Berserk Cyborg wrote: ↑16 Jul 2019, 03:56
Hmm? Hosts send their active mods to clients, irrespective of the map, which appear in the mods/downloads directory of the client. If clients already have a mod with a matching hash in that folder, then they don't have to download it again and can keep using it for as long as that iteration of the mod is prevalent.
This must have just changed within the past year or so because previously players would get mod not compatible with host if a player has a mod loaded and attempts to join a game without it or vice versa. There wouldn’t be a transfer, just incompatible versions.
Berserk Cyborg wrote: ↑16 Jul 2019, 03:56
All the devs are/were players, some more than others, so it's not like we don't "understand them".
Rarely have I seen people complain about the stats my time here. Most reports are about simple things that just aren't working as they should be, unit behavior, sync issues, or actual crashes and whatnot.
The reason players don’t complain about stats is because the vast majority do not know what the stats are. They don’t know what the possibilities are and what can be changed. All they know are damage, hitpoints, rate of fire and maybe a few other things. They wouldn’t know what to complain about.
There is a huge difference between player vs player and player vs bot but the people who do not have a lot of player vs player experience are unaware of this. It’s like the difference between looking at something under a microscope vs just looking from afar. A bot will never challenge a human to use the best strategies but in player vs player the most optimal strategies have been identified over the years. Unfortunately, the current best strategies are not strategical at all, they are unintended consequences of imbalanced stats.
Berserk Cyborg wrote: ↑16 Jul 2019, 03:56
I'm sure everyone on the dev team knows at least the basic history of the game. It's no secret the stats are far from ideal. However, balancing is often time consuming or considered rather hard to do and it requires a lot of testing to make sure everything is "just ok".
And that is what makes balancing hard: there is no known way to quickly test balance changes. It would be amazing to devise a way to automate this in some shape or form. Of course, to even begin, it would require some kind of standardized ruleset and probably some kind of custom "balancing" AI and a new script library to go with it... and a bajillion test cases to go with it.
What you speak of above is already done, this is what I’ve been working on for years… I’ve manually launched the game 1,000’s of times and simulated perfect research games to test everything is “just ok”. I don’t know that it’s necessary to programmatically have test cases. I’ve simulated interactions of all turret, body and propulsion combinations at all points in the tech tree. This was especially painful for the engine upgrades because it requires every turret/body/propulsion/upgrade combination to have a unique speed signature without crossing certain thresholds. There is a certain threshold in the speed calculations where droid speeds jump 50% which makes some combinations way over powered. Prime example being Heavy cannon, hover, mantis. If that combination gets the speed bonus it’s way too powerful (ask any player from 1999-2007). There’s still room for minor stat improvements but nobody would notice unless I pointed them out.
Here’s an example. I had to add short hit/short range back into the stats since the devs re-added it 3.3 (I think it was you, good job if it was). In my stats editor, I just insert a column, sort the weapons by weaponSubClass and initialized the following scheme:
I launch several games and test out the scheme. Altogether this takes about an hour. From experience I know that edge cases concerning short/optimal/long range are vtols and mortars while using pursue, especially grenadiers. I’m the only person that can make these changes and knows what they’re looking for. I’m certain that nobody will notice the VTOL light cannon using the short range value while pursuing unless I say something but I’ll spend 15 minutes making sure that it works. The short/long range and hit values are the reason why all non-bomb VTOL’s are worthless but I think I’m the only person aware of this. Player vs player people definitely acknowledge that bomb’s are the only VTOL’s used but they don’t know why or how to fix it. Non-bomb VTOL’s are worthless compared to the other options available at the time and only a player vs player person would know that.
I also test the stats using conditional formatting in Excel. I have it such that if any short hit value is < long hit then the cell will be highlighted red. I check for many conditions like this including duplicate object id’s, null values etc...
Another major flaw that should be considered “broke” from a gameplay standpoint is that structures and droids use the same turret. This means they only differ in their propulsion and hitpoints. Players don’t know this unless they know how the stats work. Structures should be thought of as immobile droids with extra hitpoints. It implies that all defensive structures are inferior to their droid counterparts except in the few rare cases where the map has a single choke point. Only player vs player people know this gameplay dynamic and only a stats person would know how to balance it. I fixed this by giving structures their own turrets and initiliased the following scheme
Bunkers +1 tile range
Hardpoints +2 tile range
Towers +2-3 tile range
Using the stats editor, I can set the long range for structures to = 128, 256 or 384 more than their droid counter parts. Just by doing this simple fix, defense structures are now viable (although I later found this came at a cost to WZ consuming more RAM).
I uniformly rescaled research costs and structure build times using certain schemes as well but when I post on the forums I think I’m the only person that’s understanding the implications.
Another obvious stat fixing is that 90% of all games use the high power setting and most player vs player games use 40 oils. This is entirely because of bogus stats. The power multiplier needs to be fixed but it's going on 20 years and nobody will fix it.
Berserk Cyborg wrote: ↑16 Jul 2019, 03:56
Warzone isn't exactly super popular and most players are playing skirmish or campaign.
I think there’s a misunderstanding of cause and effect. This doesn’t mean the stats for multiplayer are not important because there aren’t many players. The reason people don’t player vs player is because the stats need fixing.
Berserk Cyborg wrote: ↑16 Jul 2019, 03:56
Change can always be a good thing. There just needs to be someone to do it, and, not do something radically out of touch with the core mechanic of the game.
I think it’s already done. What I have is a good baseline. I don’t think I changed anything too radical, I kept everything in the original spirit and emphasized intuition. If there’s anything in particular about the mod then we can change anything easily. The reason I started adding new stuff is because there wasn’t anything left to optimize in the original stats. I already reviewed every attribute of every object in all the stats files and optimized everything. At that point I started adding new objects. The names of turrets/bodies/structures are not as important as their stats.
It seems there is a dilemma that people don’t want radical change but the stats need radical change. Most of these changes cannot be submitted in isolation because there is a whac-a-mole effect that causes other imbalances. That’s why all changes are bundled together into a mod because they wouldn’t make sense by themselves. EB mod should be thought of as a stats wrapper around all the relevant mods in the add-ons section. I propose to add it to the game as a baseline expansion pack. I'll leave the idea here.
EDIT: I would say the best way to test the differences is to start a skirmish game, attach an ai and pause the game. Then load the mod and do the same thing and simulate 2 games side-by-side one with mod and one without.