Testing Economic Balance Changes

[sven]

- They gain more terriroy withi nthe same timeframe of 200 turns: this seems laregely due them semengly be able to "deterr" Marauders more effectively

My own guess is that an increase in territory gain would probably be due to some tweaking I did inside the AI's routines for building outpost transports in r38171 -- I found and fixed a few cases where the AI could stall due to insufficiently aggressive outpost construction. But a higher early game ship building priority may also be helping the AI's clear out harpies and pirates earlier, which could also be helping with expansion rates in builds r38170+.

I'm still seeing issues with Marauder/AI interactions in the current build, but, I think what's going on there is typically that the AI underestimates how big a fleet it will need to capture a marauder base, and so it ends up starting a fight it can't win. Marauders should never be initiating wars against the AIs on their own -- if you see the AI getting raided by marauders, what's usually going on is that the AI tried to conquer the marauders, bounced against the marauders' planetary defenses, and then got counter-attacked.

[zolobolo]

Tried an extreme end of this config on normal, ellipse Huge galaxy but 99 stars (AI performs better with more planets and buffer space)
All factions were running under AI control to compare them all (and Phidi have been performing clsoe to Human so can serve as benchmark going forward)

Uploaded: game_8351

Spoiler: show

target_navy_investment=0.95
elseif empire.ai_state.strategy_cards?.rush
target_navy_investment = 0.9
elseif empire.ai_state.strategy_cards?.tech
if war
target_navy_investment = 0.85
else
target_navy_investment = 0.75
end
else
if war
target_navy_investment = 0.95
else
target_navy_investment = 0.85
end
end
local credits in empire
if credits > 5*psuedo_income
target_navy_investment=max(0.9,target_navy_investment)
end

Humans, Gremak and Tinkers performed the worst - these are the usual suspects expect for Tinkers
Here are my findings for Tinkers:
- Yes: AI attacked Marauders and suffered defeat loosing ships and beign raided at the same time
- Tinker AI did not attack pirate planet and single Light Cruiser altough they wastly outgun them (did not capture planet due to this) This was one of their biggest flaw as it has pulled a break in their expansion (the system effectively blocking them from spreading out)
- Asteroid Outpost production progress lost when AI changed planet build
- Too many factories are beeing built and too quickly (2 factories on a new building)
- Tinkers build 2 facotries on new planet with minor (rebelling pop) instead of market already being available)
- Tinker scout has extra munition and anti missile - replace extra munition with Research Station (not researched yet)
- AI producing ship even though metal is running negative (too many factories)
- AI can run a negative budget and needs to scrap ships next turn withouth stopping producing new ships or factories (two new factories being built under negative budget instead of market or coin focus and one the planets is newly colonised and does not have any other building yet)
- Suggest: force AI to always start with farm, mine of Market on the first 2 slots of a new colonies and never with factoy or lab. It should always be farm if Fertility=high, Mine if Metal=High, but Market should always be in the first 2 slots altogehter when researched already
- Tinkers were not refitting Desrtoyers for a very long time (100 turns) (did refit Missile Destroyers) for a tiny amount of Coin 4 (+11 income, 335 reserve)) and Metal 6 (+48 income 642 reserve) after Deflector Research done. Maybe AI does not upgrade below a minimum cost/amount of mount swap?
- Tinker Missile Destroyer does not need extra munition, Tinkers never run out of munition Please give them armor instead
- Suggest for Outpost to provide +1 Research (or +1 trade) so that they provide a minimal amount of bonus besides range but not enough to exploit them. This would effectively nullify the reasearch outpost spam exploit as well
- Building Scouts (over 2 units) should be prevented for the AI especially now that the cost wear them down much more

Summary: Coin seems to be fine, only Tinkers seem to have been scrapping ships due to runing negative budget (due to above changes to buildup) but even so the Tinkers have pushed through and were not the worst performing faction: they basically brute-force survived with the help of their ships even with low amount of colonies

In the end, military strenght was ok for 2.5 empires (Colonials are almost there . I think if we hit 3 competitive empires at turn 200 that shall make for stable influx of good mid-game stories (battles and positioning)

[zolobolo]

