I agree, and it'd also be nice to corraborate my own changes - see if they're the same. Perhaps you could PM us it.

Incidentally, started another "real" game with the Yor switched to Generic AI. They're ranked #2 and are one of the strongest civs on the map. I'd say they're done.

Unfortunately, I have not record most of the changes I made to AI values and categories. If I did, I think my changes would extend far beyond four pages. I will say, though, that I did give heavy consideration to everything said here and in other threads on the forum. What changes I did record can be found here: https://forums.galciv2.com/402576 (note: it is up to day as to the time of this post, but may quickly become outdated)

Also, In my latest game, the Iconians have researched Expert Engineering in my game. They have even researched Seekers III, in spite being in a bad starting position. So I say the Iconians are mostly fixed in my game. Maybe only a few more tweaks.

Hey all,

Another thing for the Korx: they will research Super Trade Starbases if you switch the category from starbase to economics. Which leads me to comment that, in general, the ai players seem to undervalue starbase upgrades. I would be interested to see if AIs would start building powerful starbases if they were categorized according to what they benefit (military, economics, or culture), since anything in the starbase category seems to be ignored until there's nothing else to research. Evil races in particular might benefit if they researched more starbase upgrades, since the upgrades are free after Xeno Ethics. 

Strange. In my games, the Korx always end up with the Super Trade Starbases from the start, as well as any other techs with a trade bonus. I figured it was a bug related to the Super Trader Ability. Did you not have the abilities in your games?

I'm working on this problem alongside the general neglect of culture techs.  I just went through and boosted the AI value on all starbase, defense, and culture/diplo techs to see if I can get the AIs to branch out a bit more.  Overall, all of those techs have quite low AI values.  My last test with subtle increases to culture and starbase techs did result in some of those techs actually being researched.  Some AI starbases actually had some real weapons and defenses on them.

Just imagine, AIs that make use of the entire tech tree.

I just had an interesting experience with the Iconians in a campaign level I was playing with my mod. They researched all of their economic techs much more quickly then I had ever seen. Once more, they even researched Organic Hulls and Expert Engineering within the few years that the level lasted. (No medium hulls, but the level was over fast)

Sorry for the double post, but I just wanted to add that I just played a normal game, and the Iconians researched their second econ tech fairly quickly! They didn't research their Organic Hulls and Expert Engineering techs, but that was probably because I cut the game short (was being overwhelmed by the Yor. Would have lost anyway), and because in the campaign, they were in a constant state of war, making weapons and hull techs a higher priority. I wasn't sure it could be done, but I think I fixed the Iconians!

Running into interesting circumstances now.  Boosting various techs has indeed caused the AI to pursue a more balanced research strategy.  Thing is, there is only so much time available for general research before militarization starts.  Once the arms race starts, it's on until the end.  Only after exhausting the chosen branch of weapons tech will the AI return to a more balanced strategy and finish up with important research.  There is a bit more balancing to do to ensure that important techs get researched.

• My setup ensures that Xeno Comms and UT get researched quickly by almost everyone.  They quickly follow up to Diplo I, Trade, and usually Alliances on their own.  These are important techs for a balanced empire, so I'm not keen to downgrade them.  Except Alliances, see next.
  
  • Even after I gave Alliances away for free, alliances won't form until year 7.  So the AI value can be dropped a bit.  I already dropped it some, which helped, but everyone who can will usually get it by year 2 or 3.
• I tried to delay militarization by dropping the AIV on Space Weapons and every weapon tech, with no real success.  The race starts from early year two to early year 3.  See next.
  • In the interest of uniformity, I have all races on AI 8.  This may be affecting militarization, as AI 8 is a bully.  It's expansionist and militaristic, which I think is part of the reason Thalans and Terrans are usually such beasts.
• Under the new setup, Extreme Colonization is no longer researched except by the most powerful and advanced races.  Even then it's in year 5 and only 1 of 3 did so.  The Torians, however, do research Aquatic World Colonization at a decent time.
  • Extreme Colonization is now at AIV 100.  Even such a high value can't get the AI to prioritize Colonization techs until quite late.
  • This also stops the Arceans from researching Weather Control, since it's a Colonization tech.
  • The Torian tech tree has Aquatic Colonization without the Extreme Colonization prereq.  Since each specific tech has its own category - eg AquaticWorldTerraforming, or something close - it seems that Extreme Colonization itself is the important roadblock and the Colonization category is the issue.
• Rather than try going even higher with the AIVs, I think it may be time to trick the AI into researching Extreme Col by changing the category.  I'm not a fan of changing categories to manipulate the AI, but in this case it seems necessary.  I'm thinking Biology, since it's somewhat appropriate.  Also the AI will reliably research Biology techs, which seem to be mostly population growth types.
  • I think I'll also give AIV one last shot, this time going to the extremes.  Maybe 150 or 200 or something.  I wonder what the top number is?

