Heroes 4. Player versus Comp difficulty

[pepak]

Another example is the AI being unable to leave its castle area - every day a new batch of creatures is available for buying, so it just returns to the castle to recruit them. It wasn't a problem in H3 thanks to its weekly growth of creatures, but the flaw is rather apparent in H4...

[Corribus]

Wasn't the H4 adventure AI an exact copy of the H3 adventure AI?I'm pretty sure I read that somewhere.

If that's true, that makes it even worse! The games are completely different - it would be no wonder that the H4 adventure AI is terrible. That'd be like taking the AI from Starcraft and having it try to play a game of Civilization.

Corribus

[Corribus]

Another example is the AI being unable to leave its castle area - every day a new batch of creatures is available for buying, so it just returns to the castle to recruit them. It wasn't a problem in H3 thanks to its weekly growth of creatures, but the flaw is rather apparent in H4...

I see the point you are trying to make, but you're failing to see the overall picture. What you're trying to basically say is that the AI did some of the same things in H3 as it does in H4. By using this argument, you are trying to disprove my claim that the H4 AI was much worse than the H3 AI. (I.e., if they did the same things, how can one of them be worse than the other, right?).

But whether the AI did similar things in H3 and H4 is really immaterial, though. The way the AI code is adapted to the rest of the game is just as important as the integrity of actual AI code itself. Was the same AI code used in both games? Maybe. But that does not preclude the possibility that that AI code might have been perfectly suitable for H3 but not suited at all for H4. Given the dissimilarity of the two games, it seems unreasonable to expect that the same AI code would operate with equal efficiency in both cases.

Thus you basically prove my point when you point out how a similar tendency of the AI in both games leads to big problems in one of the games but doesn't in the other. The fact that creatures regenerate weekly in H3 is compatible (as you say) with the way the H3 hangs around a castle, but the new creature regeneration method in H4 is NOT compatible with this tendency. Just because the code is the same does not mean that the AI is "the same in both games". In H3, the AI is competent because the code was built around the system. In H4 it is not (for whatever reason). I don't think you can refute that.

[pepak]

In H3, the AI is competent because the code was built around the system. In H4 it is not (for whatever reason). I don't think you can refute that.

Indeed. But I can use this to make a claim that it doesn't have to be a game programmer who makes the AI overcome its inherent deficiencies. In case of H4, it is the mapmaker who can make the difference. Before you damn the game and its incompetent AI, you should give maps like A Wind of Thorns or My Brother's Keeper a try.

[Psychobabble]

Wasn't the H4 adventure AI an exact copy of the H3 adventure AI?
I'm pretty sure I read that somewhere.

I'm not sure where that story started but, yes, that was a popular explanation which floated around on the RT boards for a while and it does make sense wrt things like wood piles and allowing heroes to die. Unfortunately I read something from on of the H4 programmers about mid last year (sorry, I can't remember exactly what it was, it was something which was posted on CH) which specifically refuted that idea. The adventure AI was programmed, badly, from scratch.

[Corribus]

Indeed. But I can use this to make a claim that it doesn't have to be a game programmer who makes the AI overcome its inherent deficiencies. In case of H4, it is the mapmaker who can make the difference. Before you damn the game and its incompetent AI, you should give maps like A Wind of Thorns or My Brother's Keeper a try.

Can a mapmaker work around the problems inherent to H4 and create a good map? Maybe. I haven't seen one yet that has succeeded (and I've played a lot of the ones that are touted to be "great") so I don't play the game much anymore. I don't deny the fact that maybe a good mapmaker can, through enough scripting and playing and exploiting loopholes, make a map that is fun to play. I don't deny that at all.

But I will damn the game nonetheless. Maybe there are a handful of maps that are enjoyable to play where the AI doesn't get hung up. H3 had hundreds. Just because you can get something broken to work with enough effort does not excuse the fact that it's broken, nor does it encourage me to play the game further in the hopes that this time, maybe it will work. That's like going to a restaurant and ordering a hamburger, and when you complain to the waiter that it tastes like spoiled fish, the waiter says, "Well, sir, you shouldn't damn the burger until you try it with lots and lots of ketchup. It will taste good, then." Maybe it WILL taste good, if you happen to like ketchup. But underneath the ketchup it's still a bad burger, and I'm certainly not going to come back to the restaurant 50 more times in the hopes that once, I might get a burger that tastes like cow. Similarly, why should I waste my time playing 50 maps in the hope of finding 1 good one (or, god forbid, try to make on myself) when I can just play H3 and get a good experience 90% of the time?

