The Worm Farmer's Gambit

[Iainuki]

I have recently found myself laid up and with only an ancient underpowered Linux laptop for company. While looking for something to do, I found T-Hawk's SMAC thread here and recalled that SMAC runs on Linux courtesy of the sadly-defunct Loki Games. Thus, I dug out the game, found a neatly-packaged form of the ancient glibc libraries required to run it, and after some adventures with ffmpeg decided to explore some variants. Like T-Hawk, I've spent my fair share of time folding, spindling, and mutilating the game, but our approaches are definitely a bit different, and so I'm kind of hoping that someone will find the contrast between our styles interesting.

I was curious if my memory of having an adequate army from only captured mind worms was at all accurate, and thus was born this variant. The variant rules:

* No normal military units with better than 1/1 attack/defense. I'm allowing psi attack and defense but they come so late in the tech tree I don't expect them to matter much. All noncombat units are allowed but must have no better than 1 armor (or psi defense).

* All native life forms can be built as well as captured.

* No founding bases within 3 squares, or 2 diagonally, of another base.

* No asking AIs for tech trades.

I ran a couple of test games before the game I'm reporting, to shake off the rust and to make sure that the variant would be playable. I quickly established in the test runs that banning all combat units (even 1/1 garrisons and police) was not practical, and that allowing 1/1 units to try to get the worm snowball going works better. In one test game I only ever saw a single capture, so allowing building mind worms provides an escape valve so I don't have to start over if things go really poorly. The limit on base spacing isn't to make the game hard (nothing could do that, short of rewriting the AI to not suck), it's to limit the amount of micro I have to do. While formers probably account for a majority of the micro in SMAC, there's still a lot of base micro to do: for instance, SMAC doesn't warn you if a base is about to grow and go into drone riots, so if you want to avoid losing turns of production for your bases, you have to periodically go through and look for bases that are about to grow and micro them to avoid riots. I also don't ever want to have to mind-staple 180 bases repeatedly. Meanwhile, more bases can also work more tiles requiring more formers to improve them, which also adds to the micro. Taking maximal ICS off the table to limit the micro will make things more fun for me. Finally, making it so I'm not allowed to trade techs takes off the table one of the standard strategies for judoing AIs on high difficulty in Civ-like games, using diplomacy to turn their production advantages against them. Again, it won't make the game hard, but it will make the early game research choices more interesting.

The Gaians are an interesting faction, and arguably (IMO) the second-strongest faction in the game after the University. They get two social engineering advantages, +1 planet and +2 efficiency, and two disadvantages, -1 morale and police. This variant is built around of the first of those. +1 planet means they capture a mind worm the first time they attack one and otherwise have a 25% chance of capturing mind worms. (I have read that mind worms have a reduced chance to capture other mind worms, which I could believe, but I don't know if this is true.) It also gives a +10% bonus to psi combat, which there will be a lot of in this variant. IMO, efficiency is the strongest social engineering bonus in SMAC except for the special effects you get from +6 growth and +2 economy. It reduces the number of drones you get from founding lots of bases when the low happy cap is one of the biggest limits on production in the early game, lets you speed up research by unbalancing the sliders without crippling energy losses, and increases the energy per base by limiting the losses to inefficiency. Meanwhile, the only thing -1 police does is prevent you from using nerve-stapling, which you don't really want to do anyways until you repeal the UN charter at which point you can arrange to increase your police rating if you want. -1 morale does actually hurt some because it makes all your units worse at psi combat, only morale doesn't affect mind worms. The Gaians also get +1 food from xenofungus, which doesn't sound like much, but 2 food tiles are rare in the early game without terraforming, and as T-Hawk illustrated, having an effectively infinite source of 2 food tiles for supply crawlers is an underrated bonus. Their starting tech is everyone else's first research target, Centauri Ecology, because it gives access to formers, meaning they can start building formers on the first turn. The Gaians do have one further hidden disadvantage that I'll explain in a bit.

