SMACX AI Growth mod

[Rocky]

Your solution of adding clean units definitely seems to have a positive effect on the AI. They seem (especially the aggressive/erratic factions) much more dangerous and determined.

- Factions like the Believers/Spartans/Hive are often the first to swarm the ocean with Clean Synth Foils. In my last game the Believers were bombarding my coast with half a dozen Foils, decimating my infrastructure. I like it. It makes them more of a threat that you can't just ignore.

- The AI loves the early Clean Police (Synth). It gives the bases some more defense and stability without slowing them down too much.

- The AI seems to pick its most productive base to produce a couple of Clean Formers early on, which works perfectly as it really does give them a boost.

- The aggressive factions (or the factions with the Conquer research path) seem to benefit the most. When they get their Lasers and Impact units out, their advantage in minerals saved by their earlier built clean units makes them potential monsters early on. I had the Believers overwhelming the Peacekeepers, the Hive decimating the University and the Spartans winning against the Believers in an earlier game.

- The more peaceful factions are no pushovers either though. Even a faction like the Morganites put up a fight. They were taking bases and units from me with Probes and combined with hit and run attacks from their Foils disrupted my invasion.

===

Some things I noticed:

- Some factions, when not having much starting land, don't seem to expand into the ocean. Do the AI's get a pre-defined Sea Colony Pod as well when they get Doctrine: Flexibility? Or wasn't that possible? I remember you were trying to do that with the Pirates once.

- I like the increased Prototype cost, but it feels like too much in its current state. It feels like it's slowing down the game, especially if you're up against an enemy who already has prototyped that weapon and is hammering you with it.

- The AI is performing much better in its current state, but is there a way to improve terraforming as well? Currently it still uses the old terraform style of Farms and Mines.

[bvanevery]

A lot of the Clean units seem to be helping.  Clean Formers is clearly a win.

I'm experimenting with Clean Scouts and it at least is not causing harm.  I find that when factions start in marginal conditions, like very little land surrounded by fungus and poor nutrients, they may make plain Scouts and exhaust their SUPPORT.  Since I started doing the Clean Scouts, I'm not sure if it's prevented the problem.  I saw 1 city that produced several Scouts and cost SUPPORT, but it had the minerals to do it and did not get choked out.  I am wondering if changing the AI Plan of the predefined unit will help.  For my next test I will try -1=AutoCompute.  Before I was doing 3=Explore, and that at least does not cause harm.

My own testing is problematic because I play the game too well.     I know exactly how to exploit everything, like a Clean Reactor at the beginning of the game.  So I've just been expanding in the early game like crazy and leaving the AIs in the dust.  Yes the Clean Reactor helps the AIs, but it also helps me.  This won't be true of everyone who tries my mod though.  And it could be, I'd kick the AI's ass anyways.  I'm just wondering, whatever happened to the days when I thought the Pirates and Morganites were scary?  Maybe I played on Enormous maps back then.  I've only done Huge for a long time now, and it doesn't seem that difficult to me.

- Some factions, when not having much starting land, don't seem to expand into the ocean. Do the AI's get a pre-defined Sea Colony Pod as well when they get Doctrine: Flexibility?

There's no predefined unit for it.  There could be, there just isn't right now.  It was Sea Former that I was having trouble with before, not Sea Colony Pod.  I have no idea whether encouraging sea expansion will help or not.  The AI in my experience often does a terrible job picking sea base locations, especially when it gets into a "3 square diagonal climb" in some corridor, none of them having any substantial resources.  Although, that's typically a late game behavior.  All I'm really saying is, this is an untested unknown.  What I can do, is try throwing a pod into the predefined units, and see what the AI does with it.

- I like the increased Prototype cost, but it feels like too much in its current state. It feels like it's slowing down the game, especially if you're up against an enemy who already has prototyped that weapon and is hammering you with it.

The tricks you want are:

• Blow an Artifact on it.
• Set up a prototype in a border city, pop a pod somewhere, so that that city is closest.  Good chance the prototype will complete.  For bonus points, make your prototype unit as expensive and exotic as possible.  Like put totally unnecessary armor on it and so forth.
• In the early game, you can just wait for techs to give you Synth or Speeder units predefined.  The general drill is predefined units that use a new weapon, armor, or chassis, come 1 tech later than when those capabilities appear.  This is so predefined units don't just give prototypes away.
• C4 Bioadaptive Resonance and C4 Single-Sided Surfaces are exceptions to this rule.  You get the prototypes immediately.  This is because they're only providing Strength 3 weapons and armors on Tech Tier 4.  Used to be that C3 Advanced Subatomic Theory was a prereq for both.  It still is for Surfaces, but Bioadap is now its own independent branch of the tech tree.  The idea is you've already struggled to get Plasma armor prototyped and that makes both 3-Pulse and 3-Res available.
• Once you get C3 Applied Physics, build a Skunkworks.

