Hey Bamdorf. Yes we're familiar with this one and it's pretty confusing. Fortunately +speed does work as intended.

We can't fix it ourselves though.

That makes sense. thanks.

Hi,

as far as you know, is this mod compatible with the high resolution mod by Logicsequence?

I would like to check the graphic mod, but I wouldn't if that means disabling CU mod.

I just started the DL campaign using this mod, and I'm enjoying it.

Thanks, it worked.

I'm afraid this is both not true.

But I've given you the benefit of doubt and repeated your test: I still saw the Yor build generic factories, although not as common as I've seen this happen in other games. I've even extended the tests in order to see if the AI follows your routines, for example by gifting him a generic factory that has 0 maintenance but 20 industry, ie. it's superior to everything else in the game. They aren't going to build it. Go ahead and see for yourself.

The truth is: The AI uses a number of selections to determine what to do on a planet but nowhere near sophisticated that you would like to have it.

One selection reflects his global needs, ie. he decides whether to build a farm, a bank, a lab or a fac.

The next step is to ignore all improvements you see under the "old" tab.

Afterwards he is going to make a selection from "newest": Usually there's only a single improvement there so this step is void in many cases. Afterwards the AI only has to rely on the upgrade chain.

But if there's more than 1 option present, the AI will make a random selection of which one he is going to choose, although the randomness will be influenced by the value of the <AI> tag of both improvements. It is essentially the same mechanism in place like the <AIValue> for techs.

Yor factories have a value of 35, the generic 15. How this is exactly calculated in the end I don't know, but this is the reason why you saw your results, as well as these results will approximately stay the same irrelevant of maint/industry output of rival factories. I use "approximately" here because the randomness dictates that the results could also vary - there's even a slight chance that these results are in total opposition to the rule. That's why it is simply wrong to deduce AI routines from a single test. If the game creates an array wherein it will role a dice between 1-50 where 1-15 = generic and 16-50 = collective you could have alot of different possible outcomes.

Nevertheless, we've also spoken of Thalans, repeat your test there: their T1+T2 factories are at AI-value 5. In my tests they've chosen the generic +6 ind fac over their +10 ind fac pretty consistently. As you can see the problem we've spoken about exists for real.

But simply increasing the Thalan value would neither be a clean or good solution. In a way the value of their factories (5 - 5- 20) makes sense, because if Thalans build alot of their own superexpensive factories it usually wrecks their game. So if they have access to the generic one they will initially rely on them, and once T3 becomes available (which is superior to all other factories) start building those. The problem in this scenario is that the AI is usually reluctant to overbuild low buildings with better ones if they aren't bound by the upgrade chain. Although I've seen this, but not really often. So in that instance the low generic factories will stay low industry and be the Thalans #1 factory to build for a very long time, because, for all I know, we've decreased the value for their T3 factories, didn't we? 

In the case of the Yor it's pretty onesided. Their facs are superior in output and maintenance, but if someone trades a better factory to them they are going to ignore it. It's still flawed although it doesn't happen that often.

Remember when we were talking about what techs should be made stealable? During that time I reported from a testgame were Drath stole Bartering Stations from Iconians and build them onto their worlds although they already had Stock Markets (!) - and they did it in a 50/50 occurrence. Backchecking on their values both are 30. That shouldn't be, Barter Stations are inferior to Stocks in every sense.

Basically there has to be some sort of value adjustment for shared improvements. The Ai value has to be linear to the quality of the improvement. The quality or worth of an improvement has to be based on 1) output 2a) buildcost 2b) maintenance. And unique+untradeable improvements need to see a bonus increase in value because they most likely will receive a more early upgrade.

For example if I compare the generic vs Yor fac, I'd put them into this order of priority:

Generic T1

Generic T2 --> better output

Yor T1 --> small decrease in output but less maint, less buildcost, increased chance of upgrade

Generic T3 --> great increase in production justifies for increase in maint & buildcost

repeat logic ad nauseum

Yor T2

Generic T4

Yor T3

Generic T5

Yor T4

Yor T5

However, it is still not going to manipulate the AI in building always the optimal factory, it simply increases its chance depending on how big the steps are. Considering there are alot of unique factories of other races which we'd have to integrate therein these steps cannot be too high otherwise the end-values will shoot through the roof.

