EPTIt seems to me as if the "winning" program was simply examing more future moves than the "greedy" program. It looked at the possible enemy responses after all. In chess programs the program gets stronger the further and more completely it looks into the "future" of the game. The same should be true for Mancala I guess.
Competitive Gaming Article by Day[9] - Page 4
| Forum Index > BW General |
|
Simplistik
2005 Posts
It seems to me as if the "winning" program was simply examing more future moves than the "greedy" program. It looked at the possible enemy responses after all. In chess programs the program gets stronger the further and more completely it looks into the "future" of the game. The same should be true for Mancala I guess. | ||
|
aRod
United States758 Posts
From the minimax theorem, a players best strategy is the strategy that minimizes potential loses against all opposing strategies. Mencala is a game of perfect knowledge. Here, every aspect of the game is visible to each player. This makes the best strategy easier to choose. However, unlike Mencala, Starcraft is a game of imperfect knowledge. Certain aspects of your opponents position and strategy are hidden from you for at least part of the game. Thus, you cannot always select the best counter to what your opponent is doing. You cannot be in your opponent's head visualizing the strategy he will choose. Thus, your strategy will not always be the best because it will not always minimize potential losses against your opponents strategy. The players that do the best in Starcraft are the ones who scout incredibly well. Good scouting increases knowledge of your opponents strategy/position allowing you to select a suitable counter. However, there are still times you do not know what your opponent is doing especially in the early game. This leads me to ask what characteristics should we look for in early game strategies especially when scouting is limited due to blocks/map distance etc... Initially I believed conservative strategies such as 2 gate >> obs are great starting builds simply because these strategies minimize initial potential losses. However, the eventual resulting potential losses from conservative strategies can be tremendous. Considering 2gate >> obs when facing a terran who quick expos. The p player's strategy puts him significantly behind in economy which often leads to a stomping generated by the terran player's superior economy. Accepting the premise that obtaining and exploiting a marginal advantage should lead to victory, the best strategies in games of imperfect knowledge are the strategies that are most likely to give you a marginal advantage. This is why FE builds often lead to victory especially on larger maps. Thus when choosing strategies, we should look for strategies that are likely create marginal advantages. The best games, in my opinion, are the ones with very subtle marginal advantages. Boxer's recent games on Blue Storm are great examples. Boxer took superior position (in one game with marines/medics, in other with vults/mines), and his following game play allowed him to exploit that position. He may have been slightly behind on economy initially, but his position was his marginal advantage which he exploited fully. Keep in mind that randomness is part of Starcraft. You'll never know exactly what you should have done until you see the replay, but I hope this little bit helps when it comes to choosing strategies. | ||
|
noname_
456 Posts
 very good article, good job | ||
Xeofreestyler
Belgium6771 Posts
Pretty much sums up how I feel about StarCraft. | ||
|
Kage
India788 Posts
If you have more insights into what makes gaming greats please do go ahead and share them here! Excellent read, again. | ||
|
Gaetele
Esper760 Posts
For example, players like July can turn what would be a marginal disadvantage for most players to an even ground or a marginal advantage. While Starcraft can be split up into essentially three areas where you can have advantage - economy, tactics and army size, players like July can sacrifice one of the three more readily. 8 Zerglings vs. 6 Marine/2 Medic could be Zergling blood for many players, but it could be easy pickings for someone like July. Marginal advantage in one area can often be "traded off" or led to into an advantage for some other area. Like BoxeR's TvTs on Blue Storm - while he may not expand and could be a base behind the game, his Vults/Mines or his M&M hold the center so that he can expand if he needs to. In this case, his tactical advantage leads directly into an economy advantage. This economy advantage could then supplement his tactical advantage or both tactics and economy could supplement an army advantage, which is common in his play now. A true good player knows how advantageous of a position he is over his player, and by allocating that advantage between the three different types, he can push the lead. While the defender may have an advantage in Starcraft by virtue of travelling distance and "home field advantage" and such, the player