Disenchanted

[Analyst Redux]

Originally posted by Captain


Right! Very true. But HOW do you get an AI that's capable of that?

Or is MP the only resort?

And if so, wouldn't you want it to have Simultaneous turns (like Risk 2, a vast improvement over Risk 1 where you could railroad over your enemies so easily if you got the cards)...
If they make MP, I hope they figure out a way to make it Simultaneous turns (submit orders, then resolve) - which would also make it more playable for people who don't like PBEM!

Simultaneous order resolution is orders of magnitude more difficult to program, in terms of AI strategizing, than a TB system. I've seen some heavy duty mental firepower take aim at that problem and fail, within the context of the game of "Diplomacy", which presents a much simpler set of strategic problems to solve than Civilization.

I think, as you suggest, that multiplayer is the only realistic future for this title, if it intends to remain interesting. The absolute best AI that the absolute best programing can come up with can't hold a candle to the deviousness of even an average human opponent. There came a point where I became bored with SMAC, and took it off my HD, but Civ II stayed on--because I was playing it against other people online. I haven't played either of those two titles against the AI in years. Rather unchallenging and uninteresting. Civ III's shelf life on my HD, absent a viable MP alternative, is likely to be short, which is really too bad. I like these games.

[Analyst Redux]

Originally posted by Sze

Do you think they dumbed down the combat system to make the AI better at estimating its chance of success?

Yes.

No other explanation strikes me as more convincing, given the overall pattern of changes to the game.

[gnomos]

Originally posted by Velociryx
Ideas for the Editor


2) A "this unit cannot attack cities" flag and/or a "this unit cannot leave your Civ's borders" flag (useful for creating Partisan/Militia defenders (especially colorless units) that could be used to whack settlers and other denizens from neighboring civs who REPEATEDLY violate your borders. If the AI will not let you be, there should be a way to keep all comers out without starting a world war. Such units, as I envision them, would be powerful enough to go head to head with whatever the standard defender is, assuming he's out in the open (but probably not have sufficient punch to actually attack a town, except in large numbers) to keep your rivals at bay. Trouble is, especially in the ancient era, there's simply no "room" given the default A/D/M values to do that (unless you wanna totally overhaul the entire system :: shivers:: ). So, a flag like this would allow me to create a colorless swordsman who could NOT leave my borders, or at the very least not capture enemy towns. (As it stands now, colorless units can attack and capture towns without causing a general war....the AI, in fact, does not even respond....not good! (as it stands now, I'm creating units that are x/x/1 - All Terrain as roads and requiring such units to cost one or more pop points as well, and even still, given their ability to capture cities without triggering a war, they're verging on being unbalancing). Such flags as mentioned above would open a whole HOST of interesting possibilities.

-=Vel=-

Sounds like the game needs INS agents to take care of illegal aliens. I think you should simply be able to "arrest" or kill foreign units in your territory without starting a war.

If the AI sends a military unit into your territory, you should be able to declare an international incident. Things like this should be treated more seriously by AI civs, getting a bad rep should be more meaningful and avoided by the AI.

[Analyst Redux]

Originally posted by jimmytrick
Thanks for the clarity Analyst!

If you are correct, and I think you are, then Firaxis has tried "addition by subtraction" in their effort to make a challenging SP game.

Wouldn't it follow then that they would have to "know" that in constructing multiplayer they need to go 180 degrees in the other direction and give us more instead of less?

I hope so.

I doubt it though.

I doubt it, as well, for two reasons:

(i) Substantially altering the game for MP purposes is diametrically opposed to their mission statement; and

(ii) These designers' actual track record, in Civ, Civ II and SMAC is not to design an MP version of the product, but to slap MP on the back end of the product and hope to hell it works. Their batting average so far is an unpromising .333--only Civ II had an MP implementation that was worth the bother to code.

[gnomos]

Originally posted by Libertarian


Oh, dear. Where to begin?

Heehee, spoken like a true software developer.

Hmm, what would it take to add a simple flag to keep a unit from attacking or moving into a city? Well, I'm not a Firaxis employee, have no acces to their code, but I'll try to make a good guess about all the changes this would entail.