More to add, but no more time.  I'll come back with discussion of weapons techs, soil enhancement, defenses, and starbases.

Strange. I have never had any problems with races researching the colonization techs, and I have AI values much less then you. (55, for most). I also don't see many of the problems you have mentioned. For example, the Arceans researched Weather Control by the third year in my last game, at least (I didn't meet them until then, so I don't know exactly when they researched it).  I think it might be AI 8 that has a problem. That's the only thing I can think of.

Thanks for reporting your findings, Tolmekian. Meanwhile, I haven't had much success getting the computer to construct powerful starbases. I got the Korx (on ai 11) to research StarbaseVictoryStrategy and StarbaseSupremacy by increasing the aivalues to 50, but they didn't build any starbases so they got no benefit from it. I'll try this with ai 8 next, since the Terrans and others who use personality 8 seem to build lots of economy starbases. 

Thank you.  It's nice to get feedback.  I'm going to keep posting on my work, since it gives me a chance to collect and present my thoughts.  It's also nice to share and get ideas.  My next test is going to be with no changes other than using all AI 11, just to see.  Oh, that and I did create a bunch of custom races and gave them the exact same stats.  That way I can compare the tech trees against each other without having the same races rise to the top every time.

Weapons

• I dropped the AIV of Space Weapons from 50 (120 in some TechTrees) to 15, left the theories at 10, and dropped the weapon all from 10 or 8 to 5.  Didn't seem to change much.  Basically, when the AIs decide it's time to militarize, they grind out weapons techs.  This led to an interesting observation.
  • The Arceans stopped researching Planetary Fortification.  I had them researching it by just changing the category from Planetary Defense to Military.  Now, they don't research it even though the AIV is up to 45 or 55.
  • They do, however, research Fleet Battle Tactics, Tulon Weapon Focus and their unique Advanced Fire Control and Advanced Battle Stations techs - even though I dropped the AIV to 3.  They're all the same category - Military - but they're red (weapons) techs.  Planetary Fortification is purple (defense).
    • My hypothesis, based on this and another case, that the ID of the tech is important for prioritization.  It's easy enough to test by changing Planetary Fort. to WeaponsID.
    • The second case involves the Korx techs Mercenaries and Cut-Throat Competition.  Even with the same Category, AIV, and prereq, there is a big difference in how quickly and how often they are researched.  They always go for Mercenaries (orange) fairly quickly, but Cut-Throat (yellow) lags behind. I'll also be testing this by changing Cut-Throat and Fortified Freighters to orange techs - IndustryID.

Defenses

• I jacked up the AIV on all defense techs to encourage research.  At 20, not much really changed.  At 30 the superpowers dipped into the defense line, but after considerable weapons research had been done.  Not enough to be useful.
  • I wonder just how a game with lots of defense would even play out.  Are defenses even practical up to a point?  Just for fun, I may change the defenses to WeaponsID just to see if they get researched faster.  The idea would be to outcompete the weapons themselves, I guess.

Soil Enhancement

• Not much to say here, only that they don't get around to researching it until very late and very powerful.
  • At first I lowered AIV, so young empires wouldn't get stuck upgrading tiles on empty planets.  Now it's too late.  Time to work for a middle ground.

Starbases

• Upping the AIV got the AIs to research a few of the early techs, but not enough to make a difference.  It doesn't help that Advanced Starbase Construction (the first tech in many trees) is just a spacer tech, without any benefit.  It's hard enough to get the AI to research the first tech, let alone the whole line. 
  • I think I'll have to add in some basic starbase modules that use ASC as a requirement just to reward the AI for going down that path, and also to see if they use them.  I like the defense modules that TOV has in his mod.
• I did have a game where an AI or two got a few starbase techs, and even Terror Stars, but didn't seem to do anything with them.  It was too late to be building constructors, apparently.
• AI 8 may help, because it goes all out for the mining starbases, which it will then upgrade to fortify them.  AI 7 also builds plenty of starbases, particularly military ones, but is too broken to playtest without jumping in to help with colonization - which I may do on a future run.

Colonization

• Speaking of AI 7, I accidentally set my Meklar (Yor tech) to AI 7 during a test.  By chance they had enough planets within their influence to carve out a small empire, maybe 13 worlds.  They stopped colonizing and built a number of starbases, but didn't upgrade them much.  Interestingly, they were the first and only to research Extreme Colonization and even went on to research the right techs to allow them to colonize the only two extreme planets under their influence - one toxic and one radioactive.  I always wondered if there was any intelligence behind choosing extreme colonization techs.  It seems there is, at least for AI 7.

