SMAX - The Will to Power - mod

[bvanevery]

I never could understand why everybody thinks Yang is aggressive

Your Hive is set to Aggressive personality, as is vanilla.  If you don't want that, change it.  I did.

with its GROWTH (!) and INDUSTRY (!) bonuses as well as its free Perimeter Defense (!). It is a well protected economical paradise -

Hardly.  Your Hive has -2 ECON, as does vanilla.

[bvanevery]

Thinker mod AI can move units around in APCs?  If it can, we'll that's new, and an accomplishment.  If it can't, then APCs shouldn't be allowed / provided.  I don't offer predefined APC units for anyone to contemplate, because the AI doesn't know what to do with them.

APC = ground transport.

Yes, everyone talking here knows that.  Why did you think anyone misunderstood?  Doesn't matter. 

The stock AI does not know how to move units around in APCs.  It's a human player exploit only.

I highly doubt that Thinker mod AI knows how to move units around in APCs.  I haven't seen it do it.

[Tayta Malikai]

*Actually, that was something else I wanted to mention. In my past few games with WtP (modded or otherwise), the AI has been curiously unwilling to attack bases of mine that have at least somewhat decent defenses (e.g. a 2-defense unit, possibly with ECM or behind a Perimeter). It makes it rather easy to guard chokepoints against even superior numbers, and leaves their units easy pickings for me since they mill about uselessly outside the base. A base with a 4-defense unit might as well be invulnerable as far as the AI is currently concerned.

Guys, tuning an AI is a huge and endless work. It is like writing a chess program capable to beat the world champions. The only difference is that SMACX is much more difficult to handle since it is brand new game not thoroughly analyzed in tons of books yet.

Thinker did noticeable work on that: a few steps on the path. WTP added some more steps, etc. It will never be perfect. Recently I've tilted it toward expansion trying to mimic human strategy. It did well acquiring more economic power quickly but reduced amount of early conflicts, obviously. I think it is good as early conflicts just slow down the expansion and let others (human) grow stronger. Yet it could be toned down slightly if people think it is a mistake. I have also added some defense focus so that it would not be big empires with absolutely empty bases easy to conquer. You tell me if AI succeeds in it.

Besides, it is all single turn single unit heuristic. No  multi turn strategy, no unit coordination, no plans. Teaching AI to fight properly would be very difficult.

Yes, we should do well to keep the scope of the project in mind. It's not "unbeatable AI", it's "AI that is challenging enough to require some effort to beat". I guess I'm just skilled enough that it's not that challenging for me.
That said, I have lost to Thinker AIs before. One was an early worm rush by Stewards, the other was a Progenitor Victory by the Metastate. Of course, back then I didn't know as much about exploiting playing the game as I do now.

The reason I talk about "symmetric defense buffs" is because any mechanics changes you make will apply just as equally to players. Except that players can make better use of them. The armour buffs are a case in point: the AI having a couple of 6-defenders is an inconvenience for me, but me having a couple of 4-defenders is insurmountable for the AI.

Ironically, it may very well be that moderately privileging attack over defense (as Nevill advocates) would allow the AI to make more meaningful counterattacks.

I do agree that an early focus on improving the economy is a good way to go. Even a +25% territory defense bonus pushes the bar for an early attack higher, so they (and the player) can mostly develop in peace for a while.
I can report that the AIs in my game are all making tonnes of cash now. Almost all of them make 50-80 credits a year, with hundreds in reserve. They are also very competitive when it comes to tech: I suspect I only remain on top because my faction has nice research bonuses. The bright side is that this means I also benefit greatly from trading techs with them.
I forget, do AIs get bonuses to maintenance and research on Transcend?

The AI might even have remembered how to build projects now that support redistribution is disabled. Stewards are building the Maritime Control Centre in an 18-minerals base. Might be too late for it to recover, though; I don't plan to start a new game anytime soon to find out, unfortunately.

Re: APC.