Preliminary design:
Update design documentation: add data element to object dictionary, along with a description of it's purpose, where it is checked and set, so the non-techies can read up on what it's for. Get these changes reviewed to ensure they are correct. Specify any changes to data files affected, such as the Civilopaedia and the bic file. Get these proposed changes reviewed for correctness. Get a mod number to track source file revisions, put it on all documentation and use it later to comment changes in the source code, so anyoen else who works on the code can find the correct documentation for the changes. Write an estimate for how much effort is involved in the change at all levels from design through testing and migration to production. Get the estimate reviewed and approved. Estimate goes to accounting for later comparison with actual effort reported for the change, so that estimating methods can be tracked for accuracy so job planning is reliable. Report all effort expended in preliminary design, including time of reviewers.

Detail design:
Hunt down the source files where the changes need to me made, detail where all the changes need to be made in the code, write up a change request or whatever form of design document Firaxis uses, get the design change reviewed for correctness. Write a test plan for testing the code changes, and get the test plan reviewed as well. Log effort for code design.

Coding:
Check affected source files out of revision control with write previledges so they can be modified.
Add data element to object definition.
Add initialization code to object constructor.
Probably don't need to add code to object destructor.
Write functions to set flag value, return flag value.
Add code to game editor to display flag to user, if it's set on a unit or not.
Add code to game editor to modify flag on units people mod.
Add code to game editor to write the new flag to the bic file.
Add code to game editor to read the new flag from the bic file.
Add code to game executable to read the flag from the bic file.

(maybe the flag is just part of a numeric mask, and doesn't need specific code to read and write it from files, but maybe all available codes are currently in use)

Add code to game to handle moves/attacks in cities.
Add code to AI so it doesn't choose to try to move unit into city or attack a city with it, otherwise it might infinitely loop trying to move a unit into city.

Compile and unit test changes. If things seem to work correctly, check source files back into revision control system, update migration documents. Log effort expended for coding.

Testing:
The QA department pulls the new versions of the source files to their stable test paltform and compiles the binary. They test the program through a variety of scenarios. First they verify that the changes actually work. Then they perform regression testing to verify that the changes didn't break anything else. If code passes all the testing QA signs off on the mod, and approves the versions to be migrated to the production platform to go into the next patch. All testing time is logged for accounting purposes.

Now, I don't know if Firaxis runs a methodical and professional software shop, or if they just hack out their code willy nilly. But most experienced software people will tell you that you have to dot your i's and cross your t's if you're building a complex piece of software and want it to be reasonably stable. And all of that takes work, and a lot of attention to detail, and strong methodology to ENSURE that you don't shoot yourself in the foot very often.

[gnomos]

Originally posted by nato
Yeah Libertarian, that is what I was trying to get at!

Making large changes is really difficult ... and because the game is already sold, no longer in the process of being made, I don't think Firaxis can or will make large changes.

All they will want to make are small changes, so those are the only realistic ones to hope for I think. However, just because they have to be small doesn't mean they can't be real important.

Well, I actually expect Firaxis to come out with a multiplayer expansion, and I think this presents an opportunity to get some of these other things changed, since the expansion will probably be sold and so will generate more revenue for the company. It would be nice if someoen from Firaxis would tell us the scope of the work they are currently doing, and let us know if people here are spinning their wheels, or if some of these changes have a real change of making it into an "expansion".

[Dis]

good analysis, analyst

And I'm inclined to agree. Although your bit about city improvements I don't agree with entirely. And Sun Tzu's does provide a building in every city, as with the Pyramids. And I doubt the ai sells the buildings made obsolete when they build these wonders. But I digress.

and I do enjoy the easier ai with SMAC over Civ3. And that is saying something. Because I really don't like science fiction games. But I'll take free market economy any day over whatevery whacked out system civ3 has . Social Engineering . Yeah I thrashed the ai. And rovers were too powerful along with choppers. But hey, I had fun for a while. Not nearly as much as civ2, but not bad considering I don't care for most of that scientific mumble jumble .

[gnomos]

Originally posted by Analyst Redux

I'm tempted to agree with your "Potempkin Villiage" characterization of the current offering, but I don't think it's quite that bad (or cynically intended). I can also point to examples of new feature/rules that have been added that the AI seems to do quite well. Pop-rushing, for example. At least, if the AI isn't effectively pop-rushing, I can't figure out how else it is doing so well in the early game. I also do see a refinement in the mass-to-attack formula which I count as an improvment over the prior offerings. Other improvements like that. But the real improvements are incremental, unfortunately, and absent the simplification of design, probably wouldn't have been particularly noticeable, IMO.