I want to explore what it's possible to do with mind worms under optimal conditions (because if I can't build a sufficient mind worm army from captures under optimal conditions, I surely won't be able to under others), which means I'm playing with different settings from T-Hawk. First, I'm playing with abundant native life. This means xenofungus everywhere and more mind worm attacks, which in my case means more chances to capture mind worms. This also tends to serve as a break on expansion, which hurts the AI but oh well, but it does make it harder to reach truly ridiculous research rates early. Second, I left pod scattering on. Pod scattering leads to a very different and IMO more interesting game even though some of the pod results are ridiculously swing-y. Pods end up being responsible for revealing most of the nutrients/energy/minerals bonuses and add monoliths, which as 2/2/2 tiles are the best non-bonus tiles you can work until the uncapping techs, which makes base placement more interesting because it breaks up the pinwheel resource pattern SMAC inherited from Civ2. More relevantly for this variant, popping pods can result in one or up to eight mind worms appearing next to your unit. While these are technically "bad" results, if you're trying to capture mind worms, they give you a lot more opportunities. If you manage to pop a full eight worms, you expect on average to capture two worms if you can set up to attack the worms. Mind worms will sometimes not attack human-controlled mind worms and just move away, which makes it easier to capture the worms if you're popping pods with your own mind worms. Third, I'm playing on a huge map. This means my opponents will be farther away (this is not relevant for this variant, without a variant rule, this helps me avoid the temptation to add bases with impact rovers rather than colony pods), but also gives more fungus to explore which means more mind worms. The huge map also increases tech costs without a compensating increase in tech production until you can actually fill up all that land. Fourth, I'm playing with tech stagnation enabled, which AFAIK as I know just increases tech costs by a linear 1.5 factor. This might only bring the late game down to 2-3 techs per turn, but I find it interesting mostly because it extends the more interesting early and midgame without unduly lengthening the mechanical endgame.

One of the things that I learned from reading T-Hawk's thread was the reverse engineering of the missing tech mechanic. I wrote up a Python program that finds beelines that avoid the tech holes, or establishes that no such beeline exists. I tested two beelines for this variant, the usual Industrial Automation and going to Centauri Empathy for the ability to build roads in fungus. With abundant alien life and Unity pod scattering, there is a lot of fungus around, and a good military road network is helpful for dealing with mind worms. It also gives the Empath Song unit ability for +50% attack against psi defense and the Green economy model (which I won't be using immediately but will eventually). Sadly, my testing confirmed that the bog-standard Industrial Automation was the right play here even though Centauri Empathy is net three techs versus five for Industrial Automation (Secrets of the Human Brain is required for Centauri Empathy and gives a free tech). It was smarter to just build more formers and clear fungus the hard way places where I need to build roads. For the Gaians, that hidden disadvantage I mentioned earlier is that it's not possible to go straight to Industrial Automation, you have to pick up one tech off the beeline before reaching the goal tech. (The same is true of Centauri Empathy, which is another reason why that beeline doesn't make sense.) Assuming I slot in another tech, there are two allowed paths to Industrial Automation: Information Networks, Industrial Base, Industrial Economics, Planetary Networks, and Industrial Automation; or Information Networks, Planetary Networks, Industrial Base, Industrial Economics, and Industrial Automation. Since the Gaians can run Planned but not Free Market, the latter, which gets Planetary Networks first, is better.

[T-hawk]

I was curious if my memory of having an adequate army from only captured mind worms was at all accurate, and thus was born this variant.

Sure, captured mind worms can be plenty enough to beat the (unmodded) AI.  It's just that worms aren't really any better or faster than doing it with impact rovers.  Fine for variant material though.

IMO, efficiency is the strongest social engineering bonus in SMAC except for the special effects you get from +6 growth and +2 economy.

This is correct, at least once you're up to enough bases for the b-drones and inefficiency losses to matter.

For the Gaians, that hidden disadvantage I mentioned earlier is that it's not possible to go straight to Industrial Automation

It's a curious property that the three next-best builder factions after the University all happen to share that disadvantage.  Gaia, Morgan, and the Peacekeepers occupy faction slots 1, 4, and 7, which are mutually equal mod 3 to each other, so get the same missing-tech holes.  So all of them have Industrial Automation missing at that same critical step of the 6th tech overall including its prerequisites and Centauri Ecology.