Every time I bring up the fact that the H4 AI is awful, they hand me that excuse. "Well, if you work hard enough, you can kind of get the AI to do what you want a little." Yeah, and if I work hard enough I can harvest my own wheat, grind it for 36 hours, and bake a nice dinner roll. Or I can buy my bread at the supermarket.

N.B. I have Wind of Thorns on my computer, downloaded at the suggestion of Wimfrits himself, but I never played it because my Winds of War disc is in another state. So I am basing this on the 50 other maps I've played at peoples' suggestions that have not quite been much better (albeit through no fault of the mapmaker) than any other ones I have played.

[Pol]

Every time I bring up the fact that the H4 AI is awful, they hand me that excuse. "Well, if you work hard enough, you can kind of get the AI to do what you want a little." Yeah, and if I work hard enough I can harvest my own wheat, grind it for 36 hours, and bake a nice dinner roll. Or I can buy my bread at the supermarket.

LOL

Yes, they did an AI for H4 again from the scratch. I vaguely remeber as I read some article about (not on RT), but it wasn't mentioned here in any more details.

[gravyluvr]

As a fan of HOMMIV I know the AI sucks. The mapmakers do a great job of working it out. In HOMM3, the AI also sucks. The truth is that if you play the game long enough you figure out where you can exploit and beat the AI.

In single player in HOMM4, the challenge is not always beating the AI player but beating the map. There is a difference.

[Gaidal Cain]

As a fan of HOMMIV I know the AI sucks. The mapmakers do a great job of working it out. In HOMM3, the AI also sucks.

Then there's quite a lot of room in "suck". The H3 AI at least puts up a fair show and does things. Mostly, the H4 AI stands still, and when it begins moving, it runs into a big stack of monsters and gets itself killed. Being predicatable to someone who's been playing the game for a long time isn't the same as sucky, at least not with the current state of cumputers- changing behavior would imply learning, which I don't think any game AI so far has been truly capable of.

[wimfrits]

The adventure AI was programmed, badly, from scratch.

That's plain odd. It bears too much resemblance to the H3 AI to make it just a coincidence imo. Seems a bit like a PR statement.
But if they indeed designed the AI from scratch, they did a rather sorry job.

But I will damn the game nonetheless. Maybe there are a handful of maps that are enjoyable to play where the AI doesn't get hung up. H3 had hundreds.

That's for a good part caused by your own experience. H3 had hundreds, certainly. How much would that number be now, with your increased knowledge of the game and your experience with using exploits?
I'll bet it would be close to zero.

Just because you can get something broken to work with enough effort does not excuse the fact that it's broken, nor does it encourage me to play the game further in the hopes that this time, maybe it will work. That's like going to a restaurant and ordering a hamburger, and when you complain to the waiter that it tastes like spoiled fish, the waiter says, "Well, sir, you shouldn't damn the burger until you try it with lots and lots of ketchup. It will taste good, then." Maybe it WILL taste good, if you happen to like ketchup.

Hmmmm.... burger.
I think that's a bad analogy. If the mapmaker succeeds, the burger tastes like cow.

Which is basically the same with H3. If the AI isn't garnished with huge amounts of resources, better developed castles, possibly additional key heroes after the most powerful dies, heaps of extra creatures; there is little challenge in a game against the AI.
So the principle is the same for both games.

I do agree that H4 mapmakers need to do put in a great deal more effort to achieve a similar challenge level.
And can perfectly understand why you feel this way about the game.

[Corribus]

That's for a good part caused by your own experience. H3 had hundreds, certainly. How much would that number be now, with your increased knowledge of the game and your experience with using exploits?
I'll bet it would be close to zero.

Actually I still play H3 maps (not new ones - usually just replaying one's I liked before or the original camaigns) and even H2 maps from time to time and I still find them enjoyable, even if I know the best way to beat the AI. Even with exploits the H3 AI can give still give me a challenge.

[wimfrits]

Glad there was no money on that bet then
Good that you're still enjoying the game. I honestly wish I could say the same.

[Corribus]

Well, to be fair I still put H4 in the computer every so often... and I will admit that I do enjoy it for awhile. The enjoyable part of H4 for me always has been hero development. But that novelty always wears off pretty quickly because the AI is just too pathetic. It's really a shame.

[wimfrits]

To be fair in return, H4 is starting to get tiresome to me as well.

Good thing a sequel is on it's way

[Corribus]

Good thing a sequel is on it's way

Well I think we can all agree on that!