At higher difficulty levels the AI gets free units when it founds a city. Therefore it doesn't need to build garrisons to maintain happiness, so its cities grow faster. This also provides a lot of extra units to explore new territory. I saw a post once from someone who built a tiny map with one other civ, and looked at the units the civ had on diety on turn 2 after it founded a city. It had something like an archer, a worker, a spearman or something, and 3 warriors. Instant happiness.

[korn469]

Analyst

welcome back, its good to see you!

First thing is you are right on the money when it comes to how firaxis engineered the rules to help out the AI. On the firaxis site, there was a link that talked about an original game that Soren had programmed, and he commented that he had to designed the rules to make the AI in that game competitive, and I'm sure that Soren more than any other person at firaxis is responsible for asking for rule changes that lets the AI perform better.

I also agree on your ten point list of AI changes that makes the AI play better. However, not all of those ten point harm the playing experience, and even some of those that do make the game a little more restrictive pay off big dividends in making the AI able to compete. However, a few of the rule changes really constrains the player and therefore harms gameplay. While it truly is a shame that AI hasn't advanced to give players an actual challenge yet without massive advantages Civ3 does have a solid AI. It still underperforms when compared to humans even with all of the rule changes in its favor, but it is far beyond the AI in SMAC. I don't think that it is only in rule changes, but i do think the overall AI is slightly better than previous incarnations, and if it had the combat bonuses on higher levels like the AI in Civ2 then the AI would be truly dangerous.

Many of us posted saying that good AI was our number one concern, but its kinda like in the story the monkey's paw, wishes get answered but always with grave consequences.

Not everyone was confident that the AI could be even mildly challenging, i'll quote yin on this (first because everyone knows him, and second because when yin makes a point you never miss it ) this is from yin's review of civ3

AI

I leave the review on this section because the faulty AI in Civ 3 is by far the most glaring failure of the game. The only times I was ever impressed by the computer AI was when it refused to trade a valuable commodity to me unless I first signed a peace treaty with its neighbor. This was great. Until I realized EVERY trade has an almost identical "request." Why should I sign a peace treaty just to buy some coffee? Also, and this is important, EVERY trick that worked against the AI in Civ 2 is working EVEN BETTER in Civ 3, mainly due to the fact that the computer AI is so incompetent at securing resources for itself that it is crippled from the start and simply never catches up. Frankly, I am stunned that such a clearly vital requirement for the computer AI was seemly overlook entirely. Again, I might not be so critical on this point alone had the computer been programmed to make good use of trade and diplomacy to make up for it, but as noted, the computer AI is nothing more than a nagger or a warmonger with little or no intelligent moments in between.

Then again, Firaxis never promised anything with the Civ 3 AI. In fact, they never even mentioned it. On this point at least, Firaxis delivered all they promised.

CONCLUSION

Overall, while Civ 3 promises a great deal over Civ 2 in a number of potentially fascinating areas, the game simply fails to deliver due mainly to horrendous AI

Now he maybe be right in a way, because if the AI was better the game wouldn't need to be dumbed down, but I do not think that rule changes to improve gameplay permanently prevents any improvement in gameplay. I feel that there was a general inattention to detail that harmed civ3 far more that dumbing down the game. Certainly using the U.N. to implement something akin to the Planetary Council in SMAC would have improved gameplay without putting a much strain on the AI. One of SMAC's greatest achievements was on focusing the player's attention away from mundane tasks. Being able to vote on who was the Council leader focused the player's attention away from micromanaging build queues and warmongering. By involving the player in task outside of the ordinary realm and giving the player other smaller objectives to reach (like winning the vote for Council leader) it made the game much more emmersive. I'm sure the minute that the player spent voting (especially if it was close) was worth ten minutes of moving terraformers around. Civ3 skimped on the extras, and it didn't add in enough macromanagement options in the interface so it appears that Civ3 has more micromanagement than previous incarnations, when that is untrue. Civ3 has less micromanagement than either Civ2 or SMAC in my opinion, however while in SMAC we could take a vaction from micromanagement by either watching a wonder movie, or listening to one of the tech descriptions, or reading an interlude, or participating in the Planetary Council; in Civ3 the game is like a task master and it forces back to micromanaging the game immeadiately.

