Article Concerning Flaws in the Rules

[Sirian]

I've written an article detailing what I've found to be the ten worst flaws/loopholes in the rules, accompanying exploits, why I believe these are problematic to the lifespan of the game, and what might be done about each of them in a future patch.

The article may be found on my Civ III fansite. Follow this link:

Sirian's Great Library

Any comments you may have can be posted here. I'll check back in and reply from time to time.


- Sirian

[Dienstag]

Nice article (and website, too!). By and large I agree with your interpretation of the spirit of the game, and I probably use those exploits along the same lines you do. It is truly sad to be penalized for _not_ doing such unsavory things as razing and starving captured cities.

That's the first I heard of the blitz-against-nearsighted-AI-threat-radius exploit. Is there significant evidence for this? I haven't noticed it myself but I'll keep it in mind. I know the AI has significant map visibility bonuses *coughcheatscough*, but I haven't seen a thorough analysis of how it determines/reacts to threats. I bet a watered-down version of how a human does this wouldn't be too hard to implement and would improve things.

[Sirian]

That's the first I heard of the blitz-against-nearsighted-AI-threat-radius exploit. Is there significant evidence for this?

Yes. Noticed the ability to blitz through multiple cities quite some time ago and set in to study it. The AI only keeps two defenders in each city, but will add a couple more to a "front line city" that comes under threat. If you move within 2 squares, the AI immediately does one round of drafting, if it can, and may rushbuild a unit, whipping or with cash, as applicable. (That it will then send these extra defenders out after bait is another issue).

I've conquered a large landmass standard map world in 22 turns, on Emperor without any ancient conquest, starting with 3 panzers, launching a golden age that brought me on average 5 new panzers a turn. This was the game in which I discovered and confirmed that the AI only detects threats within 2 squares of a city, as I had all kinds of chances to test. I made a point of blitzing as far as I could on one turn, capturing what I needed to reach farther into enemy territory to attack again on the same turn, razing everything else, starving what I did capture and using my obsolete units and/or infantry to garrison. Not one city lost to revolt during this period, but frankly, most got razed. I would also use new settlers to extend my reach, at times, but had to be careful not to cross the domination threshold.

I knowingly used the "hold back 3 spaces" blitz exploit a few times in the Apoly4 game, which sped my conquest a marginal amount, maybe three or four turns tops. (Check the tourney results when Mark posts them). It's not a game breaker by itself, but is indicative of larger problems with the AI in regard to defending itself.


- Sirian

[Willem]

Your page heading for the military tips is the same as the one for research.

[lockstep]

Great article! You should mail a link to Soren!

[KoenigMkII]

Sirian's Great Library
Well, that was great work. Well written, I spent 3-hrs (Graveyard-shift ones) reading all the text. The scrrenshots were all highly relevant and interesting.

I like to think one day I will get up to your skill-standard for starting games in such hostile conditions. Remind me not to cross you if a MP edition comes out :-)

I have to agree about cavalry being a real Blitzkrieg wonder weapon. But only Panzers get the 3 turn move rate for - the other countries 1st generation armour can actually be a lot slower in knocking out AI's than a good supply of elite cavalry.

The AI does need to upgrade its defensive units, especially if its used railways throughout the deep guts of its empire.

Captured/raised cities allow instant use of their communications, sometimes the very shape of the Civ-city raidius leaves a gap, along the diagonal, for cavalry to get at the next city, then the next etc.

Thanks for making the effort on the Fan site, its a must see for any civ player.

[TitanTim]

Your analysis of the game is commendable. Keep up the great work.

Here are some [nit-picky] notes about your web pages.
- Times New Roman is an ugly font and is difficult to read. Arial or Verdana is much nicer (not to mention your font is a little small).
- Colour your dates yellow so they stand out as individual points.
- Miscellaneous is spelled wrong on pointers.html.

[nato]

Cool site Sirian!

About your article, I really agree that razing is too powerful and not razing too risky. I don't want to be a razing monster ... this is one of my biggest problems with Civ3.

[Nakar Gabab]

One issue: The penalties for dishonorable trade dealing.

While I agree that it's out of whack for players, it's even more out of whack for AIs. I've NEVER seen an AI's diplomatic misgivings come back to haunt it. After 12 games of Germany backstabbing me, or Hammurabi landing mediocre forces to attack for no apparent reason, I'm beginning to wonder if having Polite allies makes a bit of difference. Worse, they'll STILL be able to sign MPPs, ROPs, and embargoes later in the game with other AI civs. I have no idea how they get around the "I'll give you a ROP... to hell!" but if there are gonna be penalties they should apply to the AI.

