View Issue Details

IDProjectCategoryLast Update
0023885AI War 2Balance IssueFeb 8, 2021 9:43 pm
ReporterSuzera Assigned ToChris_McElligottPark  
Severityminor 
Status resolvedResolutionfixed 
Product Version2.605 Hunter Intelligence 
Fixed in VersionBeta 2.716 The Great Balance Curve 
Summary0023885: Mk power and deep vs wide tech rebalance proposal (also some low aip relevant adjustments)
DescriptionEdit: Updated file due to problem with frigate count math. Check https://bugtracker.arcengames.com/view.php?id=23885#c59079 for most recent version.

Per the supercat post and some discussion it was generally agreeable that the state of wide vs deep tech, how powerful monotech was with good rng on ars/transports, and some other miscellaneous balance issues were not in an ideal state. Chris asked me to make a proposal so he and others could try it, and I did a bit better and just compiled them into a mod already which I've attached to this post.

I used the information created in the sheet to help work out about what the changes should be. It's a little messy but let me know if any of it is kind of confusing. https://docs.google.com/spreadsheets/d/1kTkZZFBHPmdM7gNH3x345wJR4VdVwGXZSc3jpgnra5w/edit?usp=sharing

Here's the changes:

-Player ship counts by mk are rebalanced around 2x total count rather than 3x. Frigates also follow this.

-The ship count increases are reallocated to be mostly at the mk4 and mk5 point, with none for mk6 and 7, and fewer for mk2 and mk3.

-The big base power jump in damage and effective hp for mk4 has been reduced and the curve is more smooth now. This has the side effect that the AI getting mk4 aip is not as big a deal and might make high aip games more viable. Mk7, mk6 and mk5 have their base stats to be about where mk6, mk5 and mk4 were respectively. Mk4 is now between where mk4 was and where mk3 is.

-These are the exact old and new mk numbers:
Old:
    dam ehp count
mk1 1 1 1
mk2 1.4 1.25 1.25
mk3 2.1 1.75 1.2
mk4 2.8 2.5 1.33
mk5 3.6 3 1.25
mk6 4.2 3.5 1.2
mk7 5.4 4 1
New:
    dam ehp count
mk1 1 1 1
mk2 1.4 1.2 1.05
mk3 2.1 1.6 1.1
mk4 2.7 2 1.25
mk5 3.2 2.4 1.35
mk6 3.8 2.8 1
mk7 4.4 3.2 1

-Mk techs have been reallocated to be 3 hull and 3 weapon.

-Hull costs are lower, and weapon costs are higher. Especially the last weapon which now has a cost of 8000. These are the new costs:
      Hull Weapon Turret
1 2000 2000 1000
2 4000 4000 3000
3 6000 8000 5000

-This is somewhat less science cost to max a monotech path (26000 vs 33000), so starting science is reduced by 4000 to help keep this from benefitting super low aip as much as other strategies by making you have to capture somewhat an amount of planets like you had to previously to science cap a monotech. Monotech will also have a lower power cap relative to ai strength since max ship counts are reduced 50%, and to reach the same power you generally need more fleets active than before. However, with 3 hull techs it is a lot cheaper to max more techs, and thus have more useful higher mark fleets to make up the difference without needing good to perfect ARS/transport rng.

-Due to 3 hull techs, cheaper hull techs, and AI planets for mks 4-7 being a bit relatively easier due to the mk base power curve change to remove the mk4 bump, the game start is somewhat easier. AIP for AI mk upgrades has been reduced by 10 on all difficulties 7+.

-Some mkless units have been reduced in power directly to account for reduced player power. The overlord phase 2 in particular was impossible with half the maps ars and doublers plus a spire frigate without direct changes to them, even starting with a clean map in a solo fight. All exogalactic units and overlord phase 2 had their hp, shields and damage divided by 1.75. This should still leave them slightly more relatively powerful vs the player than before, which should make super low aip strategies like used in my supercat run a bit harder since you'll need more science for more techs which means more planets. For now I left the dire guard posts and dire guardians alone, since the posts are not as strong as the overlord which is the only reason to fight them, and I felt that dire guardians never really felt dire to begin with.

