Any chance we'll have the patch by tonight for this weekend?

>> Lastly, mtw is close to chess, not to go. You know his units in advance(before even starting the battle), you watch what he does with them(there are no new units showing up out of the blue in any square of the grid) and respond to his moves. There's no "teleporting"(which ok, is a huge issue when designing an ai).

MTW has almost nothing in common with chess from a programmatical standpoint. Two humans playing against each other in MTW might feel like it's similar to Chess (one of the reasons why chess has always been popular with strategy nerds -- it employs the basic tools needed to out-think and out-maneuver someone). But in terms of the rules of the game, MTW and Chess are nothign alike.

Yes, the AI can give orders simultaneously, but we can *figure out* orders more efficiently than it can. Just because it can order everyone to attack in less time than we can doesn't mean the decision is a good one -- it just means the implementation of that decision is slightly more efficient because it doesn't have to deal with mouse clicks.

I guess my point can be reduced to: if AI weren't an order of magnitude more difficult than any other aspect of game design, then we'd see a lot better AI in most games. The folks behind GalCiv 2 are ahead of the curve in this respect -- most game AI is a huge pile of 'if this, then this, weight that and then do this' while they're actually trying to make it 'learn' -- a real artificial intelliegnce, as opposed to an artificial batch of responses to stimulii. I don't know where on this specturm MTW2 falls since the GalCiv guys are a little more open about the guts of their game than the MTW2 guys are.

They basic formula for game developement: Good, Fast, Cheap... pick two.

Unfortunately for us, and at no fault of the developers, is that publishers usually pick cheap and fast. Patches are required to make the games good. Pre-98, post-98, PC, Mac, Nintendo, or PS3, it all boils down to that basic axiom of all business. Good, Fast, Cheap, pick 2.

the only problem i have with general is that he doesnt appear to have moderator under his name.:no:

Quote:

---

Originally Posted by adembroski

They basic formula for game developement: Good, Fast, Cheap... pick two.

Unfortunately for us, and at no fault of the developers, is that publishers usually pick cheap and fast. Patches are required to make the games good. Pre-98, post-98, PC, Mac, Nintendo, or PS3, it all boils down to that basic axiom of all business. Good, Fast, Cheap, pick 2.

---

I agree with you here, however as a developer you put out a list of fixes a week or two before the release of a patch you have to expect a percentage of the gaming population to gripe and groan.

Granted the majority are happy to see the list, but there is a small vocal minority that want it released yesterday. Been that way on many a game forums....

Quote:

---

Originally Posted by SnowlyWhite

if my militia goes against his militia (same numbers, same xp) I don't know who's unit gonna win and consequently if I have to reinforce my unit or not. I have to always start from the assumption that I might be unlucky. He knows, because, well... he's the random number generator. In chess, you don't have this huge advantage because there's nothing random in chess.

---

The Total War AI never used to predetermine its combat results. The AI made it's melee matchups based on a comparision of the unit's combat stats as modified by the situational combat bonuses. The combat calculations would then be made in real time as the individual men engaged and went through their 1 second combat cycle. There was a chance it could loose an advantageous matchup due to randomness in the combat calculation, but that chance was small because the minimum combat advantage was 20% and usually 40% or more. A 20% advantage would win 6 out or 10 times for 60 man size units, and a 40% advantage would win 9 out of 10 times. If the AI's unit was stronger, it would make a direct frontal attack. If it was equal or weaker, it would make an indirect flanking attack. If it was weaker in a flanking attack, it wouldn't attack unless forced to, and would withdraw its army if its army was weaker overall than the enemy's army.

Quote:

---

Originally Posted by SnowlyWhite

Randomness actually helps a ton; ok, if the computer "cheats" and knows in advance what will be the result of that. Obviously, you could argue that, in the current implementation, he doesn't know; the idea would be to change the implementation (like - determining the result from the start and then just show the animations) not to complain about how you implemented it.

---

Increased randomness helps the weaker player to win which could be the AI or a weaker human in multiplayer. The weakness could be in army strength or in how the army is used. For instance, I've seen the figure of 30% chance for a beginner to beat an expert in Backgammon which has a high random element to the gameplay. In chess, a beginner won't win a single game against an expert. Increasing the randomness is a way to give the presumably weaker AI more winning chance, but I think it's counterproductive in multiplayer if it can't be overcome within a single battle by skillfull play. For instance, in Backgammon you have to play many games before the skill of the expert is apparent. You can do that in Backgammon because the games go very fast.

I think the combat model will suffer as a simulation if you predetermine the combat result because you won't be able to take enough factors into consideration. Apparently, something like this has been done in the archers model where a hit is determined at the moment the arrow is fired. A consequence of this is that it's just as easy to hit a moving target as a stationary target because target motion is not being taken into consideration in the model. In the old physics based model, target motion is automatically accounted for with no additional calculation because a hit isn't determined until the arrow arrives. Likewise, volley effectiveness vs target distance and orientation are also accounted for in this model.

I loved RTW (my first TW game) so I bought M2TW. I love it too. Sure it has some issues. But so far that didn't stop me at all from playing and enjoying the game! I have been playing for many, many hours and still haven't encountered many of the issues on the bug list. So how much time do you think developers should spend playing hoping to accidentally bump into a situation?

A lot has been said about the increase of game size. Maybe it's also been said, but I haven't seen it: complexity increases exponentially with size. Balancing 1 ball is easy, 2 is getting harder, 3 requires quite some exercise, and 4 is already getting impossible. :juggle2:

Note also that patching is a VERY costly business: you have to keep paying those high-earning developers while you will never receive a penny for the effort. The time they spend on patches cannot be spent on R2TW, or Napoleon: TW, or whatever comes next. So I think we should appreciate CA's efforts in this a little bit more.

I really won't wait for any patch, because I'm getting rid of those pesky Timurids today. :smash:

Quote:

---

( unless Skyline Pete will go and deliver an esky full of Fosters)

---

No one in Australia actually drinks that stuff =P


Personally I'm not too fussed when the patch comes out as I'll just have to re-mod all my changes again before I start playing.

Let's move discussion of patch release to this thread:

https://forums.totalwar.org/vb/showthread.php?t=80775