Of course I can see exploits therein, as well. Germany's sent 20 units into my land to get workers (oy vey) in the heart of the empire. So I call him up the turn before he's in striking range and sign a ROP, and a big trade deal. Next turn his hardwired little german-AI brain tells him to attack those evil workers at all costs, and the resultant diplomatic backlash leads to the destruction of his empire at the hands of the entire world.

Of course, that might make AIs smart enough not to do stupid, obvious encroachments like that...

[Sirian]

Nakar: good corollary in regard to diplomatic consequences for the AI's. They don't QUITE still do the old Civ1/Civ2 thing of continuously working as a team against the player, but there are still remnants of that. They certainly do not seem to suffer any shutdown of trading options from betraying agreements.

The AI's never, ever hold back on trading tech if someone can afford to pay the "market price" for it, in cash or goods. They do a lot of sneak attacks, too.

Yet I don't wonder if they don't need this crutch. Look how BADLY they invariably suffer from war in the industrial period. You can virtually ensure victory for yourself by getting them to fight one another. They clearly DO suffer the penalties of war weariness, which collapses their governments and sends them into communism, from which they do too much draft/whip and erode their population, as well as the inherent trade penalties. They might as well sign their own death warrants. So just imagine if they were penalized for their betrayals! The whole system might need another generation of programming and design work to stand up to that measurement.

It would be nice to have the AI's playing by the same rules, but that's just not feasible. They do need some advantages to give an experience player a challenge. I'm sure Firaxis would improve this, and apply penalties, if they could, but that may be asking for too much at this time. I'd be content just to see the diplomatic loopholes closed for the player, by way of more balanced and meaningful penalties for betrayals.


- Sirian

[Analyst Redux]

Sirian, an excellent effort at deconstruction—a few observations, if I may.

With respect to the happiness exploit, you mention the use of Shakespeare’s Theater in this regard. This is not a new idea/exploit in Civ III. It has existed, in some form, for as long as the series. In prior incarnations of the game (including SMAC), the use of a democratic government form meant that moving military units outside of cities would create 2 unhappy citizens per unit (units were assigned to cities in prior versions, of course) and air units (much more powerful in prior versions) created 2 unhappy citizens per unit, just for existing. In the old days, one would build Shakespeare’s Theater, then engage in what was referred to as “unit laundering”, the practice of reassigning air units to the ST city in order to rid one’s empire of their unhappiness effects. This would allow you do be democratic and bloodthirsty without dealing with the unhappiness component, at least to the extent you could launder units through ST. If you planned ahead, and build ST in a shield rich city, you could launder over a hundred units through it, which was more air units than even the most incompetent Civ general needed to plunder the four corners of the earth.

SMAC was even more exploitable in this regard, as one could build a mid-level base improvement called the “Punishment Sphere”, which forced all citizens in a base to be content, but cut that base’s research production in half. Any base with a PS, however, became a unit-laundering center. Thus, in SMAC, you had a limitless possibility to use the ST unit-laundering exploit, by being able to build a laundry in any (or even every) base. The research penalty in a few bases meant little compared to the upside of controlling military unhappiness as a democratic warmonger. You could even force build PS’s in freshly captured forward bases that produce little or any research in order to reassign/launder units on the fly and not give up the research penalty in older, established bases.

You’ve correctly discerned that the level of exploitation that can be achieved in Civ III, based on the intersection of the pop-rushing unhappiness and content citizen rules, is more like the high degree of unit-laundering exploitability associated with SMAC’s Punishment Sphere’s. Civ III allows you to “launder” the effects of pop-unhappiness through the ST wonder (just as always), but also, through the much-complained-of high corruption level, creates the effect of “disposable populations”—populations that aren’t really good for anything other than pop rushing or drafting, because more of them never produce more economic benefits. It’s as if you have an economically viable core, and the rest is just good for breeding soldiers, assuming that you provide enough bread and circuses to make the women forget their children who were taken from their hearths and homes to become cannon fodder—and you can slave people to death building their bread and circuses, too, since they don’t seem to notice. Pop-rushing otherwise useless frontier cities in not dissimilar from the SMAC strategy of using Punishment Sphere’s on the periphery of one’s faction, for essentially the same reason.

I agree with your reaction. Not what I consider attractive game play. But when lobbying/hoping for a change in a future patch, consider that (i) the exploit existed and was well known in SMAC’ (ii) the SMAC patches never addressed it; and (iii) the exploit has been repeated in largely similar form in Civ III. In other words, don’t hold your breath.