It's sad that the AI doesn't have routines to determine "the best improvement" like you think he does, because then, this problem here wouldn't persist nor would the game need <AI> or <AIValue> tags anymore.

Another more crude way would be to simply disable the trade of these problematic techs. Or maybe there's another solution, I think I remember you or Gaunathor write about the possibility to create multiple upgrade-targets for an improvement? Is this possible without any negative side-effects? If Thalans would simply overbuild all generic facs once they got their own T2 or T1 that would actually settle the issue.

Although it brings me something else, I've encountered a bug in one testgame:

"Market Center" is still under the Newest tab although "Adv. Market Center" is already researched. Actually the upgrade chain seems to be broken. As you can see, the AI even builds the weaker improvement - it has the same value as its follow up improvent (yet another proof that the game doesn't autopick the stronger improvement.)

However, I wasn't able to recreate this. The only thing that I could see from looking into PlanetImprovements.xml was that the Minor starting improvements have the same name in <S_InternalName> as the original buildings. I know that doing this was responsible for other bugs in another sheet. IDK, maybe the game here takes the "Adv. Market Center" as an upgrade to the Minor Market Center...(?)

Could you be a litte bit more specific?

I'm pretty sure we cannot alter such behaviour from working on xml-sheets, but a screenshot could help, and a description what to do in order to create the info not updating.

BTW some stuff, eg. adding social production, only shows after opening the Finance Window twice.

Seeing how a lot of things with AIPs have been resolved, I want to discuss weapons/defenses one last time.

This is why I think weapons are imbalanced, it's a long post, so bare with me

All weapon tiers in CU except 1&2:
Cost - 1,0 : ~1,25 : ~1,45 ratio (Mass Driver, Missile, Beam)
Maximum possible damage - all are equal, with only Missiles being more effective on larger hulls (10-15% more damage on average).
Research costs: negligible difference (28k vs 33 vs 30).

If we exclude first tier of weapons, which is extreme and weapons don't follow any rule, and tier 2, which is the only semi-balanced tier in a whole tree (especially past Small hulls), weapons allow you to outfit ships with about same amount of damage. Examples:
Small hull, 0% miniaturization, tier 2 (excluded as is quite balanced):
- Particle Beam II: 6 damage total, cost - 175 bc total
- Harpoon II: 6 damage total, cost - 160 bc total
- Singularity Driver II: 4 damage total, cost - 100 bc total
Medium hull, 10% miniaturization, tier 3:
- Plasma II: 15 damage total, 355 bc
- Photonic Torpedo II: 20 damage, 360 bc
- Pulse Cannon II: 16 damage, 240 bc
Large with 30% miniaturization, tier 4:
- Phasors III: 65 damage, 1005 bc
- Photon Torpedo III: 72 damage, 880 bc
- Graviton Driver III: 60 damage, 660 bc
Huge with 45% miniaturization, tier 5:
- Disruptor III: 147 damage, 1895 bc
- Anti-Matter Torpedo III: 176 damage, 1760 bc
- Quantum Driver III: 162 damage, 1400 bc
Huge with 60% miniaturization, tier 6:
- Doom Ray: 288 damage, 3080 bc
- Nightmare Torpedo: 360 damage, 3050 bc
- Black Hole Generator: 300 damage, 2320 bc

As you can see, past tier 2, every weapon can max out at the same or very similar damage value, with the only difference between them being the price for a ship. Beams are 40-55% more expensive then Drivers, and are also weaker on several occasions. There are no differences, they do not introduce any fun mechanics, and boil down to picking strong, average or weak option (Mass Effect anyone?), unless you play on slowest tech speed (if you do play on slowest tech speed exclusively you can ignore this post, as you probably never reach weapon tier 3 or 4 frequently enough to care for any of this). Therefore choosing Beams is just like voluntarily picking -% military production. Problem of weak Beams is magnified even more, as shields are best at increasing defense, so already expensive and only average Beams are easiest to counter.

