Being good at "strategy"

The point I was making is that at the very highest level, the game goes to the person who is most able to see the best move in the given position.

To say that an AI doesn't strategise is invalid. The best Chess AI's simply brute force the best move for the given situation, because that's all the strategy you need. And so there isn't strategy, because two players with equal ability to see what the best move is will tend to force the game into a draw.

So those aspects of strategy are pretty much invalidated by players who are able to very quickly see the best move they can make.

In regards to GM's not having perfect calculation ability, what is that glorious and utterly relevant quote about the current world champion?

Chess AI's don't just brute force. There's a reason that if you take Houdini and have it play Deep Blue on equal hardware that Houdini is going to be vastly better. It's because chess engines have become better at strategy. Also there are lots of stylistic difference between chess engines.

And yeah, sure, chess has a puzzle aspect to it where you can objectively evaluate positions and even I can beat Magnus Carlsen if I'm a queen up, but there are still stylistic differences between top players and top players still play primarily on intuition, calculating mostly to double-check.

"In regards to GM's not having perfect calculation ability, what is that glorious and utterly relevant quote about the current world champion?"

You're being annoying, what are you talking about?

Isn't that because newer chess AI's use libraries of previous matches to be able to faster judge situations?

Haha touché.

Oh it's a honest question. I thought I read about it and it makes it a lot of sense in my mind, but I am not really into chess so I don't know for sure.

Haha nah I thought you were stating this to back up your OP's point. Me liked.

That was the entire concept of Deep Blue. It wasn't really a demonstration of "AI" or "computing power". It was a demonstration of database lookup capabilities. There are something like 140 million feasible chess board combinations and Deep Blue was supposed to have a catalog of most of them and understood which ones could come from other ones and which ones were in games with the most wins. I don't think it really even knew the rules of chess so much as it understood that 82% of the time when the board looked like THIS, you needed to make the board look like THAT to head towards a win.

Which speaks, I think, to the OP's point, that it's possible for strategy in a game to just be memorizing a lot of possibilities.

But that's chess, which is a flat board of entirely predictable (if numerous) circumstances.

That might be technically true of any situation. We could look back on any real-life war and say that this was really 1 of a finite number of possible setups and results and a real general just needs to memorize all 380 billion possibilities to be a master strategist.

But I think the art of strategy is to examine a situation on the fly and understand what needs to be done next without having had time to examine or memorize all possible setups, results and eventual outcomes.


I do think that a lot of RTS games can become "played out". You can only play a given map so many times with so many people before all strategies are well understood and quite possibly only a single-digit number of strategies are even viable. (Battle for Middle Earth 1 was especially bad about this. e.g., Gondor vs Gondor, you got Gandalf, you leveled him up, you blew the doors off the other person's base and you charged in with cavalry. There was no other strategy in this matchup. Starcraft strategy gets mundane quickly too. If your opponent is doing X, you know you must do Y. There's not a big list of possibilities.)

Where I think PA is trying to excel is by presenting a much larger list of possible strategies both by having a complex game and by having randomly generated maps. The art of strategy will be in "winging it" because there is no exact precedent for the particular situation you are in. Never before has anyone been rushed from a teleporter on a lava planet while being bombed on the starting planet while a 3rd player is almost done with a Halley on a moon and you have 2 nukes ready to fire and you know a 4th player has 3 nukes he's almost done with and, and, and...

As complexity deepens and it becomes harder to find specific examples of your exact situation, strategy becomes more about skill and less about memorization.

Well I do. The fact that I think I know about this way of chess AI is part of why I think of strategy like I do.

Lol where did you get that number about chess? Please read a bit from here http://answers.yahoo.com/question/index?qid=20100306195331AAq9Zji

Strategy is the long term plan you want to use to win. It can include a bunch of conditions about how you will respond to reconnaissance as well. For example, you might decide your strategy is to quickly acquire a certain threshold of economy, set up defenses and go orbital early to get a macro advantage on a relatively uncontested planet. Of course there is some debate as to whether it is a good strategy, but it is a strategy nonetheless. You might further decide that if he goes for an early mass of bots you will respond with some bombers.