I also agree with your observation about the rules making it preferable to commit genocide on captured populations, rather than try to keep captured cities. I’m not sure who thought this was fun. I’ll go your observation one further, though, and point out that, the way the rules work, retaining a city for the wonder within has almost no benefit, therefore, no impact on the decision to raze. Wonders will not generate culture for the civilization that didn’t build them, and do not count in tallying the final points. In other words, outside of their actual effects, wonders make no contribution to victory at all. If the wonder is obsolete, or close to becoming so, its existence is of absolutely no consequence to you if you didn’t build it. If it doesn’t become obsolete, but its effect is weak, or doesn’t fit with your strategy, why keep it? In older versions of the game, I became accustomed, when playing Transcend or Deity levels, to seldom seeing early wonder movies, and having to capture wonders after the AI built them first. What I’m having a tougher time with is that wonders I don’t build have no value to me. When I go on the offensive, I become like the Taliban blowing up millennia-old statues because it’s less trouble than having them around. Again, not my idea of fun.

Your observation regarding the near-sighted AI combat horizon is keen. SMAC was the first incarnation of this program that introduced the programmers attempting to teach the AI to mass its forces in preparation for an attack. IIRC (and it was a while ago, so I may not), the programmers explained that it was pretty hard (actually, impossible) for the AI to evaluate the whole map to create a credible attack. The geometric expansion of calculations as you increase map sizes required use of a very narrow horizon for reasonable calculations. Therefore, they had the AI select a target or two (based on criteria they refused to divulge, but which weren’t terribly hard to spot, after repetition), and then had the AI isolate on the area of the map around the target to concentrate its forces. The AI defends, as you’ve discovered, on a similar principle.

Advanced SMACers discovered what you have—that this method of teaching the AI to mass-to-attack (or defend) can be neatly exploited with a blitzkrieg offense, bypassing the heavy pockets to get to the light ones. The breaking point in the AI logic seems to be the same in Civ III as in SMAC. It’s the point at which you can cut the AI into two or more distinct geographic pockets. The AI then appears to make its regional calculations based on an assumption that the separated pockets are entirely independent from each other. Counter-attacks will never be coordinated between the separate pockets, and will often be muted as the AI will, after being cut in half, put more effort into rebalancing its forces (based on the new calculations of what constitutes a “target” and where units ought to be massed as a result) than responding with a counter-attack. Or as we said on the SMAC boards, the AI runs around like a chicken with its head cut off while you mop up the blood.

This is a deficiency that you simply can’t hope to see change. As it stands, the game plugs along at a surprisingly slow pace, from a calculating the AI turn standpoint. I don’t see any possibility whatsoever that the AI view horizon can be expanded, given what that would cost in calculations. Were pretty much stuck with what we have in that respect, until the folks who write this program figure out how to tighten their code.

[Steve Clark]

bump. This is too valuable to be on the second page.

[Sirian]

In regard to the wonders, there are already enough benefits to the conquest route. I think it's rather better that the rewards for capturing a wonder are much less than for building it. Imagine if they gave any culture (much less full) that you'd get free border expansion and get rid of lots of flip pressures automatically. That wouldn't be any good.

Yeah, it is as lame as the Taliban, but the game engine pushes the player in that direction. Cities with wonders have just that much more cultural "memory" and are THE worst flip risks in the game. Whether or not I might benefit from the wonder(s), I would normally raze anyway, unless there was some reason to think I might wipe out that whole civ soon.


I had such overwhelming "Civ Weariness" from years of Civ1, that I did not much enjoy Civ2 at all (until the last two years, when I played it some more) and completely skipped SMAC and the others. Master of Orion (the original, didn't like the second one at all) freed me from Civ1 and I'm still looking for a conquest game as good as that one was. So Civ3 represents for me the "start" of such things as workers separate from city builders, and units supported by cash, not shields. I've read bits about SMAC, but still never even had a peek at it, so I actually did not know about the laundering exploit.

As for Civ2, I never had any problems being a warmongering democracy, just build Suffrage and some really large cities. How many tanks and howitzers did you need? I sure didn't need all that many, to a point of having problems supporting them. I'm sure there is plenty about that game I still don't know -- never did plug in to the internet community for it -- but for as little as I played it or read about it, I could cruise no-reload on Emperor with near 100% win ratio -- and not just space, but conquest. It was just such a time sink, I didn't play it a whole lot. Civ is still civ. Even Civ3 works on the same principles, although a few things have changed (mostly for the better).

The best news is that exploits can be ignored. They don't have to ruin the game for anybody not too tempted by them, if they don't want that. But it would still be nice if Firaxis did two things: A) Fix what they can. B) Tell us what they don't plan to fix, and why.

Forthrightness would go a long way to securing my loyalty for future business, yet it's something in short supply in the gaming industry. Gaming consumers are more savvy now, and more demanding than ever. Firaxis is about to release another patch, so I'll wait and see what they've done rather than speculate.


- Sirian