You may argue that all this brings more randomness and challenge to the game. There are a couple of issues with that, and here are main arguments against current weapons/defenses progression:
1. You hurt players who do not know about this. If they pick Beams their fleets will have almost same damage as if they picked Missiles of Drivers, but will have 30-50% less ships then Driver users and 10-20% less ships then Missile users, who's weapons are also 5-20% stronger.
2. You do not make game more challenging. On the contrary, if the strongest civ in galaxy picks Beams and loses to you when you pick Mass Drivers/Missiles, it feels more like cheating then outsmarting the AI. AI is hurting itself without even knowing it.
3. In vanilla, Beams are the best pick of any weapons. This is thanks to OP Disruptors, Phasors giving you quick access to Doom Ray and Beams being overall cheaper to research then Mass Drivers. In other words, current progression is against original design, which made perfect sense: beams cause damage almost instantly, their accuracy is higher then Mass Drivers and Missiles which can both miss a lot more, and in all Space Operas/Dramas/Movies shields and beams are primary weapons. They ought to be strongest. Mass Drivers used to be weakest, now they are a better choice then Beams, why? Only Missiles stayed more or less in same position.
4. Some tiers of defenses/weapons had different number of levels. Now everything has 3 levels. This makes the tree less interesting, especially defenses - some had 2 levels, some had additional tech in its tier (like Superior Duranthium etc.)... one tier of defenses got cut out completely (Invulnerability Field techs) for no reason, and tech costs/defense values for other shield techs got raised to artificially compensate - again, why?
5. Lost production even in case of cheapest ships is almost non-existent if the average amount of time to build a ship for the AI is anywhere from 10 to 40 turns.
6. Level 3 of lower tiers happens to be stronger then level 1 of higher tiers a lot, this was the case in vanilla on couple of occasions, but on a much smaller scale. Examples:
Particle Beam 3 > Plasma 1
Phasors 3 > Disruptors 1
Pulse Cannon 3 >= Graviton 1 (small difference in price, Pulse Cannon being more effective on small and medium hulls)
Graviton 3 >= Quantum 1 (same as above)
Stinger 3 > Harpoon 1 (on small and medium hulls with 0-10% miniaturization).

Sparrow being better then Stinger I on small hulls if miniaturization is lower then 20%.
Maser being better then Laser I and Kinetic Stream I on small hulls if miniaturization is lower then 20%.

And probably some others I missed. 

This is a problem for the AI who upgrades its ships a lot - it pays enormous amounts of bcs for ships that can end up being weaker then what it already has.

Other stuff that goes against original design:
Ion Cannons - were much smaller then Railguns - and even slightly smaller then Lasers. Not only their size when compared to Railguns, but also size mod had lower value.
Kinetic Streams - while still overpriced a little, they were even smaller in vanilla, and this was the only reason why they were expensive - ships with KS3 were almost as strong as tier 2 level 2 (Particle Beam 2).
Seeker Missiles - used to be stronger then Stingers, not weaker.
Ion Beams - were larger then Plasma.

Gaunathor had a good idea when rebalancing weapons. Unfortunately he did it the wrong way - instead of lowering damage of beams and keeping the cost, damage should be left alone and costs raised.

I made a spreadsheet with comparison of the weapons and defenses, in which I also made a separate sheet to make my own changes to fix some of those issues - note that this is a preview and consider it still work in progress (I'm working away this week so I don't have much time to play or test things). After I finish with weapons, which needs a bit of polish, I want to give defenses a go and balance them out as well.
The changes I made so far:
- Fixed most weapons being weaker then their preceding tiers. Some values of max damage had to be increased. Some size mods and size values were changed - for example, made starting weapons bigger and equally bad. In case of Harpoon I and Photonic Torpedoes I, they are not weaker anymore, but dish out about the same damage. They are cheaper as you research the following techs (II and III), so cost is not static within those tiers. Not pretty but does the job done and doesn't make next techs any less important.
- Beams are 20-30% stronger then ships outfitted with Mass Drivers. 40-60% more expensive as counterbalance. A move closer to vanilla.
- Changes to missiles to scale better on higher hulls - lower average damage then beams and comparable with Mass drivers, but scale better on large and huge hulls - in other words, they work better with bigger hulls then miniaturization alone. Goes well with original design.
- Made Kinetic Streams and Seeker Missiles smaller (so ships end up stronger) like they are supposed to be, just as in vanilla.
- Made Ion Cannons a bit smaller, same as above.
- Ion Beams are bigger, but still better then Plasma. They also scale better on larger hulls, like in vanilla. They were hardest to get right (being bigger/better) without redoing the whole tree, so I went the ugly way (like in vanilla), and made 3rd level with higher damage and price. 
- Remade Evil weapons. My concern is that AI may be picking weapon modules with highest damage value when designing ships, so it might be possible that even as the AI reaches 4th tier of weapons, it still might use Evil weapons, as they are just copy pasted with changed price and all other stats the same. I made Evil weapons slightly better in terms of damage, but also made them worse at scaling on larger hulls. 