I could make the Skunkworks easier to get.  Just realized it's also a Build tech, as it saves on minerals costs and allows prototypes to be shifted around without minerals penalties.  C3 Neural Grafting might be a more reasonable fit.  I find the latter is easier to obtain than C3 Applied Physics, probably because it's a Conquer / Build / Explore tech instead of just Conquer.  Multiple categories generally means more likely to be discovered.  Think I'll also make the tech path to C3 Neural Grafting more clearly in the Build tree while I'm at it.

- The AI is performing much better in its current state, but is there a way to improve terraforming as well? Currently it still uses the old terraform style of Farms and Mines.

There are no settings to change the terraforming behavior with .txt modding.

I could remove this penalty:

Code: [Select]

-1,      ; Nutrient effect in mine square (0 or -1)Whether that's a good idea or not, I don't know.  I suppose it would stop me having to scrap all the mines on top of farms, when I capture AI territory.     I suppose I'll try it as a test.  The stock AI is stupid and that's not gonna change, so this may be justified.

[bvanevery]

Your comments triggered a fair amount of changes.  Among them, Lasers will go back to attack 2 and will come earlier.  I think the early game is too slow getting into a shooting war.

Details AI sea colonizations MY 2163

AI sea colonizations MY 2163

Sea Colony Pods is not one of them though.  I haven't done anything with that yet.  I note that 2 very cramped factions are in fact producing Sea Colony Pods.  The Spartans even managed to found 1 sea base already.  Now it could be argued that they're not coming early enough, but to put it in perspective, I haven't managed to make my 1st sea base either.  And I'm the Unsurpassed faction of the game.  I started with amazingly fertile land, so no pressure for me to do it, I'm just filling out stuff around my capitol.  I have 1 coastal city working on a Sea Colony Pod and it's been taking a long time.

They'd have to be cheaper to get it done any faster.  I could only do that as a special case predefined unit, as I'm not inclined to mess with Foil chassis or Colony Pod Module costs.  The former would change all the early game ship costs, and the latter would change cost of land colonies.  I'm not thrilled about handling it as a special case, because if someone wants to design a different kind of sea colony pod, the cheapening wouldn't be in effect.  I could get over it if I really thought it was worth it, but it'll take substantial testing before I believe that getting the AI to make a lot of sea bases is actually a good idea.  I think it's likely to make the land factions rather weak, and the Pirates don't need the help.  In fact, it would inevitably make the Pirates rather monstrous, if they're getting Sea Colony Pods even cheaper than they already do.

MY 2165.  I've noticed that the Spartans are settling a new sea base, and the Hive seemingly already had a sea base I didn't notice before.  So, I don't think the AI making Sea Colony Pods is a problem.  Only issue is the rate at which they do it.  Which is inevitably compounded with the Pirate problem.

Much later... it should be noted that Sea Bases are extremely easy targets for enemy Foil Probe Teams.  I lost the one I made almost immediately.  I stole from it, but the effort of actually retaking the base was not worth it compared to other things I could be doing.  Granted, I forgot I could use Marines to retake the base.  Quite an oversight since I designed Marine Probe Teams to steal from it.

[vonbach]

The clean formers are nice and the AI seems to love them. Sea colony pods coming earlier would be a nice idea  just to help out the AI.
I have noticed the AI does not like to build its own units. So units like sea pods come later.
The problem with the increased prototype costs isn't that it hurts me I just plow though the extra cost or make a skunkworks.
The AI seems to be hurt more than me.
All in all the changes seem interesting and good.

[Rocky]

I think the cost of the Sea Colony Pods is fine, don't worry about changing or experimenting with it. My issue was that sometimes the AI doesn't design them, although the chance of that happening is low. In my latest game the AI also produced Sea Colony Pods early when they ran out of land. But once every few games there seems to be a bug or randomness where the AI doesn't design a Sea Colony Pod when it discovers Flexibility. Usually it designs a Sea Pod eventually when it reaches Fusion Power but it can limit the AI at times. With a pre-defined Pod you could prevent this bug if/when it occurs.