-Tech vaults are more useful to low aip games now since multi-tech lets you leverage more fleets and you probably need it to take overlord phase 2 without really good map rng, but the last mk science cost is much higher. Hacking cost to get the tech increased to 60, but reroll costs reduced to 10. Hacking cost to convert to science left as is since even if it's technically more science:hacking point efficient to hack 6 vaults to get 9000 science instead of getting a tech worth 8000, there's only so many tech vaults on the map. Doing the tech conversion for most of them can be a way to alleviate rng issues.

-As a nice side effect, there's less turret count increases in the new intended balance, so they draw less power total for a mk5-7 cap, which might make GCA turrets more useable when you're not doing a chokepoint strategy.

I played a game using this at diff8 and here's my big observations:

-I have a lot less dead weight ship fleets lying around. With the hull science cost reduction, it's easy to get a wide swath of ships to mk4 by just buying the first couple of hulls and the first weapon.

-This uses more metal and trying to replace ships faster than they're destroyed in fights is harder because they're more fragile and blow up easier than doing a monotech all mk7 kind of setup. Part of it is also probably because the damage to hp ratio at every mk level in my proposal is a little higher. It seems like it probably works out ok because doing non-monotech to not mk7 is also a lot more science cheap so you can spend a bit more science on economy.

-I actually got kicked out of some mk5-7 planets I wasn't expecting

-Things like spire frigates and golems are now relatively much more important to get for pushing into the ai core since I did not touch their stats. They're a significantly larger portion of the total power that exists on a map for you to collect now, and you may have to get one at higher to highest difficulty doing very low to low aip to be able to beat the overlord.

-I wanted to finish the game early to post this sooner, so I didn't tech vault my high fleet count weapon techs, and only picked up about half the ships on the map. So you definitely don't need to play perfect and accumulate all power on the map despite the reduction in player power. The optimization floor should also be lower, so newer players should automatically spend science more efficiently due to wide tech becoming more effective. Overlord phase 2 was pretty close with half the map of stuff and no mk7s though.

-The current balance still doesn't really push you to need to go into higher AIP brackets, even with the things I changed to make low aip harder. So at least for diff8 my defense was pretty easy using my supercat strategy and it was just a matter of griding through the overlord defenses to clear the system and having just enough power to beat phase 2.

This might warrant a bit more tweaking (like I didn't touch subfaction mkless balance, just core stuff), but it seems like a good starting point from the game I played.
TagsNo tags attached.

Activities

Suzera

Oct 4, 2020 8:45 am

reporter   ~0058960

Also thanks to Burner for helping to iron some of the math out.

Burner

Oct 4, 2020 9:10 am

reporter   ~0058961

I think this change would trivialize tech choices
I would pick 2 hulls to max (24k science), and pick one level in couple of tech to get an MK5 fleet. I often stop at MK5 even now - it will be even better with nerf to high marks
Same with turrets - 9k to get all turrets to MK4 is way too cheap

I think keeping it 4 tech, 2 hull is fine - so you have to commit to tech to get high level
Maybe we could make hull techs cheaper though (and make techs a bit more expensive to compensate) to encourage people going with the wide options early

RocketAssistedPuffin

Oct 4, 2020 11:02 am

reporter   ~0058962

This is all something I...tried to fix, long ago. I was the main one doing balance until a few months ago (April?), alongside other things. Issues stopped it though.

Something I have considered a bit of a flaw, over time reading the Diff 10 reports, where the "Max something to MK7 soon as and only get those" kept showing up.

I did do the whole...power / science thing that I've seen come up I think, before Fleets were a thing, but didn't include health for reasons long forgotten. I would include it now.

Don't know if I could actually help now though, due to faulty memory and not really following the math I was seeing.

Good luck anyhow.

Suzera

Oct 4, 2020 1:14 pm

reporter   ~0058963

I mean, nothing will fix the "get perfect rng to win" problem other than removing rng (which has pros and cons), but I think relying on 1 in a million rng to win to also basically be not-quite-cheating and doesn't really count for regular play which is what you want to be balanced. For more normal variance this setup feels better so far at least.