Aren't they considered an exploit? I understand the infantry base attack bonus is given in compensation of it incurring hasty penalty every now and then which is obviously eliminated by using APC. Should we remove infantry base bonus? They are already balanced enough by their price.
I mean: either this or APC. One should go.

I also don't see much use for "mobile in open" bonus. Mobile unit already has benefit of their mobility. Meaning it will be attacking enemy infantry twice more often. It seems like pure lore introduced bonus just to add more "realistic" elements in the game. Same as attacking along the road, artillery altitude bonus, etc. I believe they all just clutter the tactics and complicate computations which is, as always, benefits human only. Simple rules let AI heuristic survive.

The AI barely knows how to use transports as it is. Does that mean transports should be removed entirely?

The same dynamic exists with sea transports and Amphibious Pods. A sea transport moves up to a coastal base, marines attack directly with +25%. APCs just allow you to do the same thing on land.

If you really want to get rid of it though, go into alphax.txt and change the "transport capacity" parameter for the land-based chassis to zero. That will render them unusable.
Of course, if you did that, I'd just have to build armoured infantry and march them alongside the attackers to take on the enemy bases. Same price, just takes more turns.
The main benefit of APCs is in multiplayer where it is essential to carry out your attack before the opponent can react. Using them against AIs is mostly just extravagance.

As for the infantry bonus vs bases, I'll defer to what Nevill thinks about this (in general, not just this mod). I wouldn't object if it was reduced to a token +10% or something.

25% (sensor) + 25% (territory) bonuses.

Just collecting feedback on that. Do you find sensor bonus a better replacement for territory bonus? In my mind they are fragile and they also not inherent - require some efforts. Moreover, sensor is incompatible with second level improvement so may be difficult to find a good spot for.

Requiring effort is the point, as bvanevery says. You make the decision to invest former-turns to shore up your defense. This is what any strategy game is built on.

I recall you argued before that sensors are too easily gotten rid of by an attacking force. Allow me to offer the counterargument that this is also the point. Having to divert aircraft and artillery to attempt to bomb sensors with 50% chances means they are not being used to directly attack enemy forces. More effort, more investment, more turns, a higher bar to clear for a successful attack. Small advantages matter.
The same argument also applies to using probes to disable Perimeters. Far from being an easy fix, it requires investment and effort on the part of the attacker.
Admittedly, with the AI's use of defensive probes being unreliable, it does somewhat make sense to disable this ability for singleplayer games.

Also, they are not so easily gotten rid of if you place defenders on top of them. Even better, put them inside bunkers or forests with the defenders. Now they can't be gotten rid of without assaulting the bunker.
I believe you can also build a sensor and then place a base on top of it to retain the bonus, but this is definitely an exploit.

[bvanevery]

Details that's the empire

that's the empire

Version 166.  MY 2187.  This is the point at which I quit.  I've only got 8 cities, only 1 of which is any good.  I theorized that Recycling Tanks and Children's Creches were the keys to productivity, based on the last game I played.  I used the Usurpers to beeline for them.  I managed to Complete some of those facilities with my early supply pod popping.  Well... the AI spam is overwhelming, and gives me very little feeling of agency over the game.  I had thought I'd overrun poorly producing cities with better producing ones, but it takes so long to get a city into production shape, it just doesn't work out.

Caretakers press against me and hassle me.  Cyborgs used to be allies but got snitty about wars I didn't want to fight.  Now this turn they declare war on me as well.  If I could handle it, it still feels like a desperate situation, and not at all fun.  I really don't like being curb stomped by gazillions of AI cities.  Either I figure out a way to destroy all this spam, or I give up and say this is dumb, not for me.

A strange thing about this game, is I didn't build any Formers for a very long time.  I built much of my road network with only the 2 original Formers.  That was a consequence of the "get the cities productive" strategy.

Possibly, there is no value in making peace with neighbors.  They will just deploy the AI's overwhelming colony production advantage, smothering you.

I think a Recon Rover rush is the next thing to try.  If that can't clear out the unwanted guests, then I think I'm done.  This is like trying to fight WW II when you're Switzerland.