Thank you for the tips about prototypes. I had no idea it worked like that with the pods. I really like the concept of the pre-defined units. It adds flavor to the game and gives a feeling of satisfaction when you discover a tech that makes those units available.

I've been playing the latest game with mines disabled. So far it doesn't seem to hinder the AI. They actually seem to terraform faster and plant more forests. (But that's probably because it spends less turns on mines) I'm curious to see how this will go in the mid/late-game.

In my current game the Believers are ruling everything. I'm not a good player but (playing as Morganites) I expanded as much as I could with currently 16 bases around 2220, but the Believers are three times above everyone in power and they don't have a Monsoon Jungle. Luckily, they aren't really hostile to me (more like indifferent) and seem to have set their sights on the Usurpers. The Caretakers don't like me because of my Capitalist economy, and are coming pretty close with their Resonance Laser units and Foils, hopefully I can rush my infrastructure in place before they attack.

[bvanevery]

I guess there's 2 stages to this Sea Colony Pod thing:

• Pinpoint the problem.  I think we're talking about events within the first 40 turns of the game, but I've not observed that timeframe very much.  I'm usually waiting for longer results than that, whether playing myself or watching AI games.
• Let's say we pinpoint that under conditions X, the AI should make Sea Colony Pods, but doesn't.  Then we have to see if any kind of predefined Sea Colony Pod unit will change that behavior.

I do see AI factions get "8-balled" behind other factions and have little to no land to grow into.  I would like to think the answer to this is extreme violence, but the AI doesn't know how to step up to the situation.

I've also seen an AI faction making 3 bases on land and then just quitting their expansion.  Usually because they're surrounded by a death maze of fungus.  Plenty of land, but not very good land, and a lot of twisty turny passages to get to anything.  It wouldn't shock me if the AI path finding code can't reason very well about it and just chokes.

So we've got 2 kinds of barriers: other factions and fungus.  The perceived answer is to settle the oceans.  I wonder if that's really the answer though?

I think I'll start by observing AI vs. AI game runs, all ending uniformly at MY 2140.

Details runted Cult MY 2140

runted Cult MY 2140

This game deliberately included the Pirates.  Nobody had a land availability problem, everyone had basically tons of reasonably good land.  Most factions have settled at least 4 cities by now.  The Cult is on a large island or a small continent, i.e. "Australia".  They're just morons.

Details Caretakers delaying settlement MY 2140

Caretakers delaying settlement MY 2140

In another game the Caretakers have land, but they perceive themselves to be threatened by other factions around them, plus a Spore Launcher.  Their Colony Pods don't settle.  Or they're trying to be greedy and grab too much.

Details Morganites too spread out MY 2140

Morganites too spread out MY 2140

The Morganites don't have any more Colony Pods in production.  They're an Explore, Build faction.  I'm thinking of changing them to Build, Conquer, now that they've got +1 SUPPORT.  A bit more Haliburton.

Details Morganic explorational doom MY 2200

Morganic explorational doom MY 2200

That game had a lot of factions next to each other, so I ran it for awhile to see who gets stomped.  The Morganites are obviously going to die.  They have 4 cities but their "empire" is split in half by the Caretakers and the Spartans are swamping them.  I think that settles it as far as switching to Build, Conquer.  The Caretakers are doing ok but not as well as some other regional powers.

Details Consciousness knocked unconscious MY 2200

Consciousness knocked unconscious MY 2200

The new Conquer only Hive is a terror!  They wiped out a number of decent Cyborg cities.  The latter are an Explore, Discover faction and I've been thinking about switching them to Discover, Conquer.  Think this shows it's time.

[bvanevery]

New series of screenshots with the changes to Morganic and Cyborg factions already made.  Note that so far, I haven't seen 1 case where building a Sea Colony Pod was even relevant.  Granted, I've been watching games that deliberately have the Pirates in them, so only 6 factions are scrounging for the land.  Eventually I will try it with all factions random.

Details Cult of morons MY 2140

Cult of morons MY 2140

What would I even do about this?  They have an excess of production vs. food.  They could have produced units to defend this city, but they didn't.  If I put armor on the Artifacts, would they survive attacks?  Probably they'd just be like Probe Teams and gratuitously die when something else dies.  But a lone city, maybe they'd fend stuff off.

They've also only built 3 cities despite abundant land.  They've got no active Colony Pods, but 2 in production.