The core math here is just that the rough power estimate of a fleetball is sqrt(sum(dps of all ships) * sum(health and shields of all ships)). The charts below that are just dividing those fleetball numbers for mk changes, by the amount of science to reach that mark for various weapon/hull target amounts (mono weapon tech vs dual weapon single hull tech, etc).

Some extra would need to be added to perfectly account for counter-boosts like 4x vs higher armor or space plane style defense buffs, but presumably everything should be of equal power making it a wash if the between ships balance is good.

CRCGamer

Oct 6, 2020 5:07 am

developer   ~0059014

Been doing some testing with the file on the proposed changes. I have a few opinions and comments to share as a result.

First off the average time required to kill frigates has gone up but in turn frigates mostly hit like wet noodles and the reduced caps on ships that were already low cap to begin with really hurts on a per line basis. This actually benefits the AI more than the player because the AI can very easily chuck 40+ frigates around and a player is going to have a big problem dealing damage properly with the dozen or so they are likely to have.

First tier of turret upgrades seems dramatically too cheap. Being able to instantly knock out at least one T6 turret line of choice on the battle station and have decent fleet upgrade coverage is bit wild. And leads to turret rushing being a strategy of choice if you can keep up with the metal expenditure.

The unit count difference between stuff that is normally swarming and more moderate counts is almost not there. Which means swarm units end up being a bit *too* weak because they don't have the same kind of stack health to stay present long term. It also makes the value of using a double cap upgrade on a lower cap higher strength vessel the way to go in terms of overall power gain.

Suzera

Oct 6, 2020 12:18 pm

reporter   ~0059021

Everything player side gets 1/3 less count at top mark, so the total hp+shield pools are 1/3 less from counts, and numerically keeps the same relative value of doubling any particular ship line the same. No ships or categories like swarmers got special adjustments, and every strike fleet lost the same % of hp/shields at cap. Though I imagine from the player getting 1/3 less ships and damage being relatively a bit higher than hp now would make it seem like your ships get destroyed faster if you go into a particular difficulty unmodded and then do the same difficulty with this mod. Which would probably be the better explanation for why your swarmer units get destroyed so fast, especially if you were using them alone where the focus fire effect against them is stronger. Which was intended. This balance is intended to make every difficulty more difficult if you are trying to do monotech like before and get the same amount of ARS on the map, but potentially a bit easier if you do a few to several techs and get more of the ARS/doublers on the map. Each individual fleet should be much weaker used alone or in the same number of fleets as before, but your total strength employing everything at the same time in the same place is intended be about the same or a bit higher without having to clear the map. My diff8 game was mostly raiders and swarmers, and it seemed fine.

I don't really use turrets on offense since you can't just move them like you do fleet, but even with these changes I would not expect turret offense to be as fast as just using fleet on offense. Battlestations move slow and rebuilding turrets on every planet takes a while. After the earliest part and after getting unlazy about ARS unlocks I was blowing through planets as fast as my ships could move between the guard posts on diff8, which is even more doable unmodded. And with not great rng. It's probable the first turret hull should cost more though.

The time for the AI to kill your frigates should not have gone up particularly significantly. The time to kill AI frigates may have gone down because you should have about 50% less total firepower at the same ARS/doubler intake depending on what exactly you did before or after. But see again the note about this being a general difficulty shift for the harder at any difficulty setting, especially if you try to mono-tech without really good rng or use only a small portion of the ships on the map at a time.

Suzera

Oct 6, 2020 12:27 pm

reporter   ~0059022

To add another note boiling some of that down, you should probably expect to play 1 difficulty down for a similar-difficulty experience as unmodded past the earliest part of the game where you should generally be slightly stronger.

Also in total, frigates should have wound up much closer to the same as before, since while they got the same to mk damage and hp reductions as strike craft, they also got the same or slightly increased count depending on exact rounding. If that is not the case there's probably some confounding factor that interacts with mk changes which is not in the mk's xml. But it seemed like it was working appropriately in my test game.

CRCGamer

Oct 7, 2020 7:14 am

developer   ~0059062

Well I'll say that you setup has one very big very annoying issue on the player side of the fence regarding Frigates. There is no increase minimum count by one when gaining tech levels. This is why I'm complaining that you've fucked player frigate counts over. This screws up station keeping frigate counts and fleet counts. It actively hamstrings the player to choose frigate lines when they get a maximum count of two in that slot even if its all they way up to mark seven!