[Alpha Centauri Bear]

Yes, we should do well to keep the scope of the project in mind. It's not "unbeatable AI", it's "AI that is challenging enough to require some effort to beat". I guess I'm just skilled enough that it's not that challenging for me.
That said, I have lost to Thinker AIs before. One was an early worm rush by Stewards, the other was a Progenitor Victory by the Metastate. Of course, back then I didn't know as much about exploiting playing the game as I do now.

I don't think fun project will ever get to the point where AI is unbeatable on the easiest difficulty level.

Ironically, it may very well be that moderately privileging attack over defense (as Nevill advocates) would allow the AI to make more meaningful counterattacks.

Vanilla promoted exactly this "active defense" as of some sort brilliant idea. Didn't work out for AI.
Most people loved snowballed vanilla conquest. Even though they got bored with it the nostalgy lives deep down in our souls. This is a clever self-deception to return that intoxicating sense of commanding powerful invincible army destroying everything in its way. Maybe even just a little. Maybe just a hint of it. Maybe just a smell of it. 😉
AI is dumb and does what it is programmed to. It will not automatically use any mean to help itself you give it until it is added to the program. End of story.
This is what all the dances are about. We are modifying game rules to adapt them to whatever AI currently know how to do. Not the other way around.

I forget, do AIs get bonuses to maintenance and research on Transcend?

What's maintenance bonus?

AI gets bonus on research in vanilla but not in WTP.

The AI barely knows how to use transports as it is. Does that mean transports should be removed entirely?

Don't fail to unconstructive generalization demagogy. It very depends on whether it makes game better or worse which we are trying to establish.
We are steering away from vanilla offense imbalance by making assailant work for their conquest. It seems to be fruitful. General feedback is that game became more interesting this way.
Land transports and infantry base attack bonus are clearly steps in opposite direction making base capture easier. That is why I am running this discussion to convince myself and others one way or another. So far I am inclined against infantry bonus and giving it a try in recent release. Not sure about land transport. So just talking to myself out loud observing this from different angles.

From the lore point of view, transports are supposed to help units to cross something they are not themselves capable to. They were not supposed to contribute any advantage to the battle by specifically their transporting ability. So far the sea transport transporting normal land units is a perfect representation of these paradigms.
Game designers extended transporting ability on other chassis in their run for global standardization. Yet, only land unit can board the transport. So land transport carrying land units on the same body where they can travel themselves makes no sense lore wise. The air transport carrying land units would make sense but Gravship is the only air chassis capable to serve this purpose without restrictions. So it seems extending this ability on other triads is just another unused design idea that would lurk there unnoticed in oblivion if people wouldn't make an exploit of it.

You are right the exploit is minimal in SP game as many people just don't bother building them. And it is no issue in MP as rules are same for everybody. So I'll wrap it in the way that I'd like to disable this but leave it for others to decide.

The same dynamic exists with sea transports and Amphibious Pods. A sea transport moves up to a coastal base, marines attack directly with +25%. APCs just allow you to do the same thing on land.

The comparison in incorrect. This is an actual ability! 😁 And it has its own cost which can be adjusted based on benefits. And it was reviewed from this specific point of view multiple times. And AI uses them, etc.

The main benefit of APCs is in multiplayer where it is essential to carry out your attack before the opponent can react. Using them against AIs is mostly just extravagance.

Exactly. Such mean to instantly concentrate forces for strike and crush the enemy wherever you like it invalidates the mere attempt to build balanced and economical front line defense. To counter this one should need to keep defensive strength in each potential attack target base equal to all invasion forces combined which is not feasible. It is much cheaper just to keep striking: you ate my base, I ate yours. We are falling back to vanilla madness where constant advancement in either direction was the most effective way to go even if player kept loosing from other directions. I have worked hard on eliminating that and deliberately slowing the conquest so that "permanent advancing" strategy loses its economical advantage.

However, again, this is minor and I don't have strong feeling about it.

Just collecting feedback on that. Do you find sensor bonus a better replacement for territory bonus? In my mind they are fragile and they also not inherent - require some efforts. Moreover, sensor is incompatible with second level improvement so may be difficult to find a good spot for.

