CIV 4 - AI Requests

[Psychonaut777]

warpstone why such the bad attitude? I'm willing to bet that the vast majority of people on these boards are just.. civ fanatics with no experience in making games. I think it is pretty fair for us to ask for a better AI from a 4th version of one of the best games made. The reality is that some companies are starting to make extraordinary leaps in AI. Having a larger team doesn't always equate to less ability to create a finished product. If you granulate it like Arathorn suggests, then it would be possible to have more than 2 people working together on it. Maybe Firaxis' budget isn't large enough to have a large team dedicated to the AI.. this is why I suggested I'd pay more and wait longer. In the end, it will be the playability and not the graphics or anything else that make the game a lasting success. Dedicating 2x or 3x the employees to the AI instead of adding those few to a massive rendering team could be a better investment in the end. Civ 1/2/3 are still popular despite their outdated graphics because of one reason - gameplay.

 

[Diogenes183]

Then AI should not send its settler units into a warzone past a city that I just took from it that is garrisoned with 20 knights and 10 trebuchet, gaurded by nothing more than a single spearman. (FREE SLAVES!!)

I will sometimes send settlers into a warzone to replace captured cities that I abandon, but I go about it carefully. Also, I only do it if i am winning. The AI should be never build settlers if it is getting annihilated, it should build fighting units.

 

[warpstorm]

No one is asking for an 'intelligent' AI which can react like a human and thinks about it's move by evaluating different possibilities.

Read through this thread and the other AI threads closer. Quite a few people are in fact asking for exactly that.

 

[eg577]

Okay...some are asking for that, but it would be an impossibility to design such an AI at even a adequate level with no bonuses. I'm really just objecting to the argument that a better AI has anything to do with major budget issues since a pretty darn good AI would have been possible without much vast amounts of more work.

 

[warpstorm]

How much time and effort do you estimate it would take, eg577?

 

[Jon Shafer]

4 joules and 8 hours!

 

[eg577]

I don't know how long it would take. Depending on how strong you want the AI to play without bonsuses is could take some time. I'd imagine that if they spent the amount of time in C3C improving the AI as they spent in the original release making the basic AI they could have produced a halfway decent AI. An AI that probably could duplicate Diety or Sid level challenge with less bonsues. IMHO, anything that could reduce the contrived nature of the hard difficulty levels is a huge addition in gameplay value.

 

[warpstorm]

Actually, they did spend a lot of time and money on the AI for C3C, but it was mainly to adapt it to the rules changes. To do what you say would have required an additional 3 AI programmers or so (because of the short timeframe that C3C had to be delivered in). This would have blown the budget (good, experienced games AI programmers aren't cheap - many are over 6 figures a year). Then you'd have the problem of what to do with 3 specialist programmers till you need them again. Most likely throw them into the more general programming tasks, but that means you are paying a premium rate. (Laying them off isn't really a good option if you want to hire good talent in the future - word gets around). It's not an easy problem if you have to deal with actually making it happen.

 

[eg577]

Okay, I'm not an expert on developing computer games. I don't imagine you would need use 3 of your premium programmers to improve the AI, you just need 1 programmer who's good at civ and knows the code. They're supposed to spend money on the game anyways, and the number 1 way they could have improved this game was to make it challenging without being so contrived. Is that really too much to ask for?

 

[warpstorm]

You said, "if they spent the amount of time in C3C improving the AI as they spent in the original release making the basic AI". This would have taken about 3 three more programmers due to the much shorter schedule for C3C. I was responding to your thread as written. I further assumed that you would want people as good as those who worked on the original code, rather than using someone of lesser talent to make more sophisticated code.

 

[dexters]

Read through this thread and the other AI threads closer. Quite a few people are in fact asking for exactly that.

Nope. Not this thread.

Re-read my post again

Guiding Principles
--------------
1) No going back -The AI in Civ 4 should not regress back to a pre Civ 3 level.
2) The AI should at least be at good (in terms of overall performance) as the Civ 3 AI

I think its a pretty reasonable baseline expectation for a game that will likely be released 4 years after Civ3. If Soren and Co can't learn from their Civ3 work and build a better AI system then Civ4 shouldn't be made at all or it'll just be the same game with prettier graphics.

I wanted posters to add their comments about what they would like to see the AI do. I really don't think asking for an AI that is at least competent at managing the major features of the game and maybe marginally competent when it came to doing stuff that may require 'human heuristics' is calling for an Intelligent AI.

My original post even allowed by AI cheats to replicate human 'short-cut' heuristic strategy planning. I don't know what features there will be in Civ4, but the best Civ3 example is how AI seems to know resource locations before they are discovered, although other subsystems like workers and units are unable to see it (leader AI just knows where to settle).

