Balance Team group

Nomander

He mentioned that it's anything he contributed "after this post", and to "treat it as unsupported and remove it where practical." I don't think he meant anything bad, just asking for people to take care of the maintenance/cleanup.

Nomander

@Sheppy said:

I feel like too often there is clearly "one best option" in the selection and that takes away choice. I don't know how to best fix that, only thing I can say is that in most other games I think a lot more about my selection.

Some roguelikes bias rolls towards what your "build" is. I think combining that with general purpose upgrades (eco) would make picks more interesting.
You could even track what units the player has built recently to determine the build.

Nomander

Also printer is 1.5k 4.7k games and plays terribly according to his own replay (that he even thinks matters for the nerfs done) so I shouldn't waste more time here and I wouldn't recommend for anyone else to do so either.

Nomander

@Printer
Reads like AI stylized your post.

not an argument
Post replay where transports matter. AoE vs units outside engi clumps, like shields or mex storages, is irrelevant because its unrealistically apm heavy.
yes that's the point. Second sentence doesn't make sense.
Wrong because it's 3750e/s avg
Wrong number. Also not an argument.
You need to build power anyway to use the unit, this forces you to do it before it is finished instead of after.

E nerf makes arty a more viable counter without affecting total mass cost.
AoE/retargeting nerfs fix the most annoying/toxic things which were killing fab capped mex, stalled shield arrays, and engi clumps.
All of this was explained in the changelog snippet and you should address those ideas instead of just listing the (wrong) numbers then baselessly saying they're wrong.

Nomander

@DommyMommyDostya

You still need a really high number of novaxes to one shot shields through assistance with the increased burst DPS.

Count	avg DPS	Damage vs assist for avg dps after 5.3s	Damage vs assist for avg dps after 8.0s	Shields broken
3	729	10789	8821	No shields break
4	972	14386	11761	ED4 (13k hp) breaks
5	1215	17982	14702	UEF (17k HP) and ED5 (16.5k HP) shields break
6	1458	21578	17642	Aeon (18k HP) and Sera (21k HP) shields break
7	1701	25174	20582
8	1944	28771	23522

A setup of 2 Dukes with adjacency has 1960 DPS and deals 15680 dmg + 6 aoe every 8s, while costing 76k * 2 + 3240 * 6 = 171k mass and producing 2500 * 6 - 1400 * 2 = 15k excess energy.
For just that mass you'd have 5.3 novaxes not counting the 20k/s avg. energy drain.
If you count the energy in terms of pgen-storage infinite grid mass cost (best case scenario) then it's 32k + (3240+200*8)/(2500 * (1 + 0.0625 * 16)) 35630 mass total per novax or 4.8 novaxes.
If you subtract the excess energy from the duke cost (lets say you used it in advance for mass fabs) then the cost goes down to 171k - 15k * 324/250 = 152k, which is equivalent to 4.3 novaxes w/ grid pgen energy.

Also not to mention if you add some engineers the novaxes can't actually break through your shields because they still take time to deal damage. Against 7 novaxes (3 arties equivalent) you'd need 9-22 additional engis to survive the burst damage, on top of the 19-24 engis to survive the avg dps.
Meanwhile 3 dukes one shot all shields and you can't stop it, but the avg dps needs 33-42 engis to out-assist (vs 28-46 to survive the novax volley).
Additionally the engi assist costs are up for a balance pass eventually so that it's more fair across factions, since that mechanic has never been touched.

Nomander

It's definitely possible but it's not easy. The biggest hurdle would be the game's performance stopping you from doing general-purpose machine learning approaches. I think on zulip someone reverse engineered a basic simulation of the game but I'm not knowledgeable on the matter.

There's some work towards build-order infrastructure:
@seraphim-noob has worked on a build order recorder mod ("BO Logger" by Seraphim-Noob). The description says the data is for use in a BO analyzer which I found on his github here: https://github.com/TimMasalme/Bo-Analyzer
@sheppy has a website with build orders in text form manually input from players. https://fa-companion.services.atlantishq.de/maps

In my personal opinion build orders are solvable almost analytically. Your context would be mass costs and mass-seconds (for travel time). The target builds would be a sequential list of builds to achieve at what timing, since splitting focus between 100% the target or eco should always be the fastest way to build something. Once a target is achieved you can add its impact on mass/energy production to your economy's state and continue ecoing to the timing of the next target.

Nomander

ok, thanks for letting me know.

Nomander

@Nuggets said:

Everyone is responsible for activating / deactivating their own mods)

Actually we can update the init file in the game repo to mark a mod as deprecated so that the mod doesn't even get loaded into the game. I did this in my PR that implements feedback for the chat rework.
Also any mod author reading should know that the deprecation status is based on version number, so you can update your mod and then people can load it again.

Nomander

That mod overwrites the entire chat.lua file, so it's unfixable and 4z0t has to rewrite it or game team has to integrate it.

Nomander

Can you post a game log? It sounds like a UI mod issue.

Balance Team

Posts

Member List