SMAX - The Will to Power - mod

[bvanevery]

Well, 6 AM and I finally quit.  I never lost a single engagement to the Usurpers, and generally speaking they had no answer for anything I did.

An initial rash of Missile ships, terrorized their sea bases, completely obliterating them.  They never fought back at sea, and the whole thing was a game of whack-a-mole, eliminating Sensors and new bases as yet another sea colony pod sprouted.  I'm quite surprised at the number of wounds a civilian sea colonist can commit on defense, it seems quite excessive compared to Missile ships.  Also for some reason I wasn't allowed to equip them with chemical weapons, which I consider an unreasonable deficiency.

On land, it took me forever to get my cities properly garrisoned.  Since the Usurpers actually had no offensive potential at all, this was a complete waste of time.  What they did do, is keep rebuilding cities I'd already killed, in the same places.   I felt like I had an extreme lack of productivity, far more than just my -1 INDUSTRY penalty would account for.  It's like, I eventually get stuff, but the rest of the map just grows and grows and grows.

The graph said I was the 2nd crappiest empire, after the Pirates.  And that the Usurpers were substantially better than me, even though they couldn't do a darned thing to me.  How does someone manage to be better when all they do is lose?  I stole like every tech in existence from the Hive.  Didn't quite have that finished up when I quit, but I was close.  The Hive never did anything to me either.

It seems like the only thing the AI has got going for it now is spam competition.  This is very tedious for a human player, and I don't see that it's militarily effective.  It's also a huge tech leakage giveaway.

In contrast, with the stock binary + my mod, a militant faction that gets ahold of so much as Recon Rovers, is capable of wiping you out, if you're not careful and forget to defend yourself with proper garrisons.  This is even with Huge map and all the biases I've put in the tech tree towards early defense.  That's not a brag; I didn't program that AI behavior.  It says that the AI in Thinker or WTP, is making very serious mistakes about how to allocate effort, compared to the stock binary.

I think if I didn't bother to make the latest greatest expensive garrison units, and only made X offense, I could probably wipe out a nearby Alien faction pretty easily.  Or humans if I was playing an alien.  I may not need to make a navy at all.  I didn't see any basic competence at sea.

Some specific AI pathologies I noticed:

* suiciding Scouts against much stronger armored units.  Especially egregious when the armored defender is in a city.

* building expensive armored probe teams that are merely Disciplined, when they're facing a Fundamentalist empire with Commando or Elite probe team defenders.

On the positive side, I had no problem getting my empire established.  That's 2 games now where I didn't think there was any basic fairness issue with mindworms and whatnot.  And I did have quite a bit of fungus touching my cities this game.  It was only later on that I cleared it with Fungicidal Formers.

[Alpha Centauri Bear]

Something makes the game take a relatively long time to start up.  At least 30 seconds after I set all factions to random and then hit ok.  Maybe as long as 1 minute.  There's just a blank screen waiting that whole time, an uncomfortable pause that makes me think something is broken.  I noticed this behavior in both versions 218 and 215.

Could the world generation or faction placement algorithm be iterating over and over again without success, until finally something succeeds?

In comparison, stock binary has very little delay from time hitting ok to the time telling me what faction I got.

I noticed that too. Could be world generator or PRAX patch Thinker embedded. Not sure. Does it happens every time?

[Alpha Centauri Bear]

Why are the Usurpers able to settle bases on fungus?

Details incredible colonial tech

incredible colonial tech

Even if you thought Aliens should be able to do this for some reason, it makes no lore sense at all for the Usurpers.  This smells like a bug.  Heck it doesn't even make game mechanical sense, as your version of the Usurpers has the unaltered -1 PLANET.

https://github.com/tnevolin/thinker-doer/blob/master/wtp_changelog.md#version-209
To aid AI in base placement among large fungus/rocky patches. It doesn't know how to clean spot for bases with formers. I may also automatically remove fungus/rocky from base site to not confuse people. Probably I should do it.

[Alpha Centauri Bear]

The unit building and moving AI is the most complex aspect of the game. Induktio keeps improving it and I am contributing some here and there but there is still lots of work to do.
If you have time to play Thinker 2.5 as well - try it out and let me know if it suffers the same problem.

[bvanevery]

I noticed that too. Could be world generator or PRAX patch Thinker embedded. Not sure. Does it happens every time?

Yes, every time.  Smells like world generator rejecting worlds over and over again.  Seen this sort of thing so many times in so many different game genres.