Since the AI isn't intelligent and doesn't work on the human mental capacity of settling on a spot (say tundra) with the expectation of an oil resource popping up, giving AI leaders 'guidance' on where to settle becomes a close apporximation of humans fishing for resources by settling in tiles where they are known to pop. That's a cheat but I'm ok with it and I'll be ok with it if something like this is used in Civ4 (given the parameters described above). Cheap cheats however is another story entirely.

Now, lets get back on topic please.

 

[Arathorn]

Without knowing the structure of Civ4 more thoroughly and with comments like "Eliminate corruption, pollution" in Soren's material leading me to believe the current models will not be continued, this is hard. I will focus on some things in Civ3 that could be done to improve the AI significantly with minimal work. I do this because I know Civ3 and know nothing of Civ4.

"Small" improvements which could pay big dividends:
- Teach the city governor about corruption so he can pick the best tiles. This appears to me to be relatively easy -- the corruption threshholds are fairly obvious, so that food can and should be maximized once a certain threshhold is reached. Might be time-consuming each round but should be simple to program, I believe.
- City needs communicated to workers. Somewhat harder, I believe, but it should be reasonably feasible. I mean that if a city is drowning in food, the workers should know to mine more. If the city has lots of mountains/hills, the workers should know to irrigate around that city. Definitely non-trivial but feels like straight-forward calculations that the computer should be good at.
- City placement. They seemed to have this down in PTW and regressed in C3C. Hard algorithm to write, I know, but it should be better.
- Teach the AI to use the lux slider. Just some brute force on this should be sufficient. Set lux at 0% and optimize each city and count total food+shields+commerce. Set to 10% and repeat. Might make the interturns longer but this should ideally be done while the human is playing -- background processing.

These few changes would, in my best guess, make the AI significantly more competitive and should not be terribly difficult to program. I wanted to provide some concrete examples, too, of what I think could/should be done better.

Arathorn

 

[warpstorm]

Nope. Not this thread.
I think its a pretty reasonable baseline expectation for a game that will likely be released 4 years after Civ3. If Soren and Co can't learn from their Civ3 work and build a better AI system then Civ4 shouldn't be made at all or it'll just be the same game with prettier graphics.

This is perfectly reasonable.

Civ4 will be different enough in gameplay that the AI will have to be written from scratch (actually Soren said everything will be written from scratch), because the rules will be different enough that playing optimum Civ3 will be a quick way to lose. Civ4 will not be Civ3+, but rather an entirely new game loosely based on what has come before.

The best thing Firaxis could do is to get the best Civ3 (and Civ2 and maybe evn CTP) players in to actually play Civ4 ASAP and learn what the winning strategies are and have the computer players copy their play styles (build orders, forces compositions, research priorities, government forms chosen, etc.). Once the top players now how to develop cities, etc. it would be a lot easier to program these "canned" strategies in.

 

[eg577]

Sort of on the corruption note, the AI should also be able to micromanage shields and food in it's cities. Like if a city is making 6 shields for a 10 shield project then it should shuffle the workers so it makes 5 shields and an exrta food or commerce.

I think the AI needs desperately needs logic in how it builds improvements too. It seems that the AI basically thinks every city should get every possible improvement. Some cost-benefit calculation should be done to avoid building things like banks and universities where they would merely drain his treasury or prioritizing courthouse for distant cities.

edit:
...and if it is reasonable to put canned strategies in civ4 then it would have been reasonable to do the same thing for civ3.

 

[Jon Shafer]

Let's not forget the vast majority of players never get above Regent (the online communities are only a fraction of the total players). That means, from a purely business standpoint, improving the AI a great deal isn't a priority, but leaving things wide open for (Python) modding so that the community can make their own great AIs makes more sense.

Not saying that I don't want a stronger AI... just pointing out facts. As long as the game has the capability for programmers like myself to improve things (especially the AI) I won't complain too much. I love messing around with things like that and seeing what can be improved.

 

[Sirian]

Let's not forget the vast majority of players never get above Regent

Are you sure about that? I remember knowing a couple dozen folks who played Civ1 and nearly all of them could play on Prince, which comparatively speaking is like playing on Monarch in C3. Most of those could play on King. More than half could play Emperor.

A sampling of my friends may not be representative of the general marketplace, but let's not be too condescending about thinking that someone has to belong to the online community to have any gaming talent.

That means, from a purely business standpoint, improving the AI a great deal isn't a priority

I wouldn't be too sure about that, either. Most of the mass market looks at "best sellers". There are lots of marketing factors involved I cannot analyze here, but one key element is the much-coveted "word of mouth". Whether it's a movie or a game, a book or a TV show, some folks who are hyped about it will go and check it out, then blah blah blah to all their friends, posting on their blogs, etc.