Humans Findings:
- Suggest for Humans to start with two Scouts as the AI has stumbled upon the worst planet around the first time and chose that as its first colony: planet supprots 2 pops, and is poor in everything - this pretty much guarantees that the AI will not be taking off within the first 100 turns
- Due to the above, the AI ended up wit ha transport holding 1pop that could not be deploye even at turn 80 (As its two colonies only supported 5pop altogether) - this was a huge waste of the factions starting potential. Dont forget to coutn i nthe loss of pop growth these planets did not provide to the faction in the early game all the while the faction had both a 10 and 14 pop supporting planet in range right off the start
- Aternately: have Human AI only colonise planet supporting at leat 4 pop for the first 2 colonies
- Most of their initial 100 turns were spent building factories on both of its first two planets. The above finding of preventing AI to build factories within the first 3 slots should help here as well. One exception from this rule might be if planet supports >5 slots, and faction can buyout teh factory rigth away)
- Also neglected to build market on newly colonised planet with natives even when markets were already researched - the same as for other factions
- Started building Light Cruiser instead of Destroyer as its first combat ship (next to the inherited Heavy Cruiser) - this is a bad decision given the value of small vessels as I described in earlier post. They should prioratize small ships at first as these do not have shipyard prereqisite, cost, less to build, less to maintain and done faster and can cover more ground (in some cases destroyers are outright more cost effecitve as Crusiers overall espeically serving as PD)

After these we hit Turn 170 where most of the other factiosn have an effective strike-fleet and Humans not having any combat ships: the many delays above prevented them from gearing up in time

[SgtArmyGuy]

Uploaded game_8353 for review. Colonials on Hard.

Stardate 1285 had a possible bug occurring when I was attacking Bacabs: when I pressed "autocombat", the game for some reason told me that the combat against the unarmed outpost was "inconclusive" and the battle ended in a draw. I therefore clicked "undo" and entered combat manually, but as the outpost blew up, the combat once again ended as "inconclusive" and told me it was a draw. The station, I believe, was not destroyed on the strategic map (although this part I may remember incorrectly; this was the case after the autocombat, but I'm not sure what happened after I finished the combat manually)

The station was Human, so I guess it was called "Babylon 5".

This game was such a weird clusterfuck of multiple-front wars that I'm not sure if it is very representative of the new balance changes as a whole. The AIs were constantly at each other's throats, so I was able to win a lot of territory by timing my attacks right (hitting them in the back while they were fighting somebody else, and always making peace when offered). The thing is, I was able to win the whole game by building about a dozen light cruisers with basic lasers (with Accurate, Armour Piercing and Rapid Fire) and not much else. The AI's fleets were about up to par with mine, and my first initial war with the Imperials would have probably ended badly had I continued the campaign to conquer their last planets (they were defended by their main fleet, which could probably have destroyed mine). The Gremak fleets, on the other hand, seemed oddly wimpy throughout the game (tech-wise). The other opponents seemed more formidable.

I had some trouble with the upkeep costs early game, but was able to build enough markets (and had accumulated enough cash) to balance it off just before going bankrupt. Still had to do refits manually through labour, couldn't afford to make them with cash due to the cost in coin going to something like 1500 coin per 6 ships (and the cash flow being negative at that point). Using labour, I was able to refit each ship in one turn, so about 6 turns for the whole fleet of 12 but an insane amount of micro clicking through all those manual refits. This is the one thing that still produces irrelevant, stupid micro in "modern" SiS. Can we get a single "refit all" button like the one using cash, but using planetary labour instead? It could use the same drop-down menu for selecting the hull schematic you want to upgrade to, and then send the individual ships automatically to the build que of the current planet the fleet is stationed on. This would save a tremendous amount of time and tedious clicking and dragging-and-dropping. I already suggested dropping the refit costs altogether, but apparently that would open a whole new can of worms, so we need an alternative solution. Maybe this could be it?

I noticed the upkeep costs going up in the designer when upgrading hulls, that was nice! Maybe the end-game curve could still be steeper, though. After a certain point, I was still getting something like +200 cash flow per turn - although, admittedly, I still had only 12 ships at that point as well. And the game was basically over already. So maybe this game wasn't truly very representative at all...

tl;dr: Winning the game still felt strangely easy, vanilla build with the gazillion enemies in the end-game was usually more challenging (although you could not always DO anything against those armadas at that point; in that sense, the "modern" SiS feels more fair and balanced). I still miss that feeling of fighting against impossible odds, though - that is something I'm simply not getting in these experimental build games. Something from the fun factor is missing. And refits are still impossibly costly or hair-raisingly micro-intensive.

[zolobolo]

Winning the game still felt strangely easy, vanilla build with the gazillion enemies in the end-game was usually more challenging (although you could not always DO anything against those armadas at that point; in that sense, the "modern" SiS feels more fair and balanced). I still miss that feeling of fighting against impossible odds, though - that is something I'm simply not getting in these experimental build games. Something from the fun factor is missing. And refits are still impossibly costly or hair-raisingly micro-intensive.