We all complain about late game tedium in Civ3, yet this is kind of a mixed blessing. In SMAC the player would simply end the game when it became tedious because they had won long before the late game came around. The average player won shortly after shard weapons and copters became available, the good player could win with fusion powered needlejets. Virtually all players would have an economic advantage over the AI by midgame, and those players who had mastered the power of crawlers could be so far ahead of the AI it wasn't funny.

Unfortunantly when comparing the scope and variety of intellectual challenges, SMAC beats Civ3. From designing your army, to creating an utopian existance for your subjects SMAC presented the player with many options that don't involve building or warmongering directly. SMAC has better, deeper, and more interesting diplomacy on every level except for the trade screen where Civ3 soundly beats it, but that one victory is a hollow one because SMAC raised the bar for diplomacy and Civ3 only halfway delivered on it. Civ3 removed all of the nuance from diplomacy and basically made boil down to "i have this item, what will you give me for it?" then its over. Diplomacy in SMAC had more nuance, and should have been a stepping stone, and not a high water mark. Then their is unit design, Social Engineering, Atrocities, the Planetary Council, Economic Victory, etc. Civ3 lost alot of SMAC's scope and it suffers for it gameplay wise, because players have to focus entirely on parts of the game which have been dumbed down to give the AI a fighting chance. Then to top off that, it seems like firaxis forgot to add the buildings for the industrial and modern eras. There are far too few buildings in those era, and the player quickly runs out of everything to build.

While certainly changing some of the rules to accomodate the AI is a necessary evil, if Civ3 is ultimately considered a failure it will be because it didn't add any gameplay to the series, and in fact it retreated from some of the gameplay pioneered by SMAC. This is a real shame too, because Civ3 did introduce new concepts like culture to Civ3 and these concepts could have been used to increase the number of things a player could do.

I still however have some optimism that with more patches and a good expansion Civ3 will be a worthy member of the series. Yet, without some improvements Civ3 will be a pretender instead of a contender.

[gnomos]

Originally posted by Velociryx
Good point. My solution would be to simply make the upgrades free for the AI. They already get cheats handed to them, so one more wouldn't be crushing, and it'd make them more of a threat in the Industrial/Modern eras....

-=Vel=-

I hate it that the only way Firaxis seems to be able to make the AI a challenge is with freebies. Sigh. How about this? Gradually reduce the cost of AI unit upgrades as you jack up the difficulty level. I'd like to see the AI at least making some choices about what to upgrade when, not just automatically have everything upgraded. But obviously the AI needs to put a lot more emphasis on upgrading units. I hate picking spearmen out of the treads on my tanks.

[Sikander]

Some truly excellent posts here, especially in the last couple of pages. A lot of this sort of thing was predicted long ago in the wish list Civ 3 forums. I tried to explain why we would never see a decent AI for a game like Civ until there were huge advances in AI technology and computing power. I used a comparison between Civ and Chess, an AI that can beat me regularly using only my home computer and a thick book of openings and useful patterns.

Chess: 64 identical squares which cannot change

Civ: Thousands of squares of numerous different types, each of which can have a number of additional properties which can change (ie roads, cities etc.)

Chess: 32 units (16 per side), one of which can be moved on each player turn (exception: castling). Limited upgrades (pawn reaches back line)

Civ: Potentially hundreds of units, all of which can move (in any sequence) every turn.

The list gets crazier from here obviously. I feel a bit guilty because on that long lost forum I suggested that the game had to be simplified in some ways in order to give the player a good challenge. (Not that anyone listened to me, but the idea is obvious from a design standpoint) I suggested giving the player a lot more challenges in maintaining control of his empire (which historically is a lot more realistic) and turning over a good deal of the operational tasks over to AI generals and ministers who would help the balance by the fact that they would be no better than the generals and ministers of the other civs. The advantage would be that the player could be kept busy solving management problems which would require thought and creativity and would hopefully be satisfying. Where the AI could not compete in management efficiency the choice could be made to either let it alone (the SMAC solution) or to change the game in certain ways so that the poor AI doesn't have to play the same difficult game that the player does. These areas should of course be far from the HP's view, not in direct combat for instance.

I have always understood Sid's disappointment with the game as it tended to be played, as a wargame. Civ has always been a wretched wargame, in many ways 30 years behind the boardgames being released at the same time. The idea behind the game was nonetheless fascinating, and I have been somewhat surprised to see how many people tend to treat these games as turn based warcraft clones. It is this tendency which has pushed Civ 3 over the edge, (though equal credit must also go to economic factors. It is obvious that in comparison to SMAC this game has a fraction of the graphics, voice overs etc.)