Long post.  Thanks for reading.  Now to work some more testing in along with some actually play time on my AAR game.

On the note of the Korx, in my last game they actually researched Fortified Traders (after Cut-Throat Competition) BEFORE Mercenaries, which I had never seen before. Also, the Korx have researched Starbase Victory Strategy in my last game. All I did was make it military and raised the AI value to about 11 or so, and maybe a bit of luck.

I think the AI's unwillingness to research the transforming techs has more to due with the lower AI value. I didn't touch the AI value for it (or if I did, I raised it), and they research the techs just fine. Faster than me, more often than not. (It doesn't cripple them in my games as I lowered the build cost of all three)

In any case, I find your situation puzzling, and I am eager to here more about how you attempt to solve it.

Well, my test with AI 11 turns everything on its head.  To keep it short, a game full of 11s was a joy to watch.  Nice, balanced research with no weapons blitz.  Good defenses, good starbases.  I must agree with an earlier statement by qrtxians that AI 11 seems to be the best war-fighter. 

Issues would be that now I think they research Extreme Colonization a bit early - late year 1 to early year 2 - and they don't reseearch engine techs at all.  I only gave the bare-bones starting techs for each race, so Hyperdrive itself could be a roadblock.  A little tweaking and I could call it done - for AI 11 at least.  I don't know yet how I'll deal with the issue of alternate AIs.  It may come to tailoring specific tech trees for specific AIs.

Just wanted to add here that a game where defenses outpace weapons isn't very interesting if you like watching the combat viewer. I have actually encountered a draw (link: https://forums.galciv2.com/383289) after a battle that lasted for 300 combat rounds. Until I started using the original Space Weapons Mod, high-defense was common, although I never saw the draw again. It's probably best for weapons and defenses to, at most, stay on the same level.

I think I'm at a good place based on my last test.  Weapons and defenses were keeping pace, with most races in the first or second tier of weapons and defense.  The warfighting was intriguing, but painfully slow because nobody had any engine techs.  Watching wars unfold at 2 parsecs/week isn't very exciting, to say the least.

There was one interesting case of poor defense.  The Trilarian (Torian) saw the early Sakkra (Drengin) ships with the Space Weapons particle beam, so they developed shields.  The Sakkra, and everyone else, went on to develop mass drivers and armor, but the Trilarian stuck with shields.  I wonder if they would ever go for armor.  I think it would be the only reason for me to continue my slow-war test.

I did actually see a couple military starbases with assist modules.  I got to looking at the starbase module xml and found out why Battle Stations 2 and Protection Fields misstate their actual bonuses.  Yet another minor thing to fix.

I did actually see a couple military starbases with assist modules.  I got to looking at the starbase module xml and found out why Battle Stations 2 and Protection Fields misstate their actual bonuses.  Yet another minor thing to fix.

This is the first time I heard about this (that I remember). Would you care to elaborate on what you mean by this?

Hey all, I can now confirm that if you change the branch id for the starbase line of techs (starting with advanced starbase construction, through all the starbase fortification upgrades) from 'defenses' to 'weapons', at least a few of the AI opponents will much more eager to research them. This was true even though I left the aivalues at 5. I gave this treatment to every race's tech tree.

I started a 10 player large map, did a ctrl+z and let it run until year 2233. At this point, the Terrans have some very powerful starbases. For example, there is an economy starbase with production upgrades and an attack of 10/40/40, no defenses; they also have a morale resource mining station with attack 46/46/41 and 6/20/0 defense, plus mining upgrades. This is the first time I've seen defenses that strong on an computer-controlled starbase. Although, as usual the computer is being too enthusiastic about pursuing depth before breadth. The terrans may have strong starbases, but their ships are pretty wimpy.

Looking at the other races, the only other race that has built comparatively powerful starbases are the Drath, who like Terrans, are controlled by AI personality 8. The Korx (on 11) have researched some of the starbases fortifications, but only have one (culture) starbase, with 1/1/1 attack and some culture upgrades. The Altarian on 11 and Drengin on 7 likewise have a couple 1/1/1 starbases. 

For Battle Stations 2,3, and 4, it looks like they tried to add defensive bonuses, but never got around to testing it.  The code looks something like this.

StarbaseAbility - StarbaseAttack

AttBeam - 2

AttMissile - 2

AttMassDriver - 2

DefShield - 2

DefPD - 2

DefArmor - 2

That's obviously not the code, just enough to get the point across.  The game reads the ability as StarBaseAttack and assigns the values.  Problem is that it doesn't know what to do with the Def values.  What you get is +2 attack in all categories, which works fine. The starbase summary, however, says you get +12 attack, because it adds the defense numbers even though it doesn't do anything with them.