What galaxy type were you using and how many stars?

We have been discussing this and one of the main reasons for the rebalance seems to be GUI limitation handling large amount of ships and to some extent to soften the difficulty curce for new players on normal

In general I agree that rebalance is needed to cut down on large ships (medium spam in mid and large on late game) but I do say that the rebalanced AI cannot be as challenging as the current stable build in case the balance is doen against ship upkeep as:
- Lower number of units inevitebly makes life harder for the AI (in all games - just think on the change from CIV 4 to 5))
- The AI already has little understanding of the value of small ships: faster to market, more cost effective (this cannot be "balanced" out by making smaller ships more expensive to build/upkeep as that just slows down early game). Combining this with lower ships counts and we'll get situation which you describe: Player can rush all empires with 5-10 small ships and crush their 1-2 medium that htey have managed to sweat out

As is the balance is geared towards limiting fleet sizes so this is what we need to work with. To achieve parity with current stable build competitiveness of the AI, my current approach is this:
1. AI should not get any bonuses to prevent tech skipping and medium and later large hull dominance/spam (which is unthematical, unfun, and this strategy has already been applied by the AI in early builds and still lost against small ship spam of player due to above reasons)
2. Support AI building mixed fleets to be prepared for the most variations
3. To achieve the above, the AI needs to get more efficient in its building order and selection and more effectvie on the galaxy map (not ignoring low-hanging fruits)

As is buildup rate is generally at 75% (varies depending oncards and situation), but this is stil not enough to achieve minimal readiness for the AI at turn 200.
I am current playing around with values around 90% and that is almost good enough. It is all down to point #3 at this point and there have been progress:
- Merc ships are now being purchased by the AI making them reaching stable build competency in some occasions (particularly Phidi has been catapulted as generally a succesfull race to one of the dominant factions in all my plays)

Yes I agree that refit via labor is the most micro intensive work in this game in order:
1. Refit via labor
2. Building buyout
3. Refit via buyout
I do not mind the third point as it is quite efficient overall and on stable I could never really refit everything all the time (due to producing so many ships and buyouts)

On stable I only use labor for refit when it comes to captured vessels: there it is fun as they are all kind of unique by the nature of their aquisition so no problem In generall I would try not to drive the player into labor refit though as yes, it can easily end up in a micro-nightmare especialyl sicne you cannot queue this up withouth blocking the combat readiness of all the effected ships = you would need to set each ship manually at each turn

[zolobolo]

Another run with tweaked buildup, normal difficutly, ellipse Huge galaxy with 98 Stars, usualy outcome at turn 200:
- AI is gaining territory nicely, pop count and coin all are within competitive range for most of them
- None of the AIs have stalled, though Humans got bogged down during the early game again and could not pick up the pace
- The only issue is with military buildup, at this point, the AIs have no chance even if they fought the player all at once
- Some (Gremak and Orthin) have been producing a number of small ships and managed to turn battles against their enemies
- Tinkers were also buildign a lot of destroyers that allowed them to take Arda units and win, atacked them to try their combat power in practice but they quickly lost their ground marked red. Only had 6 destroyers, 1 Missile Cruiser and 1 Heavy Cruiser, but would have needed +50% to have a chance (+3 destroyers and +1 Missile Destroyer, or better yet +3 Light Cruisers)
- Colonials and Imperial are only building medium ships, that is why their military power is much less then their pop and coin would allow them (and they have the best small ship in the game: the Light Cruiser ) - this might be due to the increase upkeep of Light Cruiser: My usggestion is to tell the AI build script to build small ships specifically or take away the higher upkeep trying to balance them out: In my opinion the solution is not the make larger ships on-par with small ships efficiency, but to teach the AI to build small ships and have larger ships specilaised (siege/sniping, boarding, DPS tanks, carrier etc..)
- The large jump in player income is due to market upgrade research doubling the income rate - suggest to soften the market upgrade effect to +50% instead of +100% coin (trade route bonus should reamain as it is already nerfed too much)

In general: no matter high high the coin pressure is the player will alway be able to wiggle out of it with relative ease: the best chance the AI has is to have compatitive forces at turn 100-150 and take territory. Even if they take over other empires, their power grows rapidly with each taken planet and the successful ones will keep th game going (provide challenge and interesting fights)

BTW: I love starting around the middle of the map, and it would be cool if none of the factions would start from or close to the edge, or have this as a galaxy option: in case of Humans it was also a big disadvantage for the AI