Oh, and I'd be interested to see that Python script. I kept wanting to automate my beeline-finding, but every time I was just like "nah I'll manually do it this one time in Excel and keep playing", and that one time turned into every time.

And supply pods definitely make the game more interesting, particularly on a huge map with late AI contact.  I mostly played without them (did once with Sparta) to keep the games within what could be described as a typical setup rather than artificially skewed as I'd been doing with Civ 5.

[Iainuki]

For the Gaians, that hidden disadvantage I mentioned earlier is that it's not possible to go straight to Industrial Automation

It's a curious property that the three next-best builder factions after the University all happen to share that disadvantage.  Gaia, Morgan, and the Peacekeepers occupy faction slots 1, 4, and 7, which are mutually equal mod 3 to each other, so get the same missing-tech holes.  So all of them have Industrial Automation missing at that same critical step of the 6th tech overall including its prerequisites and Centauri Ecology.

Oh, and I'd be interested to see that Python script.  I kept wanting to automate my beeline-finding, but every time I was just like "nah I'll manually do it this one time in Excel and keep playing", and that one time turned into every time.

I can post my script, but before I do, do you know what the effect of the "traded tech" option on the formula is?  Apolyton has been down a bunch lately, so I haven't managed to find where that's discussed.  It's not very polished (at the moment I define the techs I have by changing a constant in the code) but it seems to match the results from the spreadsheet where I've tested it when tech trading isn't involved.

As for the Industrial Automation beeline, interestingly there is a way for Morgan to get it without Centauri Ecology first:
Industrial Economics
Information Networks
Planetary Networks
Industrial Automation

Of course going without formers for that long would be a bad idea .

According to my script, though, and I checked it with the spreadsheet, the Peacekeepers can research Centauri Ecology first and then go straight for Industrial Automation using either:

Information Networks
Planetary Networks
Industrial Base
Industrial Economics
Industrial Automation

Information Networks
Industrial Base
Industrial Economics
Planetary Networks
Industrial Automation

The Believers, however, can't get Centauri Ecology and then Industrial Automation without picking up an extra tech (unless my code is buggy).  They, the Morganites, and the Gaians are the three factions with that liability, and for the Believers that really adds insult to injury.

[Iainuki]

My last test game before this one had a crazy good start, with a ridiculously long river that I fit four bases onto that lead into the Monsoon Jungle, several nutrient bonuses alongside the river, and an early monolith.  I hated giving that up but it would not have been very representative.  This is a much more typical start, with the standard two colony pods on Transcend.  I almost never found in place in SMAC.  Here, I wouldn't put Gaia's Landing down until 2013, both because I wanted to explore a bit to see if there was a resource bonus in the fog right outside my starting position and because I wanted to found on the river.  Notice that the river is only three tiles long and two of them are covered with xenofungus, so the tile I did found on was the only one I could.

The only thing this start really has going for it is the nutrient bonus that I just popped from a pod with the scout, along with a mind worm.  Because of the early-game grace period, the mind worm would not attack and I would get the guaranteed first capture in 2104.  I'm not going to mention every pod pop, but I'll try to cover all of the ones with game-changing effects.  I killed some more mind worms (each kill gives 10 credits) over the next rounds and used the money to rush formers out of Gaia's Landing.

Let me expand a bit more on why I describe this land as mediocre.  Every tile here is arid, moist, or rocky, and they have no special features, so except for the nutrient bonus, without terraforming the best tiles I have available are 1/1/0 or 2/0/0.  Every land tile in SMAC is described by three parameters, rockiness, moisture, and altitude.  Higher moisture is strictly beneficial, with tiles ranging from 0 to 2 nutrients.  Rolling tiles are strictly better than flat tiles because they give an extra mineral with no downsides.  Rocky tiles receive no nutrients from moisture, can't have farms or forests on them, and give extra minerals if you build mines.  Every 1000 m of elevation increases the energy output of solar collectors by +1, but this land is all at less than 1000 m so I can't even get 2 energy with terraforming.  Rivers give +1 energy to any tile they cross.  Xenofungus negates all other bonuses, including rivers, and gives a flat 1 nutrient (2 because I'm Gaians) per tile.  Tiles with monoliths can't be improved but give 2/2/2 production.

