Spring RTS Engine

https://www.youtube.com/watch?fbclid=Iw ... e=youtu.be

some of you might not have realized it but I went deep dive into wargames and have learned a lot but this conversation touches on a lot of things.

People used to get annoyed by xyz "what is it's role" or "your ballance makes no sense..."

Things like ACES were *Cheaper* because they required singular focus and they were also more rewarding because in my eyes the player's focus was part of thier cost.. comparitively things like the spam units, you would build a horde and charge them across the map. they could do serious damage but were less effective than say specialists that had roles that would require more management in their use.

NOT to pick at an old grudge. but one of the key disagreements I used to have with argh about his nanomachines game was that he tried too hard to achieve "perfect" balance but people are imperfect and they are not going to be mathematically accurate or precise in their gaming.

Nice find - thanks for sharing!

I also agree that what they call a chess like "mechanical balance" actually isn't achievable no matter how hard you try. Plus for the majority of Spring games we are creating RTS games which add like a whole new world of complexity and mechanics to the mix. Think of reaction time / APM or units like in Starcraft where they cannot shoot while moving vs. creating units with rotating turrets that can fire while moving. That makes things even more situational than for e.g. the example from the video about how to assess amphibic units which depending on the map can lead to the amphibic attribute being game changing or worthless.

In general striving for a mechanical balance of course is the way to go but as I see it you should just scrap that idea at a certain point. This only works for a very distinct kind of game design where there is a lot of uniformity with chess being the perfect example with two factions having exactly the same pieces which can move on an extremely simple board in a turn based manner with only one unit being active at a time. The further you stray from that the sooner you should forget of trying to accomplish a mechanical balance with most RTSs and tabletop games being lightyears away from a simple design like chess.

When it comes to mathematical balancing that might be a handy tool but one yet again runs into the problem of a complexity level that's just too high. No formula can possibly decide how to bring a feature like different height levels into the game balance. It's totally arbitrary and up to the game designer to decide if that should have a strong or low impact. Even with such a decision being made, the mathematical approach still won't help that much as such a situation occuring is still an optional thing that can or cannot happen. When now going down the rabbit hole the concept of a mathematical formula would inlcude so many arbitrary design decisions that the underlying idea of why to use math as the utmost important tool in the first place has become pretty futile. For these reasons such a mathematical approach has only two possible outcomes in my opinion:

1.)
You just get lost trying to come up with equations which try to model all the complex features of the game which will ultimately fail as those equations will be filled with so many design decisions that it defeats its purpose plus at a certain point one will struggle to find appropiate equations for modelling the complex behaviors in the first place.

2.)
Despite the given complexity the forumlas are kept simple (and thus diregarding to try to model the whole deal) in a way that the applied algorithms will run into a stable state where basically everything is homogenous i.e. the same. That's of course one sort of balance but every unit becoming pretty much the same means it in the end doesn't really matter which unit you use or in other word there's no real need for your roster of different units as they're all nearly the same with a different look.

Smoth mentioned Argh and his projects - I still remember Tired with his mathematical approach for Tired Annihilation which in my opinion just ran into these problems...

A new and interesting approach would be the integration of machine learning. While it wouldn't really add anything new to the iterational process of how balance changes come from peoply simply playing the game and identifying issues this could help saving time until a generally speaking good balance has been found. The Starcraft II AI Deep Mind has shown that it's possible to create an AI opponent on par with the top players. Of course the question is if balancing things with that skill level as the general basis is wanted or not. It still could help identifying certain strategies being too strong in terms of them being pretty much the only way to go for most situations...

Will be watching the video later, but M-A's post has reminded me of Nemo's n00berhack mod which was basically 'balance through everything being massively OP' he also made a mod where prices fluctuated according to supply and demand which was a cool idea.

I have fallen for the allure of mathematical balancing in the past and found it wanting.

As much as I find Wargamings (World of Tanks) method of 'nerf tanks that are in a greater % of winning matches' metric massively flawed, the general approach of sanding down rough edges appears sound as a heuristic mechanism for balancing.


I've come around to a view that balance needn't really be equal... i.e. Sweden in S44 was always meant to be a somewhat underpowered, harder side to play. All the sweeter to win with, against the odds.

to me, good balance aims to achieve :
1- equal opportunity to win between factions on most maps when the battle starts
2- increased variety of different strategies being competitively viable within each faction


I stuck with mathematical balance for the most part on MF. I use it to assign costs automatically for most units.

Imo it's the only reliable way to get consistent balance between lots of units without lots of player feedback. Don't expect it to produce "perfect" balance (there's no such thing), but after setting it up it's a tool that helps getting good results with a low amount of effort.

It's not easy to get it right, especially as the number and range of unit attributes increases. At some point there's a need to either
- acknowledge that a unit has some unique mechanic or is so edge-case that it's best just having manually assigned cost
- acknowledge that the formulas aren't producing good results and revise them
(generally this is where a whole class of units which share some traits are either too cheap or too expensive)

I keep having to revise my formulas but it's getting better over time.