[sven]

This game was such a weird clusterfuck of multiple-front wars that I'm not sure if it is very representative of the new balance changes as a whole. The AIs were constantly at each other's throats, so I was able to win a lot of territory by timing my attacks right (hitting them in the back while they were fighting somebody else, and always making peace when offered). The thing is, I was able to win the whole game by building about a dozen light cruisers with basic lasers (with Accurate, Armour Piercing and Rapid Fire) and not much else. The AI's fleets were about up to par with mine, and my first initial war with the Imperials would have probably ended badly had I continued the campaign to conquer their last planets (they were defended by their main fleet, which could probably have destroyed mine). The Gremak fleets, on the other hand, seemed oddly wimpy throughout the game (tech-wise). The other opponents seemed more formidable.

tl;dr: Winning the game still felt strangely easy, vanilla build with the gazillion enemies in the end-game was usually more challenging (although you could not always DO anything against those armadas at that point; in that sense, the "modern" SiS feels more fair and balanced). I still miss that feeling of fighting against impossible odds, though - that is something I'm simply not getting in these experimental build games. Something from the fun factor is missing.

Thanks for uploading this save. I think you're correct -- what looked like it ought to be a challenging and interesting game just based on the AI's development curves ended up feeling strangely easy and underwhelming, largely because the AIs exhausted each other fighting amongst themselves. Your first war against the Ashdar Imperials would have gone very differently if the Imperials weren't heavily engaged fighting the Tinkers at the time when you attacked them, and that early game territory grab is really what set you up for a dominant position by mid game.

There's a lot of different levers I can push on to try and make scenarios like this one less common. One of the bigger changes that's in the current 'dev' build relative to 'stable' is that AI's on 'dev' are much less inclined to retreat once they've gotten into a tactical battle. I think this makes fighting against AI's more satisfying -- but, it also means that AI fleets are much more likely to be destroyed "off screen", as happened in that Imperials/Tinkers war you never saw.

One possible tweak here is to make the AI's tendency to cut and run if they're feeling outgunned dependent on whether or not the player is going to see the combat they're engaged in. That's a bit of an "underhanded" adjustment, but, I think it could be a good change. I think it's good that the AI empires fight each other, but, I'd also like them to be less likely to render themselves incapable of responding effectively to player aggression as a result of wars that the player never sees.

Another set of possible tweaks would be to make the AIs less likely to start wars against other AIs. That's a lever I don't want to press on too hard, as I think a world in which everyone has their knives out for the player, and the player alone, feels more artificial than one in which all the factions are squabbling amoungst themselves. That said, there's probably room to dial the inter-ai conflicts back from "constantly at each other's throats" to something a bit less extreme, and doing so might well make games on 'Hard' a bit more challenging, without sacrificing too much of a sense of realism. (There's already some code in there that attempts to do exactly this -- but, I think this playthrough demonstrates that that logic isn't really working as intended.)

I'll juggle some numbers and do some more playtesting myself later on this week.

[Dragar]

One possible tweak here is to make the AI's tendency to cut and run if they're feeling outgunned dependent on whether or not the player is going to see the combat they're engaged in. That's a bit of an "underhanded" adjustment, but, I think it could be a good change. I think it's good that the AI empires fight each other, but, I'd also like them to be less likely to render themselves incapable of responding effectively to player aggression as a result of wars that the player never sees.

A slightly less underhanded way would be to add an early ship component (hyperspace interdictor?) that prevents retreat, and make the AI very unlikely to build it.

[zolobolo]

One of the bigger changes that's in the current 'dev' build relative to 'stable' is that AI's on 'dev' are much less inclined to retreat once they've gotten into a tactical battle.

Can you please reset retreat threshold to vanilla or let me know where it can be set?

This is a logical behaviour and the change to have ships form fleet should adress the complaint some had wich retreating enemies

Lets look at it this way: If the AI pulls together its ships from this segment int oa single strike fleet and still retreats as it determines it does not have a fighting chance (accoridng to vanilla threshold), then it might as well sue for peace

Less ships also mean less retreating - as is we need more ships for the AI to even put up a fight

Preserving existing ships on the other hand is even more important then before as there are less ships built + refit makes existing ships more important to preserve

If we combine the retreat logic of vanilla with around 90% buildup, that might just be enough - I think the AI needs to be able to organise two effective strike-fleets at any time (unless they control less then 5 planets) to pose an obstacle. As the examples also show, the AI is almost always engaged in at least one war and having all of its ships put into a sigle strike-fleet makes it impossible to defend their territory and generate engagemetns with a somehwat diplomacy savy player

Dialing back the number of AI wars does not help as the AI needs to fight wars to gain territorry: at least 1-2 empires should always keep up with a slightly expanding player till mid-game (turn 200), else the player wins after any conquest.

