Imbalanced AI?

[ComJak]

So I've been running a few tests over the last week or so with AI on different maps and my empire on full automation. All settings are identical in terms of empire settings. When I let these games run for a few hours on 4x speed, I notice that very quickly, about 1/4th of the empires randomly produced will expand a LOT faster than the others. There are times where my empire has 2-3 planets (once again, full automation)and others have 1-2 colonies but other empires have 5-10 already. These empires tend to colonize the rest of the map far before anyone else and dominate the game. I have had games end hundreds of years later with empires with only 1 colony the entire way through.

What causes this huge difference in performance? I haven't tried empires of the same race so my first guess would be race benefits. But these differences seem to be far too powerful if this were the case.

Thanks for any explanation.

ComJak

[Kayoz]

I think you're right about the racial advantages accounting for initial expansion rates - but while this is a factor, I think the biggest single factor is LUCK.

Continental planets, for example, are the highest quality but the rarest - so you're more likely to see slow initial expansion rates from Humans or such races. Though if they are lucky enough to have continental planets close to their homeworlds, they'll expand well due to the speed with which they become profitable compared to lower quality planets.

Also, the home system generation seems to be an issue. I've seen some games where (normal setting), one starts with 2 or even 3 decent planets in one's home system. That can be a rather nice advantage in early games, as it's easy to protect one's initial holdings with a single fleet.

Yes, it's imbalanced. But who said life is fair?

[Max 86]

Thats a tough question to nail down exactly. On the one hand why would it be any different. The only relative experience we can draw on, philosophically speaking, is a quick examination of the nations here on earth. The answer to your question is the same as asking why didn't Belgium or Sudan or Vietnam rise up to match the size and strength of say China, U.S. or Russia? The answer to that is quite complex.

Game wise, also like real life, it comes down to things like starting position and proximity / availability to resources, cultural characteristics in conjunction with government type, strategic decisions regarding science commitment and direction. Sometimes leaders just make dumb decisions that severely impact their nations ability to expand (Mussolini) and some times they pull it off at great cost (Peter the Great).

If you ran the test and after 200 yrs or so each faction had the same number of colonies / size of empire, I would consider the game broken and unbalanced in a 'too balanced' kind of way. Interesting test though. It says a lot about the game.

[ComJak]

You guys make good points. I agree with the thought that if all factions had the same colonies, it wouldn't really be a fun game. However, it seems to me that the continental natives are the ones that do the best in my games. I may also have to tweak the galaxy settings for a more balanced start.

I wonder if the AI is capable of processing another faction as a "threat" in a way where if one faction gets too large too quickly, the other smaller factions will band together to fight against the large faction and work together to prevent a victory. This seems to be one issue of the luck based expansion. Usually, in my games, 1 faction will occupy about 40-50% of the galaxy (on med sized maps) and achieve victory while the others just passively allow it to grow and win.

ComJak

[Tehlongone]

I haven't made any proper tests but from observation it seems high growth empires do well if they are intelligent (i.e. not Atuuk/Gizurean) but the main factor seems to be location and local resources.

In any given game some civilizations just get better starts than the others and as they take advantage of this fact their power spirals ahead of the pack. Unlike some strategy games starting conditions vary pretty wildly depending on which planets you get easy access to and how crappy your neighbors are.

The craziest results I've had were from a pre-hyperspace start where one race got off it's system well ahead of the rest and controlled 1/4 of a 1400 star galaxy when the others began expanding. He had 3 decent (and colonizable) planets in his home system.

[BigWolfChris]

Best way to truly test this
Run it a dozen or more times, make a note of each race that does well
If there happens to be races that are alot more often in better positions than other races, then it's possible their racial traits might need adjusting
(In reverse, if a race is often at the bottom, their traits might need a slight increase)
If there appears to be a nice balance of which races are more successful, then we can put it down to random map generation

Personally, I'd be happy if it is really down to generator, since it means saves should often have different races as the players main competition

[Tehlongone]

I don't know what the point'd be, it's not like the races are particularly balanced. I don't think they are supposed to be. I mean, does Atuuk vs Quameno really seem balanced?

There is a pool of "good-enough" races that often dominate and some races that almost never do well. I don't remember the last time I saw a successful Atuuk/Gizurean and there a several other races that only do well when they get really lucky. Some of the races I view as best sometimes do very poorly if they are unlucky though. Although Securans are almost always big.

I view it as more of a feature than a problem, realistically for a bunch of random alien races to end up equal seems improbable. I guess they could be modded closer if one wanted it.

[ComJak]

I guess I'm thinking more old-fashioned RTS where all factions are carefully balanced. In reality, the differences between the races are really a great unique feature. I will do more testing to see what kind of correlation there is to the factions. Perhaps some factions do need a nerf but since this is hardly the competitive game, it wouldn't even be that crucial. Playing weaker factions could even provide a greater challenge than simply easy-medium-hard.

ComJak

[Cauldyth]

Yeah, I don't think the game even pretends to make the races balanced. Instead of comparing it to Starcraft, compare it to Europa Universalis. You can play as France or you can play as a one province minor in Ireland. They're very different experiences, and how well you aim to do in a given game is relative to how difficult you've made things for yourself by your race/nation choice.

[buglepong]

Luck. Independent colonies really skew early development. Try putting independent colonies to rare setting.