https://www.dropbox.com/s/q6bhw92ysb6ivom/Weapons%20reworked.ods?dl=0

This is the spreadsheet I made, I hope everything is easy to read and compare. I left miniaturization values in vanilla sheets unchanged from CU, as it is not that important. 

In general:
Ships with Beams will be strongest early, but will be costly. Every 10% increase in damage will mean 20% increase in cost over comparable damage of Mass Driver ships, 20% in 40% and so on.
Ships with Drivers will be weakest, but cheapest and fastest to produce.
Ships with Missiles will be more or less on same level as drivers, until bigger hulls are available - higher tiers of missiles will be even stronger then Beams on Large/Huge hulls. They might be pricey on smaller and medium hulls, but overall they are cheaper then Beams in damage/cost.

If you are ok with those changes I'll edit the GC2Types.xml and start working on the defenses. 

OShee, while I truly appreciate the time, effort and thought you've put into the weapons/defenses spreadsheet but tomorrow ends the CU development cycle. A change of this magnitude cannot be tested in time. 

If SD wants to take a crack at redoing the weapons/defenses they can and might.

I'll release one more update this afternoon. V5.7.1. It mostly cleans up some irregularities between TechTree.XML and RaceTechtrees. Some branches are set to different values. While this has zero effect on the game it's a simple to fix.

Also the Thalan's aren't doing so well anymore due to the AI changes so I'm giving them InterstellarConstruction as a starting tech like they had in vanilla. Without it they procrastinate their research into their hull tech for ages and miss out on Remedial Engineering and Basic Gravitonis. Both techs give a very big bonus to research and production.

Yeah that perfectly alright, go for it. But this is as far as we go.

I have to agree with OShee. Weapons were a bit more exciting before the values returned to near-vanilla. Hells, over the years I've done a crapload of games with modded weapon values and the AI seemed to do fairly well. So, I smell a weapons mod reboot.

I'll echo what Mabus wrote above. It's been great to have you all on board. I know that it wouldn't have been possible otherwise! So, humble thanks from me.

For what it's worth, I would still be interested in a weapons mod. Let's have it.

Late to the pre-release party,  had a very busy week.

Glad to see the mod and Frogboy's fixes are working nicely together!

I'm all for locking down the CU -- it has to be done sometime and I think we're in a very good place to do it.

OSHee,  you are still mostly ignoring the fact that a very small advantage in power/size ratio can translate into a very large battlefield advantage.   A ship with 11 attack vs a ship with 10 attack won't win 10% more often -- it will win almost 70% more often.   So Guanathor had to be exceedingly careful to make the power/size ratios varied,  yet not completely OP one way or the other.   He was handicapped even more by having to work with small,  integer values.   I would be very hard pressed to improve on his work.   Yes,  there are a couple tiers that are a little off and tier 3 is actually backwards if you're looking at the 2nd weapon in each tier,  but again.. small, integer numbers.  

That's not to say I wouldn't be extremely interested to see your system,  even a rough draft once you've got defenses worked out.   As I've said before,  it's not that I think there can't be better and/or more varied approaches -- I just think Guanathor's is very very good and perfectly acceptable to release with the CU.

Edit: OShee,  maybe you can break out a new thread for your weapons ideas?   Ideally this one will be winding down.