Thanks for the feedback.  For upgrading structures, the xml files specify which particular structure is replaced by each new structure.  For example, Stock Markets specifically upgrade Xeno Banks and nothing else.  So even if, say, the Drengin manage to get Stock Markets, they will not upgrade their crappy old Black Markets.  That was a big problem that I fixed in v1.  The Drengin and Korath both start with Black Markets and Slaveling Imagination Labs. However, the Drengin had traditional economic techs and the Korath had traditional research techs.  Both ended up stuck with one of those bottom level structures that they could never upgrade. That was fixed by giving Korath economic techs and structures to the Drengin, and Drengin research techs and structures to the Korath.

I have noticed that the Torians are consistently weak while the Yor and Korath are out of control.  Like I said, I need to go back and start with an even field so I can compare the tech trees against each other.

Tol,

I've been working on my own little fix mod and I thought I'd share a trick I discovered. If you set soil enhancement and its followups to AI1 the AI places it last in an existing build queue. In other words, exactly how it works for us players. This helps them build up their planets a little better.

Also AIP8 doesn't really hate Category Biology it hates ID Biology but it really can't stand the combination. If you want AIP8 to research, say, exteme colonisation with a AIV of 55 set it to ID Culture or Industry and viola! Also I'm finding that to get the AI to research something Pure Research usually does the trick regardless of ID and usually with sub 20 AIV.

Regards,

Alt.

Well here's what I've noticed. The AI will build SP and GAs even if their AIV is relatively low. It seems to use these values to place them in the queue rather than deciding if it should be build or not. I'm also positive the AI checks what a building gives it and given the choice will build a building with a higher AI.

Consider this. The Thalan technology adaptation buildings all have an AIV of 5 while the basic lab is around 12. Still the Thalans build a lot TechAdaps UNLESS they trade for say Xeno Labs. Then they'll build labs not TechAdaps. Both give research but Xeno Labs win in the AIV department. Could be the costs, etc, come into play I'm not sure.

1pp buildings are considered normal and are queued or simply ignored depending on their AI value. If you want to get the AI to build these you need to give them a high value.

Also, I've been thinking about the Biosphere Modulator vs Weather Control Zeniths paradox. Why does the BM have a AIV0 while the WCZ has a value of 155? A Mistake? Possibly but I don't think so. Here the thing. If you give the BM a value of 155 the AI will queue one on every planet when it becomes available. Sounds great except that the Altarians can get it really early in the game (they get Xeno Myst so they jump on Precursor History and viola!). It is an expensive building and takes 15 to 20 weeks to complete on an undeveloped world. Not exactly the type of building you want to clog up your AIs build queues. The WCZ is available much later in the game because Xeno Geology is Research/5. That's AI speak for Ignore Me Until Bored Of Weapons And Everything Else. But it effectively prevents the AI from getting it to early. By the time they do I'll do little damage on their build priorities.

Interesting notes, Mabus.  I did very little fiddling with the AI values for structures, but one thing I did was to boost the Weather Control Zenith AI value to 155 (I think.  I don't have my notes). Wait, I have an idea. *goes and downloads v2 to look at his notes*

Okay, nevermind.  It seems the WCZ started at 155.  What I did was change the Biosphere Modulator from 0 to 155 so the Altarians would build it on every world all the time, especially on fully developed worlds immediately after doing Soil Enhancement/Habitat Improvement/Terraforming.  I think they didn't make full use of it before I changed the AI.  With my mod, they usually don't research divergent evolution until they are far enough along to deal with the burden of building all the Modulators.  What I did for the Arceans was to reduce the cost of the WCZ by 25% so they didn't get slammed as hard when it came time to build them all.  It's just too brutal to have all your planets spending 30-40 weeks building the same thing while everyone else keeps developing.

I'm glad you're working on this, too.  Please keep letting me know what you find.  I'd love to see how planetary development can be influenced and see if maybe the AI can be guided toward more efficient development strategies.

Should've done this years ago.

As a small note, I fixed the Torian Hull techs.  It was unintentional for them to be rooted in the Logistics techs.

Well, I notice that to some extent you've gone beyond just fixing the bugs of stock TOA, and are now modding the game to your states (e.g. colonizing minors). There's always the possibility that that's deterring people.