My 1st try of version 220, it didn't happen.  This time I only Quick Started my game, hopefully relying on previous RANDOM faction settings.  I drew Cha Dawn, which does seem random compared to previous games, so I think the RANDOM settings have been retained.  I guess we'll see if the problem comes back, or has been squashed by your Thinker 2.5 code merge.

To aid AI in base placement among large fungus/rocky patches. It doesn't know how to clean spot for bases with formers.

Well, that's a cheat for the AI.  But at least in the stock binary, it would be a defensible one, considering how badly the AI does with colonists and fungal mazes.  I'm surprised that Thinker doesn't handle things ok, considering how much more AI oomph it's got in settlement algorithms.  Are you sure you really need to have the AI push out this many more colonies, on these places?  Doesn't Thinker already spew enough bases as is?

It would be preferable to actually have the AI clean spots for bases, but I understand that could be a lot of work.

I may also automatically remove fungus/rocky from base site to not confuse people. Probably I should do it.

Removing the fungus and flattening the rocky terrain to rolling, would hide the cheat better.  Then if discovered, at least you could justify the cheat in your notes as necessary to solve an AI performance problem.  Could note it as an interim step, a kind of triage, even if you never get around to doing better.  "Ideally we'd like, but...."

If you have time to play Thinker 2.5 as well - try it out and let me know if it suffers the same problem.

I don't.  I was up to 6 AM playing your mod as is, and it's getting on towards 4 PM now.  Fortunately, you just merged 2.5, so that's halfway to knowing how it behaves.

I'm going to start a new game with version 220 shortly.  My prediction is: I'm not going to end up in a human-alien neighbor war.  I may not get the U.N. Charter properly repealed (prediction, hah!)  I'm going to make a token garrison force of Synth units, and I'm going to ignore the water.  I'll build my Speeders and summarily overrun my nearest neighbor, because AI is completely incompetent, wasting its production on colonization instead of defense.  Why am I confident in this prediction?  Because I was noticing similar behavior a couple months ago when I was last testing.  I don't think the "runaway train" of excessive colonization has been averted.

But we'll see.  I don't know how new objections about unusable roads and auto-destructing empires are going to factor in.  I suspect they mean there's no value in taking over bases.  Only in obliterating them.  Which may mean twiddling my thumbs until I've got Planetary Council clearance to legally do so.

[Alpha Centauri Bear]

I have just set huge map size and start game few times in a row. The game starts immediately. Can you check if you have the same effect with vanilla and Thinker?

[Alpha Centauri Bear]

To aid AI in base placement among large fungus/rocky patches. It doesn't know how to clean spot for bases with formers.

Well, that's a cheat for the AI.  But at least in the stock binary, it would be a defensible one, considering how badly the AI does with colonists and fungal mazes.  I'm surprised that Thinker doesn't handle things ok, considering how much more AI oomph it's got in settlement algorithms.  Are you sure you really need to have the AI push out this many more colonies, on these places?  Doesn't Thinker already spew enough bases as is?

It would be preferable to actually have the AI clean spots for bases, but I understand that could be a lot of work.

It's a huge huge work nobody wants to do. And if it is only to clean place for base then this cheat just solves it for cheap. I am not sure if this cheat should be there or we should deprive AI from such areas. I don't think it really matters much. I just introduced it to see how AI expands in general and these patches just go in the way of my observations. I can turn it off if people say it makes AI overly strong.

[bvanevery]

Instant startup is a certainty for vanilla.  Years of testing on that.  Don't know about Thinker.  Kinda busy conquering.  You could test it yourself, wouldn't take you any more time than it takes me.  It's not going to be any different on my machine with a "clean install" of Thinker.  This isn't a graphics problem, the recent instant startup proves that.  The problem may be gone now.

Regarding the colony clearing cheat, my working hypothesis is that the AI fights badly because it spends too much time colonizing.  Thus, denying more colonization spots might be in your best interest.  A halfway house would be to allow the cheat for X number of configurable turns at the beginning, then turn it off.

I'm currently on MY 2153.  From what I can see of the map, here are the colonial spreads.  Cha Dawn - me - 8 cities.  Usurpers - at least 7 cities, some map blackout obscuring.  They will soon move into the Monsoon Jungle.  University - 6 cities.  Gaians - 6 cities.  Cyborgs - 4 cities.  Peacekeepers - 6 cities.  That leaves only the Pirates unaccounted for, as nobody seems to have met them.  I can see Safe Haven because I found the USS Unity crash site.