Requiring effort is the point, as bvanevery says. You make the decision to invest former-turns to shore up your defense. This is what any strategy game is built on.

I recall you argued before that sensors are too easily gotten rid of by an attacking force. Allow me to offer the counterargument that this is also the point. Having to divert aircraft and artillery to attempt to bomb sensors with 50% chances means they are not being used to directly attack enemy forces. More effort, more investment, more turns, a higher bar to clear for a successful attack. Small advantages matter.
The same argument also applies to using probes to disable Perimeters. Far from being an easy fix, it requires investment and effort on the part of the attacker.
Admittedly, with the AI's use of defensive probes being unreliable, it does somewhat make sense to disable this ability for singleplayer games.

Also, they are not so easily gotten rid of if you place defenders on top of them. Even better, put them inside bunkers or forests with the defenders. Now they can't be gotten rid of without assaulting the bunker.
I believe you can also build a sensor and then place a base on top of it to retain the bonus, but this is definitely an exploit.

That is a good point and a lot of defense strengthening do require efforts: defensive structure, bunkers (when they are enabled), removing rough terrains in front of bases, and most of all building more units for that matter! 😋
The problem with them is that they require very complicated micro management and micro tactics to make them work as you just described. And that what I don't like in such simple 4X cost-benefit paradigm game. All features are more or less stand alone. Player make the determination whether some feature is beneficial for its price and, if yes, it pays off back in more or less consistent manner. Of course, similar effects may increase each other to extent but that is also a part of "feature is beneficial for its price" deliberation. Like having both PD and TF makes base much more protected than only one of them. Features are never 100% interdependent. Imagine that Genejack Factory and Children Creche work only if they are both built in base and not each one by itself. That would be unnecessary complication too much for a game.

Same story for sensors. Their usage is too complicated and too dependent on many other conditions comparing to other defense means those just work out of the box: armor, base bonuses, defensive structures, terrain, bunkers, territory in WTP.

As for territory bonus, as I said, I don't mind reducing it to 25%. Need to try it.

Let also look at it from this angle. You said it is possible to protect sensor from destruction by placing it at base or bunker tile. Now let's think further in this direction. Would player do this for all bases and bunkers being in a clear mind and capacity? Well, of course! So they would preferably end up having all their bases and bunkers extending 5x5 extra 25% protection around them. Then what the hell? Just give this bonus to bases and bunkers and be done with it. Saves tons of mouse clicks and also frustration when you forgot to put it under the base and cannot correct it after. Sheesh.

[Alpha Centauri Bear]

Possibly, there is no value in making peace with neighbors.  They will just deploy the AI's overwhelming colony production advantage, smothering you.

Did you try conquering their thinly stretching empires? People say with faster expansion AI does not build enough defenders. Maybe this is the key?

[bvanevery]

Did you try conquering their thinly stretching empires? People say with faster expansion AI does not build enough defenders. Maybe this is the key?

No, I "bigpoxed" and the "recycling tank" production strategy simply doesn't work.  I had the right idea putting a Command Center next to the Borehole Complex, but I need to radiate cheap units.  If there's any solution at all.

No commercial 4X game spams cities like this.  I've played a lot of them over the years.

Well, at least I tested the hypothesis of Recycling Tanks + Children's Creche being the golden path.  They aren't.

My concern is that if I do find a solution, it's going to demonstrate that the AI is a one trick pony and a house of cards.  That the only thing it's got going for it is this fixation on spamming colonies.  Which sucks rocks for the player if you don't know how to react to it, and is probably going to be a complete pushover if you do.

[Nevill]

I never could understand why everybody thinks Yang is aggressive with its GROWTH (!) and INDUSTRY (!) bonuses as well as its free Perimeter Defense (!). It is a well protected economical paradise - an opposite of an aggressive faction that conquer to survive.

It's his personality (which is an actual parameter in faction files), and the way the AI plays in vanilla. Yang is the only one who can go toe-to-toe with Miriam when she goes on a crusade.