Also lucky effects of exploration. In my current game the ackdarians are twice as big as any other realm and that is due to them following the way of darkness

[Haree78]

Bare in mind that I have spent hundreds of hours play testing the AI and seeing how AI empires do in development of the extended mod.
There are certain races that are better than others in the vanilla game without a doubt. However as people have said there are a lot of major factors that will affect even the best performing races. Who their neighbours are, if they get stuck in long wars of attrition, what resources they have shortages of and how many good colonies and independents are available to them. The biggest factor is probably neighbours and independents.
However some good starters aren't very good end game. For example the Gizureans, like my race the Keskudons are fantastic starters with the right conditions but they need to have a REALLY good start to not lose end game because they make way too many enemies.
Others if they get a good start are complete beasts, like the Quamenos.
I always consider races like the Shandar as minor races because I have never seen them become a big force.

[Nanaki]

Europa Universalis is a really poor role model if your looking for reasonable balance...

As for Distant Worlds, I hardly mind imbalance, but I have seen some fairly extreme cases in my last vanilla prewarp playthroughs. Quameno with 100+ colonies, Kaidan with 90+ colonies, then humans and boskara with 20 colonies, another 4-5 races with 5-10 colonies, and then the rest of the races (about 10 of them) having only one or two colonies. Kaidan was due to a bug where an independant instantly became an empire and started out with gainax warpdrives before any of the other races even had basic warp cores.

I have been working on a mod that balances out the races some. It is not designed to remove all imbalance but rather slightly buff the weaker races while slightly nerfing the strongest races. The only issue I have at the moment is that most racial mods do not work with pirates which only gives a limited selection of viable races for pirates to pick.

[Kayoz]

I think one thing to keep in mind is that pre-warp start is NOT what DW was initially balanced for. Racial balance was never changed with the inclusion of a pre-warp start, and I rather expect racial advantages, such as growth rate and economic advantages are exacerbated in early starts.

[Modest]

Wow... That is interesting I also feel that certian races are more powerfull than others, and I like it. I would say that most of time we can speack about major empires and minor once. But I do agree that most important factor seems to be pure luck. Actually it is eoung that one race will get lucky crash bonus on colonisation technology early in game and there You go - they will grow expotentially.

ComJak - why don't You, in Your next test help one race? Make it more "lucky" via editor? Let's say, pick one of races that never did good and give them ultra rare resource on their home planet in addition to normal resources that they have? But why stop on that? Pick up another race that most of time did good job on expansion and make sure they will not have source of calson in starting system? I am pretty sure that those changes will definietly change how both races will develop in future.

[Bingeling]

I think "no source of caslon" never happens. Certain resources are scripted to appear in the pre-warp starting system. Of course, if one manually remove them...

As for races, of course there are differences between them, but random settings influence them greatly.

[Nanaki]

I would say this is only somewhat true. The only advantage that becomes substantially stronger in pre-warp is the research bonus, which enables the Quameno to get access to warp drives and colonization tech before any of the other races. I feel the issue is more that Technocracy is way too powerful than anything else. +50% Research bonus without anything to counter-balance it is too much of a good deal. But at the same time Technocracy is the only reason Zenox is not a crap race to play, which means any potential nerf of Technocracy means that Zenox would likely need buffs to counteract that.

[Tehlongone]

Pop growth is pretty important as you start with a few billions in a pre-start, if your growth rate is too low to reach max population then you'll be at a disadvantage compared to some of the others.

[Tormodino]

I've also noticed that the most dynamic games, read as fewest steamrollers empires, take place when not playing pre-warp. Pre-warp is a nice starting point for the player, but the AI seems to struggle with to get a steady stream of resources since it cannot properly account for scarcity and save for a rainy day like the player can.

Overall, the luck of the draw in terms of planets and resources, as well as the disparity in basic race power, is something I really like about DW.

The fact that balancing is not a huge consideration is good. I would like to see the AI expand a little more reliable though. A slight limiting of the extremes, if you will. Not sure how this could be solved, but I think substantially increasing the running cost of newly founded colonies could help with this. This is based on the fact that the AI does consider cost when making descisions, so it would seem the least heavy handed way of slowing runaway expansionism.

EU4 is actually a pretty good example of a way to balance expansion. Many don't like the model, and some countries have modifiers that allow them to go crazy with it, but the basic costs of colonization (gold and troops to guard against natives, as well as military commanders allowing expansion in the first place), military expansion (monarch points for integrating provinces and countries) and increased revolt risk for changing relgigion is so high that especially in the early game it is nearly ruinous for a small country to attempt massive expansion projects.
It might not be the best, but it does work. Except for France... I mean... Seriously

[Cauldyth]

Yeah, pre-warp starts seem to result in wider discrepancies in AI empires. Those starts require two very expensive techs before you can start expanding (Colonization and Warp Bubble Precursors, though you could argue that proper hyperdrives are necessary for proper expansion, bringing it to 3 very expensive techs).

I play on Expensive tech, which makes these 3 techs take even longer. If you get lucky/unlucky with tech breakthroughs/wrong-paths on even one of those projects, that can have a massive impact on your prospects relative to other empires. In my most recent game, I had a wrong-path on Warp Bubble Precursors when it was in the 80+% range. Ouch.

[Nanaki]

I think a lot of the issues with prewarp is that the AI is not really that good at prioritizing early game techs. It needs to priortize colonization and gainax hyperdrives ASAP, but it occasionally faffs around with unnecessary techs which delay it.