A lot of sales will happen because Player A buys the game and gets hyped about it, and hypes all of his or her friends, whereby Players B, C, D, and E run out and buy a copy, so they can share the experience with Player A.

LOTS of games have good graphics and sound. A decent number are grandly fun to play. How many have a great AI?

AI is part of the experience. It's not just about winning, either. The AI impacts the "atmosphere". The more life the game maker can breathe into his or her design, the more engaging the game could be. This is not just about "which difficulty level you play". The difficulty levels exist to hand the AI a handicap so it can continue to provide a fun experience even to skilled players. BUT THE AI IMPACTS EVERYBODY. It's the partner, the opponent, the rival, the animation of gameplay.

I'm sure it's an afterthought for many game producers, but Civ3 has the best AI in the genre, so it doesn't seem to be an afterthought for Firaxis.

I guess we're mostly spinning our wheels here, though. We'll know in a year or two as to what the truth turns out to be.


- Sirian

 

[Jon Shafer]

Well, yeah, AI can be a big factor on how people view a game... but usually only if it's as dumb as a rock. As much as some people complain here and elsewhere about how stupid the AI is, as you said, it's the best in its genre and isn't all THAT bad. The problem is when you get the point where you can micromanage every single food and shield you will overwhelm the AI simply because of your own diligence. It obviously isn't perfect, and has many flaws that the average player does not, but it can usually hold it's own against a normal player.

Because of all that, as I said, I don't think that improving the AI a significant amount would be a goal from a financial perspective. How many people complain about how stupid the AI is compared to how many people as a whole play it? Where are the resources best allocated to maximize the amount of customers attracted to the game? Usually it's not Deep Blue.

 

[Sirian]

Improving the AI isn't just about making it more competitive. There is also making it behave more in line with the game's context.

Besides, your line about most players not being able to beat Regent is a red herring. A typical game of Civ sees one human player and seven AIs on a normal map size. That's 7 to 1 odds, all other things being equal. To measure the player's performance against seven AIs and claim it as a meaningful measurement of how inherently strong or weak the AI is doesn't add up for me.

Players could lose scads of games and still notice the AI making "dumb" moves left or right, or lacking personality at the diplo screen, displaying a major gap in its capabilities to handle game elements, becoming predictable, or any of the other areas where it underperforms.

Fact is, -I- did not buy a new Civ franchise installment after Civ1, until Civ3 came along with a dramatically upgraded AI. The AI was -THE- selling point for me. I would simply not have bought the game without being told that the AIs would not "play as a team vs the player" -- something I had grown to despise from the earlier Civ games with a negative intensity that could frighten dogs and small children if honestly expressed.

Limitations on the AI are why I gave up on Civ1. I outgrew it and left it behind. It wasn't its inability to beat me, but rather its inability to play fair and its inability to throw me any new surprises or keep me guessing.

I did not buy Civ2. Someone gave it to me, years after it had been out. I have never even seen SMAC in action. Sid made a great game in Civ1, but I'd been waiting a decade for the AI to get some major renovation before I bought back into the concept of Civ. Perhaps I'm alone on this score, but somehow I doubt it.

I repeat, the AI is the player's opponent and companion, in most cases. Your definition of "improve the AI" seems somewhat one dimensional, Trip. I'm not sure you and I are thinking of the same concepts with those three words.


- Sirian

 

[Lennon]

I believe that the reason they don't put more effort into improving the AI (not saying the effort is zero, mind you) is simply that it doesn't pay off. Joe Average has an average intelligence. He beats the game at normal difficulty. Joe Average also happens to make up for the vast majority of Earth's population (hence also Earth's Civ or other computer game players). So they basically develop an AI that is good enough to challenge most people and for the harder difficulty levels it simply takes less time and effort to just give the AI bonuses (in other words: make it cheat). But there will always be superhumans out there who will tell you the AI is too stupid and complain that the only way for the AI to pose a serious challenge is through cheating, but these people are a minority, hence not a priority. This is the problem with any computer game and it is also a shame, but until we learn better ways of easily programming a smart AI we'll have to live with it I'm afraid.

 

[dexters]

I think though Civilization as a franchise game appeals to a higher end 'educated' crowd. Probably highly educated and people with an interest in history and military strategy and diplomacy.

Civilization isn't in the same league as say Need for Speed or Madden, which I think 'average joe' or 'joe average' term would be more fitting. Granted there will be casual gamers out there who pick up a Civ game out of interest, I think most people who show interest in this franchise tends not to be 'average' but above average in their education - note that I didn't say intelligence as I think many 'average joes' are quite smart but they chose not to apply themselves in the fields associated with smart people.

Edit: many good points made. I'll add them to the top of the thread when I get some time to sort through the posts.

Thanks guys and gals