is not to be discounted in the equation. By putting pressure on a player early game, the attacker gets the marginal advantage in that area - he is far more composed and can do more with his other advantages than the opponent. While the defender has to concentrate on fending off the attack and rebuilding any losses, the attacker (or pressurer, to be more precise) can continue to do what he would do in his base. This leads to the difference between pressure and attack. Pressure is meant to trade advantage. It is, in example, the way to trade off an army size advantage for an economical or a tactical advantage. If you sneak 12 Zerglings into a base and kill off 6 SCVs, you've swapped your army for their econ. Perhaps, also, you are forcing their otherwise-attacking force to return to defend, also gaining you a tactical advantage. Attacking is the end of trades. You trade all three of your advantages to secure a victory. For example, take the 4 Pool Zergling rush. Your economy is automatically at a disadvantage as you are not building workers. Your army size is at a huge advantage by virtue of the 4-Drone Spawning Pool. If you fail to use this army advantage to destroy their economic advantage, then the player will have a very hard time to respond. | ||
|
7c.nEptuNe
United States153 Posts
If you're interested.... here's a link to the site. | ||
|
roc)Candy
United States17 Posts
| ||
|
Moletrap
United States1297 Posts
| ||
|
{ToT}Strafe
Thailand7026 Posts
Nice article | ||
Bockit
Sydney2287 Posts
EDIT: I wish I could say more but it's pretty much all been covered  | ||
|
One Page Memory
Bulgaria2145 Posts
Very good article Day! You might don't know but it implies also to chess very much. | ||
|
EvilTeletubby
Baltimore, USA22253 Posts
| ||
|
evanthebouncy!
United States12796 Posts
| ||
|
doc.x
94 Posts
the idea is that at low levels, two players are both playing poorly so to do well, each player needs to think about making the best move. but at high levels, both players should assume that the other player is playing perfectly. in this case, when either player tries to force things to happen, they are actually overextending themselves and allowing weaknesses to flow into their player. at advanced levels where play is very close to perfect, a player can only win by taking advantage of another player's mistakes. in a chess game of 30 moves, even best players make very small mistakes maybe a few times a game. if a player takes advantage of his opponents mistakes those few times, he builds a marginal advantage which can later accelerate and accumulate into a large enough advantage to win in the end game. otherwise, if a player tries to force a win when his opponent hasn't made a mistake, then the player only overexerts himself and gives his opponent the opportunity to take advantage of his mistake. ya, marginal advantage is where it's at though. | ||
|
doc.x
94 Posts
On January 06 2008 14:17 Simplistik wrote: Hehe, and to talk about this a little further... It seems to me as if the "winning" program was simply examing more future moves than the "greedy" program. It looked at the possible enemy responses after all. In chess programs the program gets stronger the further and more completely it looks into the "future" of the game. The same should be true for Mancala I guess. that's not completely true. the program that examines more moves isn't the program that will win. in chess programs for example, for the most part until a few years ago, the program that can analyze the most positions, and analyze them most correctly, will have the highest winning percentage. this changed in the past few years when a chess AI programmer paired up with a chess international grandmaster and instead of brute forcing future variations, they applied chess principles into the chess program. so the new chess program thinks like a person would think. with this shortcut, the program would analyze less variations than other brute force programs, but the variations it did analyze were the ones that were the best ones. if you talk to chess players too, they will say that grandmasters for the most part, unless things get really complicated, tend not to analyze that many moves forward. rather, they have a very good knowledge of how the game will flow, and instead of analyzing move, they analyze the concept of the current game. i love chess. it is such a beautiful study of life. | ||
|
Randomaccount#77123
United States5003 Posts
| ||
|
AsianEcksDragon
United States1036 Posts
| ||
|
mierin
United States4943 Posts
On May 12 2011 03:10 AsianEcksDragon wrote: What if I play Grubby in War3 a million times? Wouldn't I eventually take a game when I get really lucky with RNG and loots? Odds are, he'd probably lag out of a game after 1 million games and you'd receive a victory. EDIT: On topic, I think this is related to the BW vs. SC2 debate a little bit...it just seems to me that too often any advantage you take in SC2 instantly becomes way more than marginal. | ||
|
MattyClutch
United States711 Posts
| ||
| ||