People would be upset if Gaia started drilling boreholes left and right, and so Yang not raising an army raises eyebrows instead.

Note that I talk about relatively early game, the 2150s. Yang has a lot of bases, but all of them are protected by scouts. He has plasma armor, he just isn't using it. And he is in at least one vendetta.

Can you tell me which parameters answer for expansion? I noticed three parameters:

expansion_factor=100
ai_production_expansion_coverage=20.0
ai_production_expansion_priority=2.0

Are there any more that determine when AI needs to expand? It's pretty good for times of peace, but they need to better react to hostile environments.

[bvanevery]

Details Morganic fiefdom

Morganic fiefdom

Version 166.  MY 2166.  I quit the game at 4:45 AM on account of fatigue.  I played as a random faction, and with the Pirates in the game, so that I would not get them.  Also because that would have 6 factions on land instead of 7.

I started in isolation and pursued essentially the same bigpox strategy.  That seemed to be the natural flow.  The difference is this time I emphasized developing minerals.  Much as with the Pirates, minerals are the only things that count.  Farmland produces such a huge quantity of food that one hardly needs to worry about nutrient specials.  Energy, I doubt it matters.  For some reason I had lots of energy this game, much more than I used.  I don't entirely think it's lack of competition for supply pods, although that contributed.  I did not go Free Market for fear of mindworms.  On the last turn, I had quite the horde coming for me, and it was annoying.  I had Trance Scouts to wear them out, so it was merely a hassle.

I'm having doubts that a Recon Rover can terrorize much of anything.  I have a sinking feeling that the +50% friendly territory defensive bonus is going to nullify any attempt to rush.  I suppose they could pillage.

Starting far away from the AI factions is definitely less hectic.  I'm undecided if it's a strategic advantage.  For instance, I wasn't close enough to anyone to start the war I had in mind.

[Alpha Centauri Bear]

I never could understand why everybody thinks Yang is aggressive with its GROWTH (!) and INDUSTRY (!) bonuses as well as its free Perimeter Defense (!). It is a well protected economical paradise - an opposite of an aggressive faction that conquer to survive.

It's his personality (which is an actual parameter in faction files), and the way the AI plays in vanilla. Yang is the only one who can go toe-to-toe with Miriam when she goes on a crusade.

People would be upset if Gaia started drilling boreholes left and right, and so Yang not raising an army raises eyebrows instead.

Note that I talk about relatively early game, the 2150s. Yang has a lot of bases, but all of them are protected by scouts. He has plasma armor, he just isn't using it. And he is in at least one vendetta.

Heh. Now I understand why everybody think he is aggressive. Because it is set so to be.

However, now I think it is a mistake for him to be set like that. They have no inherent bonus supporting that. They should be better off growing large population and building infrastructure sitting behind PDs which allows them to have half of usual defense from start.

Can you tell me which parameters answer for expansion? I noticed three parameters:

expansion_factor=100
ai_production_expansion_coverage=20.0
ai_production_expansion_priority=2.0

Are there any more that determine when AI needs to expand? It's pretty good for times of peace, but they need to better react to hostile environments.

expansion_factor=100
Thinker parameter used to determine base number limit. Should have less or no effect anymore now when production choice for colonies is largely overridden.

ai_production_expansion_coverage=20.0
ai_production_ocean_expansion_coverage=40.0
An expansion demand calculation parameter. One colony (en route or in production) is demanded per so many populatable tiles around the faction bases. Populatable tile is the one not claimed by others, not within other base radius, etc. Each tile weight it is also reduced by distance to nearest own base so closer spots have highest demand impact while the rest of the globe is not that much. Actually, now I realized I need to reduce it quadratically (1 / distance ^ 2), not linearly (1 / distance) as I am doing now as the number of tiles also linearly grows with distance. A fruitful discussion! 😁
The demand goes between 0 and 1 depending on number of colonies already there (en route or in production) those partially satisfy it. No more colonies are built when player already have so many to cover all the populatable tiles around.
This does not define how dense AI will be placing colonies. Merely the demand for them. Ocean usually being less populated and easier accessible has lower demand.
Actually, this is one of the models you (or others) may help me with. I was trying to compute the top number of colonies (land/sea) that faction needs to not overproduce them. The above formula may be a little crude. However, this is just heuristic. Should not affect AI much even if one colony is mistakenly left unused at the end.

ai_production_expansion_priority=2.0
This the multiplier helping colony production beat other production choices. So with this value it converts demand scale (0-1) from above into production choice priority scale (0-2).