All non-bonus tiles are capped at 2 nutrients/minerals/energy no matter what improvements you have on them until you develop three techs with some nontrivial prerequisites.  The bonuses straight up add +2 to the nutrient/mineral/energy yield but more importantly they remove the cap for that resource on that tile, meaning that you can get more than 6 total nutrients + minerals + energy off a tile.  Nutrient bonuses in particular are important in the early game because they are the only way to get a tile that gives more food than the citizen working it takes to support.



Gaia's Landing's nutrient bonus is on a flat moist tile, making it 3/0/0.  I start off working the tile to grow ASAP while foresting the tile.  A farm and solar collector would take it to 4/0/1 for 8 turns of former work, or a farm and mine would take it to 3/1/0 for 12 turns of former work.  A forest takes it to 3/2/1 for only 4 turns of former work and can expand to adjacent squares for free.  On mediocre land like what I have to work with here, forests are the best choice. Meanwhile, I popped another nutrient bonus from a pod for my second base, on a rocky tile that starts as a 2/1/0.  A mine with a road would only make it a 1/2/0 (1/4/0 after lifting the cap), which is still worse than a forest in net productivity and doesn't help the city grow.  You can make a rocky tile rolling by spending 8 turns, which allows me to put a forest on the second nutrient bonus.  With the former from my second base, I immediately start in on this, but it takes until 2119 to finish the first step, much less the second.  Because the base grew to size 2 on 2119, I should have planted a forest on some other tile before leveling terrain on the rocky nutrient bonus.

[T-hawk]

I think the "traded tech" piece of the missing-tech formula only happens in multiplayer. I'm pretty sure I never had to use it in any of my runs.

Yes, the Peacekeepers can get to Ind Auto with no missing techs. But it's still the same pace as Gaia and Morgan, the 7th tech rather than 6th (including Centauri Ecology), since their starting tech of Biogenetics isn't on the prerequisite line, that tech is just equivalent to whatever off-line tech Gaia or Morgan have to pick.

For early terraforming, yes, forest is best on anything worse than moist/rolling. Rolling tiles aren't quite strictly superior to flat - some terraforming (roads and solar collectors) takes longer on a rolling tile. Solar collectors are basically never worth the terraforming time - go build forest on a new tile instead or roads to the next base site - improve your overall productivity in total more rather than making every tile perfect - labor-turns are a scarcer resource than workable tiles, unlike any of the later Civ games.

Mines are useless before Ecological Engineering to raise the mineral production cap, that rocky nutrient would only get to 1/2/0. Yeah, rocky-nutrient tiles are a pain to decide what to do with; leveling plus forest is probably best, but do any other easier terraforming jobs in the area first.

Are you going for the Planetary Transit System?

[Iainuki]

All units when moving onto fungus have a chance to cause a mind worm to spawn, the chance is higher for the first time you move onto a fungus tile (1/3?) and smaller for all subsequent moves.  In 2109 my mind worms spawned a mind worm and followed up by capturing it, which was pretty lucky.

While all this was going on, I researched Information Networks in 2112 and went to Planetary Networks.  Around this time my bases both hit size 2 and were generating enough energy that I could increase my research speed by unbalancing the sliders to 30%/70% in favor of labs, an early game advantage the Gaians get from their inherent efficiency bonus.  Planetary Networks finished in 2120 and I switched to a Planned economy.  The efficiency hit forced me to pull back the sliders to 40%/60%, but I thought the growth and faster production were more important than tech at this point.  In 2020 I popped Industrial Base from a pod, which was extremely lucky, and went straight from there to Industrial Economics.