Yes, I'm better at colonizing than the AI is.  I know how to completion scum, and I've gotten pretty good at masking the long colony production times.  I also started on pretty good land, lacking only a minerals deposit so far.  And I'm the Cult, giving me lotsa money from early exploration popping pods.  I'm now going through a phase of building Rec Commons and Rover Formers to follow, as my road network isn't enough for my reach.

I have prototyped Plasma armor with a completion scum.  Gatling guns are coming by sheer long term brute force.  Didn't pick up a completion scum on that one.  I don't need those techs yet, I was just being a boy scout for any upcoming trouble.  Since the Usurpers are my nearest neighbors, I wasn't wrong to do that.  But so far, we have a Treaty and aren't touching each other yet.  If we do touch, I predict they're dead meat.

Anyways, empiricism says 60 turns of settlement cheat is totally reasonable.  Wouldn't shock me if 80 or 100 turns of cheat turns about to be reasonable as well.  Whereas, when I took the screenshot of 4 bases on the front line of last game's Usurpers, all lining up in a huge mass of fungus, I thought that was pretty unreasonable and goofy looking.  Hard not to notice that, and seemed pretty excessive given the sprawl of that empire elsewhere.  That was in MY 2198, near the 100 turn mark.

[Alpha Centauri Bear]

Instant startup is a certainty for vanilla.  Years of testing on that.  Don't know about Thinker.  Kinda busy conquering.  You could test it yourself, wouldn't take you any more time than it takes me.  It's not going to be any different on my machine with a "clean install" of Thinker.  This isn't a graphics problem, the recent instant startup proves that.  The problem may be gone now.

There is no point me testing it. I do not observe it.

Regarding the colony clearing cheat, my working hypothesis is that the AI fights badly because it spends too much time colonizing.  Thus, denying more colonization spots might be in your best interest.  A halfway house would be to allow the cheat for X number of configurable turns at the beginning, then turn it off.

Oh. It definitely does. That is why I see how to nicely deprioritize expansion in favor of protection. All AI tuning is a tedious work.

What's completion scum?

[bvanevery]

Completion scum is when your production of a new unit is at 10 minerals or less, so you change your production to the most desirable unit or facility you could possibly get, just before popping a pod.  You do this at the city most likely to be closest to the pod.  It is extraordinarily profitable in the early game, especially for the Pirates where I first perfected the technique.  Needed to belt out Clean Synth Transports and Sea Colony Pods in a hurry.

Later I realized it was also useful on land in a lot of contexts.  For instance, if you completion scum a Command Center, you can devastate a nearby enemy, much faster than you'd expect a CC to be built.  Your long build time for CCs actually means nothing to me early game.

If you don't understand completion scumming, you might want to go up that learning curve, and decide whether you want to nerf it somehow.  It's very powerful.  Not quite thermal borehole, supply crawler, and condenser powerful, but it really gooses early game spread.   And the AI has no idea how to do it at all.

In fact, the WTP AI doesn't even seem to know how to pop pods.  I wandered all over the ocean last game, popping pods it hadn't picked up.  I could swear that previous versions of Thinker I've played, cleaned out all the oceans quickly.  Can't remember about WTP.  But anyways, the AI is leaving a lot of early money lying on the table.  Not good.

[bvanevery]

I keep getting distracted from the original thing I was going to post about.  As the Cult, of course I'm acquiring wild hatchling mindworms.  When they survive a combat, why are they never leveling up?  Is this intentional?  In the stock game, winning with a hatchling guaranteed turns it into a larva.

[Alpha Centauri Bear]

I keep getting distracted from the original thing I was going to post about.  As the Cult, of course I'm acquiring wild hatchling mindworms.  When they survive a combat, why are they never leveling up?  Is this intentional?  In the stock game, winning with a hatchling guaranteed turns it into a larva.

https://github.com/tnevolin/thinker-doer/blob/master/README.md#promotion-probability

[bvanevery]

Details lotsa fungus cheating

lotsa fungus cheating

MY 2198.  This is typical of AI empires.  I definitely think the fungus cheat should be cut off after 100 turns at the latest.  This is plenty enough cities to make a core empire.  It gives the unfavorable impression that the AI doesn't have to think, plan, or play by the same rules you do.

Cutting off at 80 turns may be more appropriate.  I remember the Usurpers having a substantial number of fungus cities by then.  60 is probably too early.

I'm still at peace with the Usurpers so I'm just going vertical with my 8 cities.  Slowly putting trained Fusion ECM Synth Garrisons in place, using only 1 Command Center that did in fact take me forever to build.  All factions are pretty even on the graph.

I'm finding that WORMPOLICE is an advantage, in a game where it's hard to have Police units.