I am currently playing with the last one to see if it reduces spamming to the level where they build something else as well. The coverage parameter may need some tweaking but it logically does not control the urge to build colonies. Merely the potential need for them.

[Nevill]

Heh. Now I understand why everybody think he is aggressive. Because it is set so to be.

However, now I think it is a mistake for him to be set like that. They have no inherent bonus supporting that. They should be better off growing large population and building infrastructure sitting behind PDs which allows them to have half of usual defense from start.

Not just that. Hive bonuses in vanilla lend themselves incredibly well to ICSing and the Zerg playstyle of creeping spread. They can't pop-boom, but they have free bonuses from Planned and Police State, which gives them +3 Growth, +2 Industry, +2 Support and +2 Police. What it translates into is a lot, A LOT of bases, set no farther than 3 tiles apart from each other to be able to move/rehome scouts that work as a cheap police option, and share terraforming between bases. Since they only get credits from the tiles they work, it is incredibly important to settle bases on rivers and never let forest/solar tiles go to waste. And when you get 10 minerals refunded from your 24 minerals colony pods, you want to spam those. It is not rare to see chain colony pod production, with the new base starting on a new CP right away while a police scout is retransferred from another base.

Which results in tons of small, underdeveloped bases with no infrastructure at all (there are no drones, why bother?), only busy with producing colony pods, police scouts and formers... and occasional rec tanks. Hive can also afford to skimp on defences as police scouts behind perimeters are not easy to bypass, and the sheer bumber of closely-positioned bases prevents blitzkrieg conquest.

What this also means is that while they are underdeveloped, Hive is unsurpassed at producing a horde of cheap, untrained units and swarming its neighbours. A size-2 base can produce 5 minerals on forests, which allows it to build a 2-row unit in ~3 turns, and a 3-row unit in ~4 turns on Wealth. And with how many bases the Hive have, it's a lot of units.

Hive can completely dominate early game, and gets a second breath with a fusion reactor which makes everything cheap in vanilla. That's why people think it's aggressive; because it plays best as a conquest faction, not a builder one. It's only if they fail at murdering their neighbours, that it's time to try a different tack and build up.

Take away cheap units, and yes, a builder is the only thing you'll have left. But that is a departure from the original behavior.

[bvanevery]

Details sea exertion

sea exertion

Version 166.  MY 2231.  3:15 AM.  This is the point at which I quit.  Once again I played bigpox against the smallpox.  I had a perfectly viable empire, and I actually managed to produce some untrained sea units without any Genejack Factories.  The Cult of Planet was the 1st hostile empire to come near my territory. 

I found that my own Isles were completely useless against their units, but they could pillage.  I pillaged the heck out of cities I eventually took over, which was a mistake, and a big part of why i finally quit.  I found that my ships could blow away their Sea Formers without much difficulty, so it would have been better to go around killing those.  As is, they rapidly rebuilt everything I pillaged in their back waters.

Their Isles were surprisingly resistant to being killed, perhaps because my ships weren't trained.  Eventually I wore them out by sheer brute force, and then they only had various kinds of Scouts on defense.  Those are extremely easy to kill with Chaos guns.  The Cult put up no real defense.  Sometimes it would manage to get an artillery piece built, but we'd duel for awhile and then finally I'd kill it.  Most of the Cult's defensive value was the sheer bloodymindedness of their friendly territory bonus.