Similarly in chess. You might decide as white that if he does a Sicilian defense you will castle queenside and then try to push forward on the kingside to break through his defense.

Tactics are the details that may disrupt that plan. If your opponent builds artillery near a portion of the base economy you wanted to use to go orbital, that is a tactical consideration. It may also be part of his strategy to use artillery creep to disrupt you in general.

In the chess variation, if he tries to bring pressure to your A-file pawn and you have to spend moves to deal with that, it's a tactical disruption to your plan to assault him kingside. It may also be part of his strategy to win on the queenside.

Every RTS game will have both. I think StarCraft has more tactics both because the game is more known but also because they use fixed maps. If you watch somebody cast a game, the first thing they will talk about is the strategy and tactics of the map in question. For example, they might say that on Metalopolis with close positions, the zerg will enjoy some advantage because they can use mutalisks to quickly cross the gap and harass and still easily retreat.

In each of these cases, you can see how you can plan your strategy around the tactical advantages the situation presents. On a planetary annihilation map with water, your can build a strategic plan around the tactical opportunity to do a ship to shore bombardment. On a map with two planets and an asteroid, you can build a plan around fleeing the main planet to quickly take the other two and annihilate the original.

http://www.chess.com/chessopedia/view/mathematics-and-chess

There's a difference between "possible moves" and "possible board configurations." Chess doesn't have 300 billion board configurations. I'm also not sure that those really high value numbers are taking the actual rules of chess into account. e.g., your first move has only 20 possibilities.

I love that the discussion in this forum has moved towards higher level things like strategy vs tactics, ui improvements that would help focus on macro, and just general strategy ideas. To me this means that the balance is getting better and people are really starting to see this as a real game with a lot of depth. It has been fun watching how much the conversations here change after each new game update. It's like this community is growing up and having more and more abstract conversations as the game matures.

I wish people would finally refrain from using the terms strategy and tactics altogether. They are clearly defined only as the translations of the ancient Greek terms they are derived from. Outside of that narrow context the terms are redefined in any way seen fit by those who wish to use them.
Why not talk about plan and execution instead? It is sufficiently close what most people think of when hearing strategy/tactics and is far easier talked about on an equal footing.
It is clear that one is useless without the other. So if we equate strategy with planning the OP's question should and could be asked only after another has been answered, that one being: Is PA a game that profits from or even requires planning? Or for that matter, are there any RTS games that truly require planning instead of memorization and speed of task execution?

Not necessarily words have multiple meanings to them and they change over time. We have to figure it out through context.

Bearing in mind that RTS technically includes any strategy game that runs in real-time (as opposed to turn based) then certainly the answer is yes. I don't think anyone would debate that Hearts of Iron III requires strategy, as in long term plans that aren't clearly laid out for you, and it is a real time strategy game, albeit not remotely like Starcraft.

Where RTS games like Starcraft go wrong is in the strong paper-rock-scissors construction of the game which means that for given strategy X there may only be a single counter-strategy Y that works. I think that's something pretty easily avoided in a game like TA/PA where long term planning and troop movement count for more than simply deciding which unit-build strategy to go with.


Maybe that's it in a nutshell. Games like Starcraft are mainly about the strategy of unit building: what to build in what ratio at what time. TA/PA places far less importance on this but far more importance on things like map control, which requires a more fluid strategy.

Anybody here played SOSE?
And by 'play' i mean online vs human beings, particularly team matches.
One of the best strategy games ive ever played

True but this is far from strategy.

I did indeed and I do concede that my comment is really too narrow-minded now that you pointed that out. I'm just allergic to twitch gameplay and improvement by memorization I guess. No hard feelings =)

Yes I have played it but there is a problem you need to keep the scale small or it lags.

Depends on your PC.

Im really bad at splitting my forces in SOSE, and only ever do it (ANd in PA) when the travel time and defences mean that you either have 2 army's, or you die where you can't defend.

And so usually I death ball in SOSE, but can't always afford to do that in PA, as a proper assault can happen in minutes, meaning tanks struggle to death ball on a multi-front war.