I kept exploring and found another nutrient bonus for a third base. It's on an arid flat tile so starts as 2/0/0, but once forested it would be as good as the moist flat bonus at Gaia's Landing.  There's also an off-shore nutrient bonus here covered with xenofungus.  These can be very worth developing since with a kelp farm they give 5 nutrients before Gene Splicing (the tech that uncaps nutrient yields). For comparison, the best land tile before much later tech is a rainy nutrient bonus with a farm and condenser that gives 7 nutrients. However, kelp farms only take 4 turns to build and randomly spread on their own like forests, while a condenser takes 12 turns and a farm another 4.  A tidal harness which adds another 2 energy (3 after uncapping) takes another 4 turns to build and can coexist with the kelp farm, while a condenser precludes other improvements.  The resulting 5/0/2 tile produces more nutrients + minerals + energy than even a forested nutrient bonus on land.  The downside to ocean tiles is that sea formers require two extra techs, Doctrine: Flexibility and its prerequisite Doctrine: Mobility.  (They also take more minerals but the tech cost is the limiting factor.)  I don't know when I'll have time to get them.

The colony pod is sitting on my fourth base site, which I picked because it has a small river.  I have been building roads to future bases to speed up colony pod movement and also putting sensors underneath my third and subsequent bases to reduce the chances of a mind worm appearing out of the xenofungus and killing the lone garrison unit in each base.  Sensors give visibility on units in two squares and +25% defense.  This is a bit paranoid but while I've lost formers and crawlers to mind worms in this game, so far I haven't lost a base defender.  The cost is that it slightly slowed down my expansion in the early game, though usually only by a turn or two.  The real driver of this is that because of the early nutrient bonuses the speed with which I needed to build colony pods to keep my bases in the happy cap, currently at 2, exceeded the number of terraformers I could build for building roads and sensors for my new bases.  If I'd had an monolith or mineral bonus at one of these early bases I would have had more formers by this point.

It was around this time that I built my first rover formers, using the exploit where I take the chassis from a probe team, available with Planetary Networks, and add terraforming.  Players familiar with Civ4 know that the best UU in the game is the fast worker, and rover formers are good for similar reasons in SMAC.  Rover formers cost more and are no faster at building improvements than ordinary formers, but they accrue turn advantage because they can move onto tiles that only cost 1 move without wasting their turn and then build a road for ordinary slow formers to follow them.  Their most valuable contribution was building roads to new base sites.  I tried to keep in between a 2:1 and 1:1 ratio of slow to rover formers for the next phase of the game, depending on whether I needed more roads or more tile improvements.



I popped a Unity rover in 2132, which sped up my exploration.  I didn't get any others so I eventually built some scout rovers using the same probe team trick as for the rover formers.



This shows my bases in 2145.  I had finally found some land worth a damn around Autumn Grove, which has a rocky minerals bonus (0/7/0 with a mine and road), a nutrient bonus, and some rainy tiles.  Greenhouse Gate was a filler base that didn't expand my military perimeter.  I applied a Civ4 trick by sharing The Flowers Preach's nutrient bonus.




I finally popped a monolith north of Autumn Grove.  A monolith isn't just a great early game tile, any unit you send to it will automatically gain a bonus morale level permanently if it's undamaged, and this also affects mind worms, which gain a life cycle bonus. Effectively, the first monolith means a morale upgrade for your entire army, and this is honestly late to see the first one.  I settled my seventh base here for the monolith and the energy bonus, as I still desperately needed more energy to speed up research.

2128 03 The Flowers Preach
2138 04 Song of Planet
2139 05 Greenhouse Gate
2145 06 Autumn Grove
2152 07 Last Rose of Summer

And to answer your question, T-Hawk, I am going for the Planetary Transit System.  I hope to get to that in my next report, but we'll see.

[T-hawk]

You haven't yet lost a base defender to worms, because worms attack bases at a -50% combat penalty before year 2150. Be cautious from now on, though.

Rover formers are good value only for roads, as you mentioned. They cost twice as much as regular formers, but the only thing they do twice as fast is build road on flat tiles. 2:1 is too many, I think, but judging it by ratio also may not be right; just have a rover former going in each of the directions where you want more roads.