Shortly before I quit, someone framed me for doing something to my ally Domai.  The dialog box said he declared Vendetta on me.  I forgot to check whether that was true.  I tried to talk to him but he wouldn't receive my hails.  Although I had my cities sufficiently garrisoned, having a sudden front with him annoyed me.  This is the other part of why I quit.  A frame job like this is highly unusual.  I've hardly ever seen it happen.

Allies don't really fight for a damn, unlike in the stock binary.  At least in stock, they might contribute to taking over some bases, if you soften them up first.  I think all the AI factions are too busy spamming.  It's super obsessive, and not convincingly effective.  Granted I quit, but now that I know what's going on, it doesn't look difficult to wipe out sea bases.

Another thing that annoyed me, is the Cultist colonists that escaped on land.  They just founded new cities, which means yet more work for me to clean up.  It would have been better to bring a mixed force to prevent that.  However this also implies controlling the land, and the geography was such that that never happened.  In a sense, I think I overestimated the Cult's ability to bother me.  I didn't accomplish much, slugging away at them.  But I did steal a fair amount of tech before I caught up.

I made no SE choices for quite awhile.  Then when I was finally battling the Cultists, I went Fundamentalist.  Later, I went Green.  That made my people a lot happier and was a good move, despite the loss of growth.  By then I had Children's Creches in most places so it wasn't so bad.

I swept up an awful lot of supply pods.  I believe I accumulated 10 Artifacts, which I didn't use for anything.  Other factions seemed subpar at sweeping up land supply pods.  The seas did get emptied though.

Too many sea bases and no real defense of them, seems like a problem to me.

[Alpha Centauri Bear]

Framing happens to me too just not that often. Not even every game. I guess computer just randomly decides to frame then it randomly decides who then it randomly succeeds at this. That is why it is so rare that you get framed. Otherwise, it is pretty common situation.

[Alpha Centauri Bear]

https://github.com/tnevolin/thinker-doer/blob/master/wtp_changelog.md#version-167

Some (relatively) minor changes from development point of view but they may affect AI behavior, of course. I don't think drastically, though.
Want to push these out to free my mind for other pending issues.

Also few words about these three priority parameters.

Vanilla and Thinker use deterministic if-else approach to suggest production. That is simple but difficult to tune as tiny modification in condition may just flip to another option without much explanation to the designer. Vanilla being even worse as its code is in ASM.
WTP uses a standard artificial intelligence design when each choice is given a priority and the highest one wins. This way the competition is easy to debug and, more importantly, each option can be evaluated in isolation using its own algorithms, etc. It works fine so far. However, I like to mix it up with Vanilla-Thinker choices too to 1) introduce my changes incrementally and 2) dilute WTP choices just in case they are not very nicely tuned so AI behavior doesn't change abruptly between versions.

Unfortunately, vanilla and Thinker do not provide priorities to compare with. Just a single suggestion as I stated above. Therefore, I have arbitrary assigned priorities to them for my own convenience.
Units are medium since I didn't work much on unit building and need legacy code do that.
Projects are high since I don't have project handling code yet. So I trust legacy code on that.
Facilities are low since I have done some facilities already and want them to take priority if I decided they should be built.


I am also thinking on calling in any AI designer on reddit. Not that I cannot do anything without them but this may speed up things and let some talents to be put on use.

[Alpha Centauri Bear]

https://github.com/tnevolin/thinker-doer/blob/master/wtp_changelog.md#version-170
Enjoy!

These are new popups. I've created text myself. Feel free to suggest more catchier text if you like.

The formula for probability is linear as everything else in SMACX.

Code: [Select]

exposing probability = 0.05 + 0.01 * PROBE

Numbers are configurable. With numbers above -5 PROBE never exposes a spy. Although, this is quite an extreme value for this effect.

This works well with one caveat: probability. It may vary in pretty wide range. Once a 20 times it could just work on turn 1. Or it could sick for 40 turns sometimes. PROBE rating affects this but random fluctuations are wider than average PROBE effect which would be difficult to notice. People don't like it.
Nevill proposed a fixed turn counter that may be a more deterministic alternative to it but it requires to put it somewhere in saves which is quite finer work. So please test this one first and see if it is satisfactory.