Details incompetent academics MY 2140

incompetent academics MY 2140

The University is an Explore, Discover faction.  Now they're the only one.   They've got plenty of land, they're just not getting the settlement job done.  They seem to have been distracted by the discovery of Progenitor Psych.  The Caretakers are distant neighbors to the west, but they're not in contact.  The University has researched PrPsych on its own.  I've moved Hypnotic Trance to that tech.  They've made a few too many Trance Scouts, although the problem is not bad.  If Trance Scouts turns out to be an AI obsession, I could move it somewhere else. 

Details obsessive Probe Teams and Foil Probe Teams

obsessive Probe Teams and Foil Probe Teams

It doesn't end well for the University.  They never expand.  The Caretakers hassle them with a Battle Ogre, mindworms, and spore launchers the whole time.  The University is a Socialist Police State and builds Probe Teams obsessively.  They march in front of the Battle Ogre and are promptly killed.  They also try to build some Foil Probe Teams when they should be tending to other needs.  They actually had Recon Rovers in time to do something about their situation, but they don't use them effectively.

Details dregs of Drones

dregs of Drones

In other news, it looks like the Free Drones got bottled up and squashed by the Cyborgs and the Believers.  I'm surprised at that, because recently I set the Free Drones to Build, Conquer and they became pretty terrifying.  The Cyborgs did the actual conquests, indicating that their new Discover, Conquer focus actually works.  I think the Believers merely blocked routes of expansion.  Now the Believers and the Cyborgs are in a shooting war, so I'm interested to see who comes out on top.  They're the 2 most powerful factions on the graph.

Eventually in MY 2299, the Believers have been slowly wearing down the Cyborgs.  They seem to outproduce them, the tech is about equal, Probe Teams are not much of a factor for either, and the Believer empire is more centralized and radial.  The Cult of Planet is doing better than everyone, as is made it to the Monsoon Jungle and took off.

The Cult is currently an Explore, Build faction because I was worried about them getting Extremist.  However that's the only Build tech they actually need, so I'm going to try them as an Explore only faction, like the Gaians.

[bvanevery]

I'm now shifting my testing focus to the University, which seems to be the most pathetic of factions.

That said, they did make it to the endgame of my last test game, because they started on a large island and had no land competition.  The game came down to a 4-power stalemate, between the Cyborgs, the Spartans, the University, and the Drones.  Technically the Data Angels were still in the game but I think they got wiped and then escaped.  The Cyborgs finished the tech tree but did not have the productivity to finish the Voice of Planet before MY 2500.  I quit watching at MY 2480 as it was boring.  The whole thing was dominated by mindworm combat, despite the availability of Singularity reactors and other endgame goodies.

Details double pathetic

double pathetic

In a new game, we finally see the kind of circumstance where someone might want Sea Colony Pods, although a Transport and regular Colony Pods would actually be better.  This is a curly arm of a twisty, globe spanning supercontinent.  Someone's gonna get crushed!  I just changed the Gaians to Explore, Discover as thematically they've been that way for awhile.  I've also changed the Morganites to Build, Conquer as they tend to have stiff opposition on land.  The Pirates have been a pure Build faction since version 1.33, as they have a huge moat.

Details Morganic sea luck MY 2114

Morganic sea luck MY 2114

The Gaians managed to make a bit of a land claim before making peaceful contact with the Morganites.  The latter have been gifted a Unity Transport.

Details Gaians seaworthy MY 2118

Gaians seaworthy MY 2118

So here we have a reasonable lower bound on how soon a land faction could become seaworthy.  Notice also that there hasn't even been time to get a 3rd land city settled.  I think it's safe to say that making Sea Colony Pods in the first 20 turns of the game is a non-issue / would be a non-productive strategy.

Details no Sea Colony Pod designed MY 2118

no Sea Colony Pod designed MY 2118

That said, they don't even have the option.  Is it merely a 1 tech delay until the Unit Designer runs and makes the unit?  Or is it an ongoing bug?  We shall see.  Right now the Gaians have 5 Tier 1 techs.

Details Gaian next tech MY 2128

Gaian next tech MY 2128

The Gaians do not gain a Sea Colony Pod with this tech.  They were making a Recreation Commons in 1 of their cities, which I regard as a big mistake this early in the game.  I could move the Recreation Commons to a later tech, although I'd need to figure out what to do with Social Psych.  Upon learning Biogenetics, the Gaians switched production to a Clean Synth Speeder and a Synth Police Wagon.  They have a 3rd Colony Pod at the Morganic border but have not settled yet.

