Balance and Design -- Learning from Others, and a Plea to the Developers and Community


I am an avid Homeworld fan. I am also an avid StarCraft fan. I am posting this because I believe it is something that needs to be addressed early on, because if it is said in a later stage of multiplayer development, it could be too late.

If anyone knows much about StarCraft, they know that StarCraft 1/Brood War are much different than StarCraft 2.

Some things in StarCraft 2 were improvements – unit pathing, graphics, sound quality, and so on. However, many people do not consider it to live up to its predecessor, despite these improvements.

This is due not necessarily to bad asymmetric balance, if you define balance as the equal opportunity for each race to win – the non-mirror matchups are usually pretty close to 50%. Rather, it is due to design. Broodwar was an incredibly designed game, with various distinct strategies and tactics for each unique race to choose from, and player skill often being noticeable with lots of things scaling with skill.

StarCraft 2 is still fun for many, but there are some design flaws within it – sometimes you’re forced to just sit there and mass units while defending for a big, powerful army, some units are virtually unusable because other units that exist counter them so hard despite the counters to those counters, some units were really hard to control compared to the things they were meant to combat, some things did not scale much with player skill, some things were balanced around “kill them before they get there,” etc. The expansion coming out hopes to solve some of these problems by tackling the core design of the game, but such effort for reworking and making huge changes would not be such a difficult thing had the game been designed well in the first place.

What made it a poorly designed game (in many people’s opinions) compared to its predecessor? Well, part of it seems to be not tackling the design problems early on, rather trying to use what many people call “band-aids” to keep the game balanced while not bothering with the fundamental design issues before many patches have been based upon the flawed design.

So how does this relate to Homeworld? Well, similarly, Homeworld 1 was an amazingly-designed game, despite its lesser graphics. There were many different strategies and tactics to choose from, there was plenty of room for player skill to have an impact on the game, etc. Homeworld 2, however, is deemed by many to be inferior (though still a great game). Subsystems, HW2 tactics, and the new research system were all great additions, but RNG weapons, the loss of tactics that scaled with player skill (formations, HW1 tactics), and things like “rush to Battlecruiser or kill them before BC” were disappointments.

Homeworld 2 was balanced. It had better graphics, and some interesting additions. However, its design was considered by many to be inferior, and therefore many players considered it to be the inferior game.

Now, I realize that HW:RM Multiplayer can most likely be balanced with the 4 races. However, I do want to stress that the most important thing initially will be to ensure that the design is good. Give us many strategies and tactics to choose from. Give us things to scale well with our skill. Make each ship have a role to play, depending on the strategies used. This is my plea to the developers.

My plea to the community is to be persistent on these design matters as the first priority. The StarCraft 2 community is riddled with balance complaints amidst those who knew about the design issues – this is partially what led to fixes to the design being so delayed, and instead just tweaks to units for the sake of winrates taking the first priority. Balance is important, but please, focus on the game being balanced around a good design, not designed around a good balance.


A well programmed “probability RNG” is statistically the same as a physical ballistics model. I don’t think the way it’s programmed in this game is likely that good, since you get these weird bending beams and bending bullet things, but still. Just because it’s RNG doesn’t meant it has to be bad.

I really don’t understand RNG complaints
Broodwar had lots of RNG, and was a better competitive game than SC2. You should know that, since you said it’s better.
Dota has lots of RNG while LoL has none, and is the better competitive game over LoL.

I don’t think the RNG in attack rolls actually makes nearly as big of a difference as the RNG in the unit behavior and flight modeling. Like when two intercepter squads fight and one kills the other while just losing 2 or 3 ships itself, that isn’t from the RNG on the chance to hit, which is really minor. That’s mostly because of the ridiculous RNG in the movement that let the winning squad get in like 3 times more attacks.

HW will never be competitive the same way Starcraft was. It just doesn’t require enough APM. There’s not enough “stuff to do”. It would be nice if the design was better though, yes.

Even HW1, having better design and balance, well there wasn’t much to do. You built a bunch of scouts, you put them in a sphere, and you moved them around hopefully better than the other guy and hit your aggressive defensive hotkey repeatedly, also hopefully better than the other guy.
If things went even, then you made a wall of covettes with another wall of repairers behind that. And you put your scouts in a few claws and ordered to defend them.
If it was still even, well frigs and caps vs frigs and caps were pretty straight forward. Both would try to use salvage corvs with support frigs repairing them. It came down to who either pulled off the capturing or sniped them better, or maybe something happened with a defense field.

So eh…
Though until people figured out the optimal things, I guess it was interesting. And the better player did usually win. It’s just, the best players all did the same thing.

HW2 had some good ideas with subsystems, and asymmetrical balance, but it was very poorly executed.

Comparing Homeworld and Homeworld 2, many people preferred the physics system over RNG. Now, RNG weapons as they are currently aren’t my main concern, though I believe the system could be better adjusted to work a little more like the Homeworld one (I saw a thread in the mod forum about it).

Does it limit the player skill? Probably not that much. Honestly it’s not a huge issue to me, just something that could be improved upon. I wouldn’t say BW was heavily reliant on RNG, it just had it in the game. But to be honest I would have rather had a flat damage reduction up cliffs instead of random hitting.