CRCGamer

Oct 7, 2020 8:51 am

developer   ~0059064

Alright on a more cool and actually helpful note what you need to do is actually give back the 1.1 scaling per "mark" to frigates. Just stop adding an extra 1.1 one mark earlier than the base file had it. I'm perfectly fine with having 1-2 less frigates by high marks. But it is effectively a dead slot in a fleet or ARS with the way you tuned it. Because most frigates found in the wild are singles and very rarely a double. Meanwhile some starting fleets like the raid fleet start with three... and if you tested using the raid fleet the single "double" number would look fine because six frigates is fairly comparable to the 7-9 the vanilla file would let you have.

That definitely doesn't work out for Logistic Stations that start with a single Assault Frigate at mark one. Stuff like that gets one extra at one mark gained and then never manage to snag another frigate being available outside of your scaling for Mark V and a Frigate Coordinator being taken. In which case it would be a grand total of three ships. Instead of seven to nine ships being available. Military Stations start with two, and only scale to four with your math in the file.

Suzera

Oct 7, 2020 5:33 pm

reporter   ~0059079

Anything with 1 count should get two at least by mk 5 from what I tested. If not something odd is happening because that would mean it seems to sometimes rounds up from 1.99 to 2 and sometimes it doesn't.

I did run my test game with average amounts, which I think always gave 2 for a 1-2 count due to rounding, so that could be it. Thinking about it, I don't think the initial count is ever a fractional number in the background, which multiplying by two like I was would rely on the keep it less swingy. I think giving back the 1.1 repeatedly scaling and just leaving one off would work. Which I've updated and attached to this post.

Worst case I think frigates were slightly underpowered before, and this could make them relatively somewhat better than before at higher marks.

Suzera

Oct 7, 2020 5:39 pm

reporter   ~0059080

Oh, "one extra at one mark gained" not "one at mark one and no more". So it was kind of working as intended with consistent rounding in that regard. But yeah, that would cause some frigates (especially with starting counts of 1) to have much less than 1/3 less count as without the mod, which was not intended.

Chris_McElligottPark

Feb 8, 2021 9:43 pm

administrator   ~0060475

A lot of these things are accomplished in a similar way to what you proposed!

Issue History

Date Modified Username Field Change
Oct 4, 2020 8:43 am Suzera New Issue
Oct 4, 2020 8:43 am Suzera File Added: SuzeraSciencemkRebalance.zip
Oct 4, 2020 8:45 am Suzera Note Added: 0058960
Oct 4, 2020 9:10 am Burner Note Added: 0058961
Oct 4, 2020 11:02 am RocketAssistedPuffin Note Added: 0058962
Oct 4, 2020 1:14 pm Suzera Note Added: 0058963
Oct 6, 2020 5:07 am CRCGamer Note Added: 0059014
Oct 6, 2020 12:18 pm Suzera Note Added: 0059021
Oct 6, 2020 12:27 pm Suzera Note Added: 0059022
Oct 7, 2020 7:14 am CRCGamer Note Added: 0059062
Oct 7, 2020 8:51 am CRCGamer Note Added: 0059064
Oct 7, 2020 5:33 pm Suzera File Added: SuzeraSciencemkRebalance-2.zip
Oct 7, 2020 5:33 pm Suzera Note Added: 0059079
Oct 7, 2020 5:33 pm Suzera File Deleted: SuzeraSciencemkRebalance.zip
Oct 7, 2020 5:39 pm Suzera Note Added: 0059080
Oct 7, 2020 5:46 pm Suzera Description Updated
Feb 8, 2021 9:43 pm Chris_McElligottPark Assigned To => Chris_McElligottPark
Feb 8, 2021 9:43 pm Chris_McElligottPark Status new => resolved
Feb 8, 2021 9:43 pm Chris_McElligottPark Resolution open => fixed
Feb 8, 2021 9:43 pm Chris_McElligottPark Fixed in Version => Beta 2.716 The Great Balance Curve
Feb 8, 2021 9:43 pm Chris_McElligottPark Note Added: 0060475