Details Hive land bridge MY 2128

Hive land bridge MY 2128

Unbeknownst to the Gaians, the Hive has triggered an earthquake and bridged the land just to the south.  They have not met yet.  I suspect that earthquake probability has to do with settlement and land constriction, that it's not completely random.  It may be a partial safety valve for bad map layouts.  It's probably not going to help the Gaians a whole lot though.  Introducing a new enemy may even be a big negative.

Meanwhile, the Morganites have 2 Colony Pods milling around fairly unproductively.  They should just settle somewhere.  Well I guess actually exercising brains for this sort of thing, is Thinker Mod's territory.  I can't do anything about that.

[Rocky]

I agree with you on making Lasers available earlier. Although I do like weapons later in the tech tree, it sometimes does create stalemates between factions early on.

I agree with you about adding Conquer to certain factions as a defense against more aggressive factions. With the tech tree being specialized the Conquer path is more powerful.

I've been testing out the game with Mines disabled in the alphax.txt. It doesn't seem to hinder the AI and it actually seems to improve on it a little. Without mines, the AI is terraforming faster and adding more farms/collectors and early forests.

[bvanevery]

All of my various faction focus choices, like whether to do Conquer or not, may have been valid when I made those choices.  But this mod has been developed for a long time, and the tech tree has gradually changed in all kinds of ways.  It seems to me pretty clear that Conquer now, with Lasers going back to being early and puny, is better than Conquer a long time ago.  And conquest as a strategy doesn't have the same balance and implications as in the past, because armor and Sensor Arrays got a lot stronger than back in the dark ages.  Sure you can try to rush with Recon Rovers, but I seriously doubt that AIs are going to be competent enough to take out other AIs that way.  I mean, Synthmetal is available as a Tier 1 tech nowadays.

One thing that troubles me though, is factions often don't build Command Centers when they should.  They're already a Tier 1 tech so I can't offer them any earlier.  I could cheapen them.  On the other hand, I don't want the AI to obsess about making them too early.  It was doing that with Perimeter Defenses, to the exclusion of colonization, so that's why it got moved to a Tier 3 tech.

I can't really see the aesthetics of Mines going away, pleasing me at all.  I like my big "7 minerals" mines on a Minerals bonus square.  But kudos to you for experimenting.  I did turn off the mine on farmland penalty, although I haven't paid attention to the results.

Details Gaians too happy MY 2133

Gaians too happy MY 2133

The Gaians have gone back to building a Recreation Commons.  This is not a smart course of action for them.  They already built a Clean Police Wagon, so they can go to size 3 without problem.  They still need to colonize.  They don't have enough Clean Formers to make aggressive use of a larger population anyways.  They haven't even gone Police State and they could, which would allow 3 police units.  Here they are just sucking.  Now that the use of Non-Lethal Methods is early game and mainstream for all factions, even Democratic factions, I really think I should put Recreation Commons later.  It's a waste of time.

I've noticed that the AI also does the same sort of wasteful thing with Hologram Theaters, if it gets the chance.  This is perversely a problem of the University.  It's actually a Tier 3 tech, you have to discover Optical Computers.  Most other factions don't discover the tech because it's in the Discover part of the tree.  I might have to delay Hologram Theater until Tier 4.  It really shouldn't be needed in the early game at all.  I've seen AI factions completely fail to research it for long periods of time, and they don't seem to be negatively affected.  And as a human player, I certainly don't need it, when police are always available and useful.

MY 2136.  The Gaians sign a Treaty with the Hive, and go Democratic.  They're building a Children's Creche.  With only 3 cities, their development is arguably too vertical.  On the positive side, the Clean Formers and Clean Synth Speeders have done their job of preventing loss of SUPPORT.

I could do the same with Colony Pods, if I perform some heroics of reassigning costs.  Just make supporting Colonists no longer part of the game.  I'd have to make the Colony Module less expensive, so that when combined with a Clean Reactor, the cost comes out the same as current Colony Pods.  This could be abused by a human player though, who could design a Colony unit without a Clean Reactor.  That might be a price worth paying though.  Strictly speaking, I could disable the Colony Module so that designing units with it is impossible, and then only provide predefined Colony units.  However I'd rather not.