True, but that's why feedback is nice.  It's a bit surprising that the colonizing minors are such a big deal, given that I also release the mod without them.  Matter of fact, now that I've been testing with smart, non-colonizing minors, I think I prefer it that way.  My first game will definitely use them.

v2.1 will have at least three versions - dumb minors, smart minors, and colonizing minors.  Hopefully that's not too confusing.

That's a good point, Agorwal88.  To my recall, I transferred the weapons and defense costs directly from the old TechTree, to keep things similar.  Those numbers are way too far off, though, given the increased effectiveness of defenses.  I'll make some changes before releasing v2.1

This is exactly the kind of input I was looking for.  Thank you.

It says I'm new, but really I just forgot my old account info; I got GalCiv II ages ago.

After reading the whole Weapons Fix thread, taking note of how you four or so went about a systematic analysis, then reading all of this, and downloading it, I am impressed with how effective the AIs are.  I usually don't play very seriously and keep them pretty dumb, but even on "Normal" intelligence where they seem sorta dim-witted, they are scary.  The already noted (ages ago) fact that they actually bother with trade now was very satisfying, and with everyone on AI 11, the scheming is really creepy.  The Terrans convinced everyone to go to war so no one was left to defend the Arcean; had they conquered them, they would have had me surrounded on three sides with their other borders on the edge of the map.  Sneaky buggers.

I did find that they were fairly proactive with acquiring starbases.  Many did not get far enough to research meaningful weapons; I was playing on a medium map with I think normal difficulty and planets were a little short, but I should mention the Drath had INSANE weapons by the time I foolishly declared war on them to stave off an impending Influence Victory, and they only had two or three planets.  Most everyone else had a pretty lame military.

While I sorta agree that the trade bonuses should be higher, a whole order of magnitude is far too much.  My coffers were insatiable and every race had trade routes with everyone and their treasuries were ridiculously stuffed.  I'm just going to throw a number out on gut instinct, so 3 times the vanilla rates is as far as I would go.  But I wasn't playing with truly intelligent AI's.

Also, about removing the starbase module fees: while it may make the AI's bother with upgrading better, it alters the tradeoff of the modules.  That is, I can upgrade all three categories of attack for the exact same cost of upgrading only one category.  The costs are there to make specialization cheaper than generalization.  Without specialization, the rock-paper-scissors system of damage falls apart and starbases may become over-powered.  On the other hand, things like Battle Stations are pretty weak compared to the later specialized weapon modules, so maybe it won't have much effect to remove the cost on those; they are already pretty expensive to get at in the first place. 

Things like the Influence modules, though, make it too easy to build super influence bases; my instinct is that +367% influence should not come as cheap as the number of constructors it takes to get it.  To incur a higher cost to those on a per-installation basis that is not a road block for the AI, you'd have to cut them up more so you need more constructors to build them; Franchise Center becomes Star Mall and Franchised Marketplace, for example.  Take whatever the cost of a constructor is, and shave off maybe 25-50bc to represent the maintenance to get the constructor to the starbase and the general hassle.  Use that amount to recreate the cost without making the AI trip-up on installation in the first place.

That is, of course, if they ever bother installing the influence modules; my AI's only built them when they needed to counteract locally distressing levels of influence.

Please note that I am speculating to a large extent; analysis of what the AI's and players do with the starbase modules would be necessary if you want to find out if any of that holds water.  I know you've done much of that, but this is more of a play thing than an AI thing.

Interesting, and good to know.  Thanks.

There must be some other factor(s) at work, though.  My defense costs are 1/2 the corresponding weapon cost, yet the TP difference is 6 fold.  A 1.5x modifier should bump my difference from 2 fold to 3 fold.  Perhaps Defense ID techs have a 0.5x modifier?

As it is, I guess if I triple my defense cost then defenses should end up at 1/2 the TP cost of weapons.  Does that sound good?

I don't recall checking defenses-I haven't done much with TA in over a year.  It's certainly possible I missed that-or that there was another ninja update which changed the values again.