Another thing to consider that's talked about in the video is "what attributes should come with an associated cost increase? "

Should a unit that can fight in more domains be more expensive (ex: all terrain, or amphibious)? That'd make it suboptimal if those "extra" domains don't feature significantly on most maps.

Should there be "free perks" for some units? Generally i'd say one should't get "something for nothing", but it's debatable. Being all-terrain or amphibious on my game has no cost penalty atm.

Should units be balanced by their raw stats relative to an arbitrary reference (i do this), or should the relative strength of possible enemy counters and other available options on the same tech path be considered in the unit's cost?

There's an infinity of balanced states that lead to enjoyable gameplay. One can look at BA and ZK and see a light assault tank with nearly identical stats and widely differing stats for skirmishers or static defenses, yet both are considered balanced by the respective communities.

raaar wrote: ↑11 Feb 2021, 04:55I stuck with mathematical balance for the most part on MF. I use it to assign costs automatically for most units.

Imo it's the only reliable way to get consistent balance between lots of units without lots of player feedback. Don't expect it to produce "perfect" balance (there's no such thing), but after setting it up it's a tool that helps getting good results with a low amount of effort.

I think what might be interesting to consider is if the balancing is in its infant stages, where things come together for the frist time, or of it's more of a situation where a new units is added to an existing roster with preexisting, reasonable balance.

When creating my *A mod, apart from having a general idea in mind of a unit's costs and power, I went straight ahead and made certain "versus" scenarios. Let's say a new tank has to be introduced to the game which is a single target hitter. Maybe there already is another single target tank that's lighter / faster so I start with doing 1v1 situations and look how much health the stronger tank has left in the end and then have a look at the costs. Then I go for another tank that has a strong area of effect weapon affecting multiple targets at once. I once again test the two units in a 1v1 situation. Then I increase the unit numbers that fight at the same time so the tank with the area of effect weapon, which should have been inferior in the 1v1 situation unless there also is a huge price gap, should start becoming way more valuable now and I check if this is the case cost and unit stats wise. Doing just that already gives a nice first iteration balance with a need for some fine tuning with regard to weapon range and unit speed. So it's not about just putting the units next to each other so they are in weapon range but it's also about letting them advance a unit with superior range and having a look at if it's possible to either micro a very agile unit so the new unit's shots keep missing or even using superior speed to always stay out of range. I don't see any way to incorporate these tests, which actually don't take that much time, in a mathematical formula.

This method might only work this well for a game like Maximum Annihilation though, where several tech levels exist and all units have a very distinct role and purpose. There are archetypes like the glass cannon, the bullet sponge, the long range artillery, the AoE damage dealer, the fast raider, the multi-purpose unit, the static target hitter or the allrounder plus specializations like being amphibic or all-terrain. Plus a unit's properties are very static (i.e. no veterancy). When balancing a game like Warcraft with heroes, level-ups, upgrades and special abilities, this concept might still be applicable but it will probably take more time to perform all these tests as a unit's worth in a certain situation can strongly vary by receiving a new ability...

raaar wrote: ↑11 Feb 2021, 04:55Should there be "free perks" for some units? Generally i'd say one should't get "something for nothing", but it's debatable. Being all-terrain or amphibious on my game has no cost penalty atm.

Interesting. In MA there is a distinct penalty for such an ability which either is represented in a change in costs or unit stats. When talking about an all-terrain unit I try to compare with another non-all-terrain unit that has a similar role and purpose. The all-terrain unit then has to be significantly weaker in terms of cost-efficiency (it won't be trash but the non-all-terrain unit should clearly be superior and thus the better choice when fighting on flat terrain). Keeping them on the same level would be very unfair in my opinion. The reason for this is for example when playing on a hilly map which has certain narrow passes which basically are the only way to get to the other side of the mountain range. Those passes are usually strongly fortified. All-Terrain units now enable a player to completely bypass this. If such a unit has the same efficiency as a normal tank that would have to play the barricade runner I think that's quite unfair as this is a huge advantage. The same goes for amphibic units which can gain access to just lightly defended regions. I did try Metal Factions a few weeks ago (nice game btw ) and as I see it getting around a fortification and into the heart of the enemy base can hit the eco very hard with one such successful attack being able to pretty much decide an entire game even though it might be no immediate victory (although a commander snipe is very well possible when talking about more than just hitting the economy). For tabletop games which aren't about basebuilding and new unit creation this might be less fatal as in this case such a maneuver would just lead to a very efficient flanking of the enemy leading to a "combat like any other" where one side got an advantage through clever tactics. Plus there is another thing - I'm not too much into the very strategic tabletop games but I think for the most part there is no thing like fog of war. So for those tabletop games it more or less is obvious that an all-terrain unit is incoming. In Metal Factions such units can very well remain undetected until they have reached their destination (especially amphibic units suddenly emerging at a coast). So not only would amphibic units have the possibilty to strike at a weak spot but they also can easily have a stealthy approach. Together which such a successful attack being heavily punished with if done right being the game winning moment, I don't think such an ability should be "free of charge"...