Hm, there's another way.  I could leave the module costs alone, and simply outright assign a cost to a predefined Clean Colony Pod unit.  Yes you could design something else, but cost-wise you wouldn't want to.  Sounds like a plan.

MY 2140.  The Gaians capture their 1st mindworm.  That's pathologically late for them, and a direct consequence of having no land to maneuver in.  They have a 4th Colonist milling around now, and another in production.  They still don't know how to make Sea Colony Pod units.

MY 2145.  The Gaians should have settled their 4th city, but didn't.  They were right on a correct spot on the Morganic border.  They still have Treaties with both the Morganites and the Hive.  Next year they just walked away.   They're just abysmally stupid.  I've seen this non-settlement behavior quite a number of times.  Generally speaking it's a benefit to me, the human player, when the AI exhibits this paralysis of encroachment.

[Rocky]

I agree, removing Mines is not an ideal solution and you miss out on stuff like bonus minerals. On the other hand, the AI generally doesn't seem to do a good job harvesting minerals from rocky squares anyway. The nutrient bonus for mines you proposed is a good trade-off. A human player would only build mines in specific areas so it's more of a benefit to the AI.

The terraforming without mines reminds me a bit of the Kyrub AI. (Except it only places mines on rocky squares) It grows forests early so they can spread and then follows it up with Tree Farms/Hybrid Forests.

Is it possible to "tell" an AI to emphasize building certain facilities like Tree Farms? Or is that something you can't directly control?

[bvanevery]

Is it possible to "tell" an AI to emphasize building certain facilities like Tree Farms?

Only in the sense of cheapening the cost of the facility, or making it available earlier.  I'm definitely not going to make it available earlier, as it's even more valuable for a human player.  For similar reasons, I don't think the cost should be lowered.  Having factions focus on Build rather than other options, would tend to get them to B4 Environmental Economics sooner.  I do have a fair number of Build, Conquer factions nowadays, so they might get there on their own.

I think it's important to remember, that Clean Reactors is now my primary strategy for increasing AI output.  Giving away everything is not a good idea.  Aside from making "unintended" cakewalks for human players, some people do combine my mod with Thinker Mod.  And that AI is evil.  You give that thing an inch, it will take a mile!  So no, no free lunches for the AIs.  Improvements always have to be kept on a tight leash.  That's why it's a game design problem, and not just a "give everything to everyone" problem.

MY 2174.  The Gaians learn E2 Centauri Empathy and go Green.  They have 2 Colony Pods that have been milling around forever.  Valid city sites are available, the AI is just not taking them.  It sucks!

I think this experiment is pretty much concluded from an "identifying the problem" standpoint.  The question is whether Sea Colony Pods is actually a solution.  The AI does know how to make Sea Colony Pods now, but I don't know when they became available.  They're not choosing to make them, despite having 3 cities.  They're working on a Command Center, a Network Node, and a Colony Pod that they clearly don't need.

It wouldn't be completely unreasonable to make Sea Colony Pods cost the same as land Colony Pods.  Yes they contain the equivalent of a Recycling Tank, but the ocean has no minerals.  I will consider experimenting with that, but first I need to change when the Recreation Commons and the Hologram Theater become available.

Details capital stupidity

capital stupidity

Oh, the punchline: the Morganites didn't do any better.  4 idle Colonists, despite piles of land they could settle!  What a joke.  Or a tragedy.  I think I'll cry now.

If unarmored civilian Clean Transports were predefined and cost less, I wonder if the AI would use them to bridge land gaps?  Something to experiment with, but first I'll be making all those other changes.  It's also risky to have the AI obsess about such things.

[vonbach]

It wouldn't be completely unreasonable to make Sea Colony Pods cost the same as land Colony Pods.  Yes they contain the equivalent of a Recycling Tank, but the ocean has no minerals.  I will consider experimenting with that, but first I need to change when the Recreation Commons and the Hologram Theater become available.

It would make the AI's life much easier.

[bvanevery]

Upending Recreation Commons has really turned into a lot of work.  Still in progress.

[vonbach]

I tried playing as an aquatic faction. I had to give up in frustration twice. The AI just threw Isles of the deep at me every time I moved until I couldn't take it anymore.
The AI sets certain things in a much higher priority and it also depends on what AI. Some AI will spread like a weed. Others will build more.
From what I've seen the AI will start colonizing (depending on the AI) when it has certain things built. Like Recreation commons that and things like  a Children's Creche or a Recycling Tanks usually have top priority.