Since this is TA and you can see the tp for a tech by right-clicking it without actually having it available to research, can you verify whether that's the method you're using or if you're researching things first?  I ask because TA's inflation taking effect at certain thresholds rather than DA's marginal increase per could possibly be skewing the results if you're comparing pre-threshold and post-threshold values without adjustments.

I suppose so, if that's your intent.  I've no idea really what might be "good" values for them, though (see: haven't played TA, or GC2 for that matter, in forever); I was just trying to be helpful.

Edit: Uh, oops, don't think I can answer that question.  I had thought it was a strict 1.5x red tech multiplier, which means if you matched a defense tech and a weapons tech to be equal at game start, they'd be equal throughout, even after inflation increases, but it appears I have other data that suggests that the multiplier increases to 2x and even a bit more.

Apologies; as stated, I haven't messed around with TA in quite some time.

Additionally, if memory serves, the military starbase module tree is red as well, which means it also suffers from this.

Numbers are from turn 450-500(map size = immense) and about 3/5 of the research is done, and tech traiding is in use so plus about 20 unique techs from others.

When i start new map(same settings) numbers are weapons = 1085-3614-7236-10858 and defenses = 405-1355-2714-4070(with drath).

This mod is one of the best mods i have tried and increasing the cost of the higher end defenses is good solution. Thanks.

In 2.1 are smart minors also colonizing? Good be fun...

Wow.  Lots of great comments.  I'll tackle them one-by-one and try not to make too huge a post.

Thanks for your feedback.  Your comment on influence starbases prevented me from removing the ModuleRequirement from the influence modules.  Could you imagine just 2 or 3 constructors for a ~150% bonus? I'll keep an eye on the other stuff.

It looks like the conclusion is that inflation is hitting the red (weapons) techs harder.  Boosting up the higher defenses seems to be the answer.

As for smart minors - what I'm calling smart minors still behave like minors, but do so intelligently.  I currently plan to release 3 versions of 2.1 - dumb minors (no changes at all), smart minors (minors that simply perform better), and colonizing minors.

1 - I think the test will repeat.  I'll be doing another one in the next couple days.  The key is having high star and planet density so they can colonize enough planets to remain competitive.  The one pitfall is if they get stuck in a corner or someplace with lower local density, then they could get stuck.  Your AAR inspired me due to your tight clusters.  I never play or playtest with tight clusters, and seeing your AAR gave me the sudden idea to test those conditions.

2 - I changed the Category and AIValue of Spore Weapons early on, before realizing that AIP 11 wouldn't use them.  Actually, it was because AIP 11 wouldn't research them.  Now that I'm going back to AIP 7, I need to fix it.

3 - For smart minors, I simply left them as AIP 5, then cut-and-pasted that magic block of code that gives the AI its abilities: AIAbilites (50), CPUUsage (100), FinancialResources (100), and Aggression (50, or whatever).  While releasing v2 I noticed that it wasn't present for the minors, so I put it in on a lark.  Upon testing, it turns out to make a huge difference.  I guess the default value must be for minimum AIAbility and CPUUsage if the code isn't there.  Don't know about FinancialResources, but maybe they were getting screwed on that, too.

I had also added that code to the Dread Lords in v1.1.  I've never tested them with the mod, but my past experience tells me that they were pretty dumb beforehand.  If the change is as dramatic as for the minors, woe to all who get the Dread Lords event.

Thank you very much for this reply.  I can confirm your results because I noticed during my work that the Drengin simply would not make use of their special mining modules.  I had simply written it off as a bug I couldn't fix, but it turns out those modules were at or near the bottom of the list.  Since I added a handful of new modules, that means even more modules just got knocked out.

Offhand, I can think of one possible fix for this.  In the mod folder, there is a separate folder for starbase modules where you can put your own xml file(s?) that will be used to add modules to the game.  Perhaps it would be possible to break the StarbaseModules.xml into two or more separate files, both small enough to be completely read.

Now, maybe it's simply a problem with the AI being unable to deal with so many modules, so no matter how many lists you have it just won't look past a certain number of module or amount of code; but maybe, just maybe, breaking up the xml could work.

Looks like I have something to do tonight.  And so much for a not-too-huge post.  Sorry, but lots of ideas here.