I did try sticking StarbaseAbility - StarbaseDefense in the middle to see if it could handle it, but it just uses the first StarbaseAbility line and ignores the second.  So I just removed the Def values from the xml.

Protection Fields has a similar issue, but a little different

StarbaseAbility - DefenseAssist

StarbaseAbilityValue - 1

DefShield - 1

DefPD - 1

DefArmor - 1

I forget if the code for an assist value is Def or Assist or whatever, but you get the point.

The result is +1 shield assist, but the summary adds up all the numbers and says +4 defense.  StarbaseAbilityValue isn't used with any other Assist, Attack, or Defense module, and really shouldn't be here.  The name and description of the module imply that it's a shield bonus, so I removed the ability value and the other two assitst categories.  That leaves +1 shields and a proper summary.

A few remarks on AI:

It's probably best to leave AIs with a default personality of 8 on 8, even if you have all the other AIs set to 11. I also find it's not that good an idea to take the Korath off their original personality, as it's the only way they'll really make use of spore ships.

There are differences between individual race personalities that share the same AI, as well. It's hard to pin them all down, but besides the Korath and spore ships, I've confirmed that the Drath personality is designed to buy large numbers of foreign ships, many more than the Terrans, Thalans, and Krynn do.

Whew.  After many hours of modding and testing over the past several weeks, I'm pretty much ready to declare my work done.  The TechTree itself is fixed, along with the TechTrees of the Arcean, Drath, Drengin, Iconian, Korx, Torian, and Yor.  Minor tweaks could be done to suit various tastes, but each tree allows for balanced research spread across the entire tree.  Overall, I would say 95%+ rating.

Final notes:

• Balancing and final testing was done with custom races set to AI 11 "generic", all with identical bonuses and no Super Abilities, in a gigantic galaxy.  Could the balanced, long-term approach fall apart in a small galaxy?  What about different AIs? We already know that AI 8 will use the tech tree in a very different way.  In any case, this is an excellent starting point for customization.  Much better than the original TechTrees, in my opinion.
• The AI just doesn't do a whole lot with starbases, for the most part.  At least not AI 11.  They don't usually research past Advanced Starbase Construction/Starbase Victory Strategy, and I'm not inclined to force the issue any further at this point.
• They do, however, tend to go for the Terror Star techs after ASC or SBS.  Since they don't seem to use Terror Stars and the techs have no other value, I'd like to curb this.  I already have the AI value as 1.  I set the fleet attack techs to 0 and that seems to slow research significantly, so that's the next step.

There may be something else, but I'm largely satisfied.  Now for the Altarian, Korath, Krynn, Terran, and Thalan.  Hey, I thought there were more.  I can actually see a light at the end of this . . .

More notes on the AIs:

Before starting all this modding, the game was always dominated by the Terrans, Torians, and/or Thalans, with the Krynn, Altarians, and sometimes the Korx usually also successful. The other races were rarely, if ever, major contenders.

Following the modding, the Altarians are now up there as one of the races that always does well, and the Drath, Iconians, and Yor are now in the ranks of AIs that can succeed. The Arceans and Drengin still aren't very good, but they've had a few successful games and I'm still working on them anyway. The only race that hasn't put in a good game now is the Korath, who I'm not finished with yet.

Although the "big 3/4" will still always do well, it's come a lot closer to the "anyone can win" feel of DA.

One thing I've noticed with the Arceans is that Weather Control seems to delay Extreme Colonization to the extent that they don't get in on the extreme planets race.  It's not that they don't research it fast enough, it's just that Weather Control and Extreme Colonization are expensive techs.  It looks like they are getting priced out, essentially.  My Drengin hold their own fairly well, but they aren't one of the top races.  A bit lackluster overall.

One of the best things I'm getting is simply exploring all the tech trees.  I had no idea how diverse they were.  I didn't know at all about Mercenary Academies, Space Cannons, Biosphere Modulators, and everything else.

I noticed that Xeno Engineering, which is, surprisingly, required to unlock most of the higher mining techs only has an AI value of 5. It would be a good idea to raise that to at least 10.

More reports. First, in one of my games, the Iconians researched Self Healing Hulls! *Cheers!* As I said earlier, they are fixed.

Secondly, I had been trying to play with the evil races on AI 11, and while the Yor did fairly well with this AI, the Drengin and Korath did only marginally better. In my recent game, I experimented with AI 8, and the results are downright terrifying. The Drengin's military might is without equal, and they know it. Granted, most of that might comes from their super dominator ships, but the game is still early enough that they are not obsolete yet, and they are putting them to very good use.

Here is a representative sample of the Drengin Army. (Luckily, there after the Letzlandiands, and not me.)

It's nice that you've made progress.  Whatever changes I made after my initial mod died with one of my  hard disks.