Strategies
Advanced Prediction
Flexible Tactics
Arguably Faster Learning
Able to command all units at once
Usually More efficient w/ resources
Instant Macro management
Fixed that for you :D
Re: (Score:2)
Re:Breakdown (Score:5, Funny)
Worst. Poem. Ever.
Re: (Score:2)
Another advantage to the AI could include knowing the map layout and what the player has at all times, which is something the original starcraft had so the AI would know whether to rush you or not.
That's not an advantage AI has, that's just cheating. Circumventing game rules to see full map is just plain cheating, if a player did it you wouldn't say he's any better than the guy who doesn't see the map, same rules apply to AI. It's equivalent to making things 50% cheaper for the AI or giving it free units... it's just plain cheating.
Re: (Score:2)
Potato Potahto, I would agree that it isn't fair, but neither is being able to command over 100 units individually in a miniscule fraction of a second. When this kind of information isn't readily available to AI, they tend not to be difficult. I've never seen a human opponent lose against an AI when the Human goes on the offensive first.
AIIDE web site (Score:3, Informative)
The aiide conference web site [aiide.org] has been Slashdotted... even though Slashdot didn't link to it. :-)
competition announced? (Score:2)
Re:competition announced? (Score:4, Informative)
bah (Score:2)
The cheapest way for the computer to beat the human is to simply open up multiple fronts that cannot be simultaneously micro-managed. Even maintaining one significant skirmish at the front will prevent the human from micromanaging the base back home.
I've been a fan of RTS from the beginning but haven't seen anything really exciting since Total Anihillation. Maybe I've missed some great ideas but pretty much every one I've tried since has been met with the initial glee of pretty graphics and then the crushin
Re: (Score:3, Informative)
If you like TA, check out Spring at http://springrts.com/ [springrts.com]
This is an open source fully 3d replica of TA. They've now built it to the point where it is a base engine that can host one of several mods - mostly based on TA style models and concepts, although a few are completely unique. AI's are plugins that can work over several mods if the author chooses so.
My favourite is the Complete Annihillation Mod - http://springrts.com/wiki/Complete_Annihilation [springrts.com]
The "chicken" mode has a weak AI, but enough brute force
naughty ai would win (Score:5, Funny)
My AI would design its base to be a rough representation of a naughty picture on the minimap. Human players would always lose as they just let the AI build away to see the picture get a higher resolution.
Re: (Score:3, Funny)
I hope they do it for SC2 (Score:2)
I'll buy StarCraft 2 if they have frequent (1 per year or more) competitions like this. Hopefully Blizzard will release its own easy to use API for AI competitions with the release of the game.
Ah I get it (Score:2, Insightful)
Perfect micro will warp gameplay (Score:2, Interesting)
I imagine that a computer's ability to control units with instant reflexes and frame precision will make AI Starcraft a completely different game from anything we've seen. Watch some Tool Assisted Speedruns and see how the gameplay of a person playing frame by frame transcends that a skilled human playing normally. Games are designed, tested, or balanced with the expectation that a player cannot press a button thirty times a second, anticipate the frame in which a projectile which hit, or issue commands at
Re: (Score:2)
Re: (Score:2)
no, but Blizzard does want to be lazy and outsource AI development.
Re: (Score:3, Informative)
RTFA.
The competition is being held by Expressive Intelligence Studio at an AI conference. Blizzard has nothing to do with this, AFAIK.
Re: (Score:2)
Starcraft AI? Well what do you suppose that is for? Notably you think an open competition won't just be sucked up by Blizzard as a result? duh.
Maybe you should again RTFA or learn what Starcraft is.
Re: (Score:2)
Re: (Score:2)
It would actually be very difficult to make it better than the random BNET player. Starcraft players are getting very good. And it is pretty hard for a computer to beat a human for an RTS in general (without cheating of course).
Re: (Score:2)
Well, computers have infinite APM, and AI bots will have to (ab)use this to win against other bots. It should also help them against humans. Do you consider this cheating?
Re: (Score:2)
Re: (Score:2)
Isn't Jaedong the currently top-ranked player?
Re: (Score:2)
Re:Does AI have to be good? (Score:4, Informative)
Depends where you look. Last month's KESPA ratings [teamliquid.net] (the latest, at least on TLPD) put Jaedong at #1 and flash down at #6. In fact, the last time he wasn't #1 in that ranking was March.
Re: (Score:2)
Step 1: Playoff/deathmatch style showdown to find the the best of the AIs
Fallacious
Re: (Score:2)
Re: (Score:2)
Take a look at Sins of a Solar Empire, although it's in space...
Re: (Score:2)
Re: (Score:2)
Sins of a Solar Empire is a very good game. Some people complain that it doesn't have a campaign to play through; it's strictly custom games, but that didn't bother me. I sunk days/weeks into it back when I got it and I still go back to play periodically. It doesn't have age advancement exactly, but research is a very important part of the game (to allow colonization of certain types of planets, give new abilities, improve your weapons, etc).
If/when you get sick of the regular game, there are also lots o
Re:blacksheepwall (Score:5, Informative)
It must be very difficult if you cannot click the link under "rules"
#
Programs that attempt to cheat will be disqualified
1.
Bots must disable the perfect information flag in tournaments 1,2 and 4
Re: (Score:3, Insightful)
Re: (Score:2)
A player could feel more satisfied if it plays against a computer with the same knowledge and resources as a human player, because then it would have to play more like a human. With such cheats, the player will feel annoyed that the computer always attack when he is the weakest, without real knowledge, or can attack with twise the units he know is the maximum at a given time.
What always bothered me in 4x games is that I'd do my initial planet/city grab, then sit tight building defenses while shuddering at the big enemy force groups moving by. (Master of Orion, 32k ship stacks anyone?) Then I'd finally be able to put a death fleet together and relish some big clashes. Guess what? The enemy megafleets go away. Argh! Sometimes I just want to smash metal on metal and watch the sparks fly. It's no fun when the enemy doesn't come out to play. Ultimately the human wants to win but it'
Re: (Score:2)
Well, I wrote some Game AI that you cannot overcome. It didn't cheat at all. It simply used my strategy to win. Unbeatable!.