Monoliths give the morale upgrade even if the unit is damaged, the popup just doesn't say so because it tells you about the repair instead.

[Iainuki]

In 2152 I popped my first Unity foil from a pod in the far north.



It didn't get much exploration done because I brought it back to my bases to transport alien artifacts to my core without having to escort them across worm-infested wilderness.  After it did that and I tried to pop some more sea pods, it promptly got eaten by an isle of the deep.  I would find two more artifacts after this one.



These are my units as of 1255.  I was still running very light on formers at this point and only had 2 rover formers, which wasn't remotely adequate when I was trying to forest all the flat and rolling tiles at the same time as building roads to two new base sites at a time.  Every base I had was building formers, rover formers, or a colony pod at this point.

I'd finished Industrial Economics in 2133.  While this tech is an important milestone for all the other factions except the Hive, for the Gaians who can't run Free Market it's just a useless prerequisite on the road to Industrial Automation.  Since I couldn't go straight to Industrial Automation, I picked Biogenetics as my off-beeline tech. The other reasonable alternatives were Doctrine: Mobility and Social Psych, but both those techs had no immediate benefits, only opening up other techs.  Biogenetics came in 2150.  It gives Recycling Tanks, which I built in two food-poor bases to prevent them from starving when working three forests after I would build the Planetary Transit System, and the Human Genome Project.

I also got Zakharov's comm frequency in 2137.  I paid him off with a tech, signed a treaty of friendship, and promptly started ignoring him, as I was reasonably sure that talking to him again would only result in more demands for tech or money.



In 2161 I popped seven mind worms from a pod.  They killed the rover in this screenshot, but I brought other rovers forward and captured two in 2169, killing most of the rest.

In 2162 I finished Industrial Automation.  The Planetary Transit System and Wealth are both quite strong, each by itself would make the tech worth researching, but supply crawlers are the reason that Industrial Automation is the tech you beeline at the beginning of every game.  Supply crawlers let you increase minerals and energy output when you are still limited, as I am at this point, by a happy cap of 2, and they also make it possible for a collection of size-2 bases to collaborate to quickly build secret projects.  I'd saved up about 500 credits from worm kills and energy income while waiting on Industrial Automation.  I adopted Wealth, pulled the sliders back to 50/50, and started buying supply crawlers as fast as I could.  At this point, I had no way to make more expensive supply crawlers to upgrade into, and this early buying them is just as efficient.  While I massed up the supply crawlers necessary for the secret projects, I had them work spare forest tiles and converge onto Greenhouse Gate and Gaia's Landing, as these were my most central bases.

I started research into Social Psych because it's a prerequisite for both Ethical Calculus, the gateway to population booms, and Secrets of the Human Brain, on the way to Centauri Empathy.  In 2164 I popped Applied Physics from a pod, which was actually really annoying since it noticeably slowed down my research while not being a prerequisite for anything I would be researching for a long, long time.



In 2167 I found the Hive.  Yang demanded 75 credits which I paid, refused to sign a treaty, and thereafter stopped talking to me.  I'm not sure what I'd done to earn the instant hate since I wasn't running Democracy and wasn't the top faction in score.  Their land was nothing special, a fungus-covered river and some equally fungus-covered offshore energy bonuses, so I pulled back to Memory of Earth and left them alone in the short run.





I'd continued exploring and founding bases.  I kept hoping to find better land than the featureless arid and moist plains I had to work with, but the map did not avail.  I was pretty sure the game was laughing at me when I found Sunny Mesa, the one landmark in SMAC that provides no increased yield.  (It's 1000m elevation so you could theoretically build solar collectors on it for a whole 2 energy per tile.)



I was up to 10 bases by this point with formers prepping the site for the eleventh.



The "Secret" Project screen always makes me laugh because it shows the projects that all other factions are working on and how many bases they're being built in. The only one of these projects that matters is the Virtual World, which the University had been building since 2153. I was nervous about losing it but not nervous enough to prioritize it over the Planetary Transit System: I figured the minerals I would get from the Planetary Transit System would speed up the Virtual World enough so that I could still beat the University to it, and the Virtual World had no immediate benefit for me.