Given the choice between a closer (but duller) battle with the AI which I will tend to win anyway, or a sandbox game (like SMAC tends to be once you learn one of the many strategies to beat it), I'll take the sandbox, at least in a competition between Civ 3 and SMAC. At least SMAC was a lot of fun. I enjoyed comparing my performance and methods with others playing SMAC, and taking the many dynamic elements of that game and twisting them into new configurations.

Civ 3 should have been a brand new design, but instead it was just another tired retrofit of a game engine which is far past it's prime. After playing many hours of Civ 1, Colonization, Civ 2, Master of Magic, SMAC/X etc. the last thing I needed was to have the game engine stripped down and left naked before me. I know it too well already.

[Libertarian]

Analyst:

A fantastic contribution. Thank you.

Your theory seems to go hand in hand with what I perceive as Civ3's greatest weakness: the disappearance over time of interesting decisions. Frankly, my biggest gripes (unary unit movement and bizarre unit activation sequencing) would be less of a problem if the units presented me with interesting decisions to make. But because the game is dumbed down, there's no decision whatsoever. I must improve the terrain. Period. That's it.

Playing Civ in the late game reminds me of those dreams wherein you're trying to move but can't. You feel like you're slogging through waste-deep syrup for no reason other than that you can't do anything else.

-----

Vel:

I had recommended a book to a guy with whom I worked. He was a recent convert to libertarianism. The book was Atlas Shrugged, more than a thousand pages. Over the course of the next few weeks, he came in each day and discussed with me the passages he'd read.

When he finished it, I congratulated him. He said that reading it was a life changing experience. He also submarined me with a surprise announcement, "I'm dyslexic, you know. That's why it took me so long." No, I didn't know. Turns out he's not just dyslexic, but severely so. He showed me how he had to sort of translate in his mind as he read, much as I would have to do reading upside down from a mirror.

He had put himself through Wake Forest University, and is now a very successful general contractor. I admire him very much, and consider him a role model.

You might be used to playing second fiddle, but while you were on break, you got moved to first chair. Sit in it with pride of accomplishment. Just be sure you don't sit on your laurels, though. They belong on your head, and not under your ass.

[moomin]

Analyst: great work. I've thought about this stuff too, but I've been too lazy and indifferent to even consider the systematic approach you've made. Impressive.

If I could offer some kind of comment on the stuff you posted, the "dumbing down" seems mostly to concentrate on the map and the units moving on same. Many meta-functions are not really done considering AI. The diplo victory, for instance, is considerably weaker than in SMAC, not only in in how it's done but in how aggressivly the AI pursues it. I occasionally let the AI build the UN back when I still played Civ3, and it did not do simple things like trying to bribe everyone before taking a vote - it would lose as often as win.

Compare this to the threats and bribes in SMAC. The dumbing down seems not to have done much for the AI in this instance. But then again there's the possibility of the entire diplomatic victory being one of the features that got axed to ship 'till christmas, and what's left is just the hook upon which Firaxis intended to hang something interesting.

[moomin]

[OT] Atlas Shrugged

Originally posted by Libertarian
Playing Civ in the late game reminds me of those dreams wherein you're trying to move but can't. You feel like you're slogging through waste-deep syrup for no reason other than that you can't do anything else.

[...]

I had recommended a book to a guy with whom I worked. He was a recent convert to libertarianism. The book was Atlas Shrugged, more than a thousand pages.

Lib, you amaze me. While I agree that end game tedium is a showstopper for Civ3, I always thought AS was a showstopper for objectivism. If you managed to slug through that you really should have no problem with Civ3. And giving it to a dyslectic is nothing short of sadism, you know.

This isn't a political flame, btw. My take on AS is much like my take on Civ3 - nice idea, horrible execution.

[Libertarian]

Reading Atlas Shrugged is like reading any other piece of literature. You need a discerning eye and mind. Take what's good and discard the rest. I did that with Objectivism. I like, for example, its core ethic. But I reject its core aesthetic.

Rand's work is so easy to parody. The characters don't converse; they give speeches to one another. All I wanted my friend to take from the book was exactly what he took: a man's achievements belong to him. Peaceful honest people ought to be free to pursue their own happiness in their own way, free from the coercion and fraud of others.

If you're interested, here's a thread at Straight Dope, where I wrote a parody of Randian love scenes.