On January 25 2019 06:37 brickrd wrote:
actually, SC2 fans should understand a thing: the purpose of deepmind entering starcraft is not to provide us entertaining bots to satisfy our intellectual curiosities about the nature of competition in RTS. it's to promote and showcase machine learning. they're doing something historically and technologically significant whereas we are playing games. we should be thanking them for taking any interest at all and for giving SC2 visibility rather than criticizing people who are professional researchers working in their own damn field.

I dont agree with gratfulness stuff, but I agree that try to improve what they have to a general AI. and having a player level micro will accually be better for it. Their goal was not to win, but beat a human in decision making in a complex incomplete information situation.

No_Roo

United States905 Posts

January 24 2019 22:43 GMT

#22

On January 25 2019 07:32 StormsInJuly wrote:
Asking for fairness, or trying to get an AI to play as human-like as possible by imposing restrictions on it, is sort of missing the point.
But I agree it's interesting to place restrictions on it and see if it can still beat pros with a hard cap on 200apm, then 100apm, etc.
Seeing it get juked hard by warp prism play makes me think it's a bit too early to start handicapping it like that though.

The reason why this is important because the game is balanced around what's possible with the given input devices. Remove restrictions and any AI will beat a human: + Show Spoiler +

That is to say, it's already been done, almost a decade ago, hence, not terribly interesting.

StormsInJuly

Sweden165 Posts

January 24 2019 22:50 GMT

#23

On January 25 2019 07:34 Charoisaur wrote:
The point of this deepmind bot is to prove machines are smarter than humans

To steal a quote, asking if machines are smart is about as relevant as asking if submarines can swim. Computers are faster at a lot of things but until now, we haven't been smart enough to translate that speed into winning a game of Starcraft, APM limit or no. If it's true that deepmind got 200 years of effective practice in during a few years, well that's an unfair advantage and you'd certainly expect a human who could grind SC2 for 200 years to be the best pro player by far.
But I don't really disagree with you, forcing the AI to economize with its actions will yield more interesting results.

On January 25 2019 07:43 No_Roo wrote:
That is to say, it's already been done, almost a decade ago, hence, not terribly interesting.

I'm not aware of any AI consistently beating a pro-level SC/2 player before today but please correct me if I'm wrong.

No_Roo

United States905 Posts

January 24 2019 23:08 GMT

#24

I'm not aware of any AI consistently beating a pro-level SC/2 player before today but please correct me if I'm wrong.

What AI is consistently beating pro level players after today? Alphastar isn't, because it isn't playing starcraft2 yet. It's currently relying on having access to additional in game information that would get a player banned on battlenet if they got caught having access to it.

jy_9876543210

265 Posts

January 24 2019 23:09 GMT

#25

On January 25 2019 06:39 Waxangel wrote:

Show nested quote +

that's actually pretty interesting to hear

so you'd say MaNa's max of 700APM is more realistic?

[image loading]

If they provide such plots for all progamers, I wouldn't be surprised if they find that many zerg players has 2000 or 3000 peak apm while the other zergs or the other races has 600.
I believe it depends on something called "keyboard scan rate" or "keyboard polling rate", some keyboards specially designed for that purpose has extremely high rate, so you just press "sz" once and 100 zerglings are made, while it seems that T and P don't need it much.
But that's not the micro ability of human players, and I think DeepMind team didn't notice that and assumed that that's how progamers micro.