In 2172 I finished the Planetary Transit System with 12 bases built. The PTS not only brings all current bases up to size 3, it also pacifies a drone in bases of size 3 or smaller. This brought my happy cap up to 3 (1 for Transcend + 1 police + 1 PTS), and took me from 25 to 32 labs each turn, a 28% increase. This is about what I expected, because approximately a size 2 base works a total of 3 tiles (the base square and 2 others) while a size size 3 base works 4, giving a one-third improvement.



For the first time, that put me on top of the other factions in score, though Santiago still had more population than me.  I assume she had landed in the Monsoon Jungle.  I wonder if the bonus morale the Spartans get had given her enough of an edge against the mind worms that she was able to expand better than the other AIs?

[shallow_thought]

Enjoying this .

[Iainuki]

This was the project status in 2175.  One note I forgot to add to my last report: to speed up the Planetary Transit System, I used my first alien artifact (adds 50 minerals, 20 more than a supply crawler) on it.  Alien artifacts can also be used to research a random tech *after* you build Network Nodes.  As I wasn't going to be finishing Network Nodes for another dozen turns at least and many of the available techs were not useful in the short run, I thought finishing the PTS in 2172 was more important than a random tech.

I think of the Planetary Transit System as one of the four must-have early projects.  The other three are the Human Genome Project, the Virtual World, and the Weather Paradigm.  "Must-have" is an exaggeration, it's certainly possible to win without any of them given how weak the AI is, but these four projects are the ones with payoffs that far exceed their cost.  T-Hawk covered the importance of the Human Genome Project for golden ages in odd-sized bases because of the cap on the benefits from psych spending, but it's also one of only three ways to increase the happy cap in the early game without changing your social engineering choices to increase your efficiency or police rating.  For the Weather Paradigm, I typically spend at least 1000 minerals on formers over the course of the game, spending 200 minerals to reduce the total number of formers I need by one-third would be worth it even if that was the only thing it did.  However, it also enables boreholes, raising and lowering terrain (which is mostly useful for boreholes), and condensers before the normal techs. Drilling boreholes on energy or mineral bonus tiles offers 0/2/8 and 0/8/2 tiles immediately and lets you get a head start on boreholes everywhere else while you research the necessary techs.  The Virtual World adds the effect of Hologram Theaters, which cost 60 minerals and 3 energy maintenance, to Network Nodes, turning them into facilities that give +50% research and psych and pacify two drones for 80 minerals and 1 energy maintenance.  Like the Human Genome Project, the Virtual World is strong because sources of happiness are limited early in the tech tree.  Unlike the Human Genome Project, the Virtual World just makes it cheaper to reach the same happy cap, rather than giving you a higher happy cap than is possible without the project.

Since the University had started building the Virtual World in 2153, a realistically pessimistic estimate of 10 minerals per turn suggested they could finish it in 2174 at the earliest, but probably I could still get it if I went for it as hard as I could.  Finishing the Planetary Transit System first was a risk, and I wasn't willing to risk the Virtual World further even though both of the other projects had more immediate benefits. No AI had started either, which could have meant that only the Peacekeepers had Biogenetics and no one had Centauri Ecology, in which case both the other projects were safe.  I pulled the sliders back to 50/50 to get more cash and kept rushing crawlers as fast as I could.



I finished the Virtual World in 2178, two turns before my estimated deadline.  At that point I figured I was home free, so I unbalanced my sliders to 40/60 so I was getting 32 science again.  The University immediately switched to the best project they could pick from my perspective, the Command Nexus, which is all but completely useless in this variant, and finished it the next turn in 2179.  The Morganites also finished the Merchant Exchange the same turn.  The Believers switched to the Weather Paradigm and the Morganites and Spartans to the Human Genome Project, but they hadn't been building those nearly as long, and I had a steady flow of crawlers converging in my core. As I expected, I nabbed the Human Genome Project in 2180 and the Weather Paradigm in 2181.



This is how the early projects looked after the dust settled, a perfect outcome as far as I was concerned.