Well, that doesn't count - a city on a 1 tile tundra island with no substantial resources in vicinity, well maybe only fish, but that's still too little for anyone to want to spend more on upkeep
Announcement
Collapse
No announcement yet.
Damn You Blake!
Collapse
X
-
Originally posted by Maniac
Is Blake's mod only for unmodded Warlords, or would it work as well with for example the Chinese Unification scenario?
Comment
-
It can hardly be worse than the stock AI.
For example, I recently played the Peloponnesian War scenario as a 2 player PBEM and the AI vassals on both sides were pretty useless, to put it mildly.
Try any of the scenarios without the latest patch and the AI is poor and easily exploitable. At least Blake's AI can build a decent economy and will therefore do better.Never give an AI an even break.
Comment
-
Originally posted by Kuciwalker
On the other hand, Blake's AI is an expert system for the core rules. If you had a big difference from the core rules, say for the China scenario, couldn't it play even worse?
For example, if I modify some civics, will AI:
1) recognize then differently,
2) or chose them in same way as with original rules, regardless how different they are
If 1) is correct, then it's good (choosing civics depending from their values, not their "name"). But if 2) is right then it's bad, really bad for moders. Exempt if they are coders too, capable of making customized AI for their modified civics.
Comment
-
Originally posted by Kuciwalker
Aren't civics defined in the XML?
That is the problem.
If moder just changes XML values and AI is hardcoded to like some civic more then other based on their original values (without checking current XML), then it's bad thing for moders (inefficient AI in such mod, without custom DLL for modified AI).
Comment
-
Hmmmm. Even Monarch AI doesn't do well on Highland maps. I think they put down too many cities that remain low pop and uncottageable. Lots of cities in the tundra region that are trying to pick up a single Iron or Deer. I'm outresearching them by a long way, but then I have a capital with 12 settled GPs and can run 100% science at +50gold a turn.
Is it possible to teach the AI to pay attention to what map settings are in effect and adjust play accordingly?
Comment
-
I'm sure it's technically possible, but I think it might be like having a smarter AI by difficulty level: too hard to do to be worthwhile to implement. You'd basically have to write a different AI for each map type, and switch depending on settings.Age and treachery will defeat youth and skill every time.
Comment
-
Originally posted by Blake
AFAIK it will work for any scenario which does not have a CvGameCoreDLL in it's mod/Assets/ folder, AFAIK none do. If that isn't the case, then I'm pretty sure that putting the DLL in the Assets folder would cause it to use the improved AI.
But as others are wondering, does the AI in Warlords 2.08 and in your further work look at what effects the civics give, or do they just think "Don't use caste system" or even worse eg "Don't use the fourth civic of the third civic category"?
Comment
-
It looks entirely at the effects and magnitudes, altough it is tuned somewhat for the default civics and isn't properly normalized by column.
For example Universal Suffrage is worth 8.33/city, while slavery is worth "6/city" in peace and "12/city" war. If a mod has a column with a civic with cash rush and a civic with pop rush (and no other effect), it will typically choose US in peace and Slavery in war, even though that's not really meaningful, both are rushes.
A better example would be Emancipation, which tends to weigh in at somewhere starting around 40/city but is basically uncapped and could easily be 160/city, any civic with an emancipation effect is going to override any other civic in the column, assuming it isn't the final civic of the game. This hasn't been changed because it's not a problem in the default gameplay.
Another example is that some civics with multiple effects are unfairly weighted in favor of one effect. For example for Organized Religion nearly all the weighting is from the +25% to buildings, rather than the ability to build missionaries without monasteries. If a mod for example had a civic which lets you train "uber overpowered ninja assassins" - and that's the only thing the civic does, then that civic would not be reconized as being good, as the civic code does not try to assess the overpoweredness of the unit which is allowed to be trained, or how much is being gained by the free training ability. It assumes that the unit is so-so and the alterantive option is reasonably cheap.
And the nationhood drafting effect is rated assuming it's a reasonable unit. Again, it wont make any distinction between drafting uber crappy units, or uber-overpowered units.
The changes I've made are certainly no worse than the old setup, in fact I tend to think I've better normalized the weightings, however nor did I go to any special effort to ensure that every possible mod (or even any particular mod) would work sensibly.
I think in general they will work well with most sensibly designed mods, but it certainly wouldn't be hard to design a set of civics specifically to "embarrass" the civics code.
Comment
-
Wait a minute, it seems Blake has made an even better AI? I can't even beat the original AI on noble I consider myself a CIV veteran, but I lose against the AI on a regular basis and nothing is modded. I am used to losing now so I don't even realize it anymore. I never say "oh let me play a game of CIV4", instead I say "let me go ahead and lose a game of CIV4 today" When and if I ever get good enough, maybe I can try Blake's AI, but its been over 10 years now.-PrinceBimz-
Comment
-
Do some strategy reading, it wont hurt at all-- What history has taught us is that people do not learn from history.
-- Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning.
Comment
Comment