[bvanevery]

Ok, after 5 hours of nearly uninterrupted play, I quit.  MY 2230 and I'm bored out of my mind.  Nobody does anything.  Usurpers never so much as breathed heavy in my direction.  I was starting to make Genejack Factories.  Other factions made various Secret Projects, which seem totally unprofitable to pursue myself.  I've got all sorts of tech that has no value, because things would be too expensive to build, and only the most minimal of defense is worth bothering with anyways.

Exploration kept me going for awhile, but then when the pods are all dried up, there's really no reason to be out there anymore.  Unlike the stock game, trying to capture more mindworms is mostly fruitless.  You will get one occasionally, but it is not worth the intervening mouseclicks.  Using your mindworms to fight other mindworms is pointless, you'll just get killed half the time.  So I don't.  I just evade and go pop another pod.  When those are gone, there's nothing left to do.

Used to be I'd go execute Fungal Towers to make some money, but I don't think I get any money for that in WTP.  So the  usual suicide ritual of sending 3 untrained Scouts against a Fungal Tower, to reduce my SUPPORT, is pointless.  Unless there's one nearby in the way of my expansion, which there wasn't.

Marr had this huge but crappy empire north of me.  And if I go try to take it over, everything up there is going to auto-detonate.  There's no reason to do anything up there, because there's no profit in it for me.  I'd only go if he was a threat, and he isn't.  Also, threats can generally be dealt with by taking 1 or 2 cities.  Then factions will shut up and leave you alone for awhile.

What's the point of this game?  Sit around and level up my cities forever?  If that's all there is to do, well I don't think I'd pick the Cult of Planet to do it.  They're not the most useless faction for building, but they're not what I'd call advantageous.

Sitting around waiting for stuff to get done building, takes a long time.  That's not much of a game.  I feel like I just went through 5 hours of foreplay, after which, I could not be bothered with SMAC porn anymore.

I never made a social engineering choice.  They're all bad.  You're better off in neutral.  Until someone declares war on you, at which time Fundamentalist is the obvious slam dunk total giveaway.  But that never happened, so I never made any choice.  I contemplated going Green Something once my cities got up to size 7, as I wouldn't need them to grow for awhile and could delay on Hab Complexes.  But even with 8 cities fully terraformed, only 1 of them made it to size 7 before I quit.

I had tons of money and nothing at all useful to do with it.  Unlike my mod, I can't just drop it on Secret Projects.  They're too expensive.  None of the buildings are worth anything and nobody's doing anything, so, why bother?

Like why did I build Biology Labs and Network Nodes?  It doesn't make any difference to research, I can just buy other people's research.  Except for the tech for Recycling Tanks, because it takes forever for someone to build The Human Genome Project and they all block on trading the tech until then.  It's the only tech they actually do block on.  I don't know why they're so uniquely fixated on that one tech.  It's really weird that they'll trade lotsa techs with SPs, but not that one.

The Cult possibly could have made use of the BLs for a mindworm assault, and I sorta had that partly in mind after getting the Genejack Factories done.  But chemicals are best against aliens and BLs don't help with that.  I was trying to talk myself into a hybrid assault model when just all sorts of feelings of meh washed over me.  Don't care.

Human history sure would have been different, if seizing foreign lands never had any value to it.

What to do, play the "take the 1 distant city that made a SP" game?  Well since rails don't work, you're sure gonna be pushing units a long time to do that.  Not worth it.

Play the "early land greed" game?  What if I don't like that really?  What happens when I'm bored with pushing colonists around?

Pretty much I feel nihilism.

I wonder if WTP is oddly vulnerable to Single City Victory?  I might test that.

[bvanevery]

There is no point me testing it. I do not observe it.

Quick Start once again came up instantly.  This time I drew Zhakarov, so all the previously established RANDOM factions are still set properly.  The original problem happened when I set all the RANDOMs by hand.  Not this time, but next game, I'll try that again.  If the problem doesn't recur then I suppose Thinker 2.5 fixed it somehow.

This game I'm going to try to jump nihilistically forwards with X Recon Rovers, spending the least possible on offense and doing the most damage.  Of the 4 X's, I don't think eXplore lasts so long now, and eXpand and eXploit seem pointless.  That leaves eXterminate.  The formal question is, just how broad a circle of destruction can be imposed how cheaply?

BTW when I play, as a matter of habit, I don't change the faction default research foci.  This allows me to perceive the faction more like the AI does.  Like whether it gets hopelessly stuck in the tech tree somewhere.  So far that has not been a problem.  There has been way more tech easily available on the open market, than I think I could ever possibly need.