The AI should fight as usual, and give up as soon as defeat is clear: preserving ships and letting another AI empire grow. This way the player is facing fleets even on smaller empires and the larger ones have a chance to keep up

[sven]

Can you please reset retreat threshold to vanilla or let me know where it can be set?

If you go into Lua state\AI\tactical_ai\CloseAndAttack.lua, and change get_retreat_thresh() so that it always returns 0 (instead of doing what it's currently doing), that should about restore the vanilla retreat threshold.

I'm not certain how big a difference the current retreat threshold settings are having on emergent balance -- but, my guess is that they are tending to make most conflicts more destructive for all involved participants.

[zolobolo]

If you go into Lua state\AI\tactical_ai\CloseAndAttack.lua, and change get_retreat_thresh() so that it always returns 0 (instead of doing what it's currently doing), that should about restore the vanilla retreat threshold.

Not sure this is what you meant but changed it to this and tested again:

Spoiler: show

function get_retreat_thresh()
local attack = all_attacks[get_attack_id()]
return 0
end

Uploaded new test game: dame_8362

Result (AI only run):
- Unsure if the above change did what it should have but the AI performed well: three competitive empires at turn 3 (Colonials made it just barely but I consider their expansion history to wailify them as such)
- Tinkers held on to their ships perfectly, and have a great fleet, a minor one and some additional ships. If the other two empires had such power this would be a very good situation fo a player to engage with
- Humans stalled early game - AI needs support like more money, starting tech or additional ship - or AI needs to cut back on factory building
- Gremak had trouble building upm noticed this several times for them
- Suggest for AI to build +20% trade ship above trade pool capacity to facilitate pop transportation withouth loosing the income from trade (this can lead to unexpected ship scrapping after planets are conquered)
- Major Tinker strikefleet had a tendency to attack Outposts and carry a lot of troop transports around with it: if there is a mayor fleet, I think it needs to prioratize other mayjor fleets or valauble planet and ignore outposts (single ships can deal with these)

[nathanebht]

New game as Orthin with r38511, 88 star, box shape map on hard. Started pretty much in the in the middle of the map.

AIs expanded aggressively. I had a reasonable size empire but was obviously worried that I couldn't defend everything with 4 AI neighboring empires. Didn't get many heavy cruisers produced before 2 AIs declared war on opposite sides of me. Lost one planet. Took one planet. Got a few more heavy cruisers produced and then the humans declared war.

The humans were pretty much boxed in a corner by my empire. They had a massive light cruiser fleet with a mix of other ships. They promptly bombed and captured my only 2 manufacturing planets with shipyards. Figured I lost the game at this point.

I had decided not to research planetary defenses as I couldn't afford the upkeep cost for their effectiveness. Perhaps this would have deterred some AI aggressiveness?

The humans could have easily kept going and bombed out more of my planets. If they had more transports, they could have captured without bombing. After their initial effective push they lost their mind. They sat at a planet for a few turns and then pulled their fleet back for no reason.

[zolobolo]

With the two changes to buildup and retreat, AI is extremely competitive on Hard thanks to their early jumpstart + persistent fleet size, most of the empires are large and have growth potential

Is not my target but might be good to know

Recommend to reduce Ashdar Light Cruiser upkeep for Scout as they are very expensive for their purpose and effect

Might also be worth swapping up the PD gun and the Scanning module as the AI also pays the same amount but gets half of the PD capacity out of the cruiser - though admittedly the design is more recognisabale from among other ships when the scanner is on the wings

Might also be worth reducing the upkeep for the Ligth Cruiser themselves by 1 or increasing mid-size ship upkeep as a Boarding Escort Cruiser only costs 4 Coin which is more then a Ligth Cruiser Scout so the balance has shifted heavily to these designs

[zolobolo]

Just had a game on Normal, where almost all factions (starting with the Orthin) have declared war on me but the Orthin offered up 3 of their total 11 at turn 151 after having lost an Escort Cruiser

Granted, they do only have 2 other combat ships left, this is quite extreeme especially considering that the player is now engaged in war on all fronts

Needless to say, ship count is still a problem (in this game I took away thw buildup modifications)

[sven]

the Orthin offered up 3 of their total 11 at turn 151 after having lost an Escort Cruiser

This feels like a bug in my equations. Could you upload this save?