SC2 suffers from different micro-limiting (and even micro-punishing) RNG elements that you can look at JaKaTaK’s videos for. BW did not have those elements.

There was an established metagame in Homeworld, with the Scouts and things as you say – it wasn’t a perfect game. But it wasn’t like the “rush to BC”-type thing we are seeing. Fighters, Corvettes, Frigates, and Capital/Super-Capital ships all had their roles. While it might have been difficult to use different things, it was still viable.

But like I said, it wasn’t a perfect game. I would love formations back for their skill potential, but if Sphere formation is a problem, then get rid of it. I’m fine with improving on the design of both games, though in my personal opinion, HW1 had the superior design out of HW1 and HW2.

My main point isn’t “make HW2 like HW1,” it’s “focus on balancing around good design, not designing around good balance.” HW2 certainly had some new and interesting design features, but it lost features that HW1 had, and honestly I like HW1’s features better (and so does the community from what I’ve read so far).

1 Like

Don’t get me wrong, I agree with your greater point. Just thought the picking at the RNG was weird.
I tried to fix a bit of the design in my mod, but lots of it is locked in source and not moddable, and the other stuff… well I don’t have the drive when I have other things to work on. Modding support also isn’t good enough (Someone should be able to click to download a mod immediately and relaunch with it applied when attempting to join a game with a mod that’s on the Steam Workshop)

You may mean mine.

But HW2 definitely has some sort of “probability” factored into the RNG. I just don’t like how arbitrary it is. It’s not what you’d realistically expect for sure. But I’m pretty certain a fighter standing still is much more likely to get hit than the “base” accuracy.
And ultimately, no one understands how it works, which is pretty unacceptable. :confused: It’s like having some complicated algorithm for uphill miss in Broodwar, and keeping it a secret.

There’s also things like setting a delay on the damage based on distance, and nullifying the hit of pending hits when a tight maneuver is made.
There is a lot you can do with “dice roll” attacks that is still very resource and network intensive, but isn’t really truly random and will average out to working nearly the same as real ballistics. Like if you ran a replay where one used realistic ballistics, and another used a decent probability RNG with a few little “tricks”, you’d get roughly the same results.

I’ve done way more intensive things in “mmo” environments, so yeah.
I really like to avoid real collision detection, hitscan, ray tracing, and so on, but that’s not an excuse for the “rng” being bad. It’s a lot cheaper to take vectors, cones, velocity, and work out if it should hit, when it’s roughly the same result when you average a bunch of ships shooting each other.

1 Like

Yeah I saw you made an improvement, but I was talking about a post saying something about putting the RNG at 0, having it fire when pointing at the enemy, having the bullet follow that straight path, and putting a damage point when the bullet’s sprite or whatever makes contact with the ship sprite. I don’t know how practical that would be though, maybe there’s a better way to play with the RNG than that would be. Like similar to your mod perhaps. The other way would be a better emulation of a physics engine without being a physics engine, but I don’t know how possible that would be or if it would be preferable (though some people seem to prefer the physics engine).

A 3rd option would be to put 100% accuracy on everything and have different damage values against different ship classes – basically like what SC2 does – but that just doesn’t feel “Homeworld” to me. Maybe a hybrid where frigates and up have less accuracy and all strike craft are balanced around 100% accuracy – could be fun but again it still doesn’t feel Homeworld to me.

Ah that. I believe that comes at a considerable resource cost.

Like it might not lag when everyone has decent bandwidth, and decent computers. But you put one person in the match that barely runs HWRM as it is before that change, and suddenly it lags for everyone to keep back for their sim.

That function is designed to calculate a few dozen projectiles. It also has many unnecessary extras to the method to deal with acceleration, tracking, rotation, strafing, that surely makes it far slower than an actual ballistics function made to handle just a straight line, predictable projectile.

And in the end, like I said, you can make much more efficient code based on prediction and probability. Collision and such is extremely expensive. It’s much cheaper and generally the same result to fake it well.

Like look at that other game that just came out, Cities Skylines. There’s no collision detection on vehicles. It’s just some pathing based on flow over cost, or something along that lines. It’s much cheaper to have a vehicle map down cost under it to actually detect collisions, let alone possible upcoming collisions, for so many entitities. And you get lots of entities for lots of bullets. And Skylines isn’t multiplayer like HW is.

I played HW1, but I wonder if its bullets “physical ballistics” was actually that good. It used its own cheap tricks to work on computers that old and online.

I really think it simply needs a better probability RNG algorithm, and formations & tactics fixed.
You can make a predictive algorithm good enough to where you can just lie to people and tell them it’s physics simmed ballistics. Shrug.
People just don’t really want to believe how much repeated events in games (and life) average out like that. Yeah, it is satisfying to see a projectile fire in a straight line, hit something, and make it explode… I’m totally not against taking the missile function and trimming it down so it’s more efficient and better suited for ballistics. That’s actually simpler to do, programming wise, than to make a good and even more efficient algorithm to do the same thing based on probability.

1 Like

I think the main issue people have is they don’t know how much exactly their units are going to be worth – you don’t know if your 5 Interceptors are going to deal as much damage as the enemy’s 5 Interceptors. If you take the average, of course all Interceptors are the same value. But in actual game scenarios, such isn’t necessarily the case.

If you keep them alive long enough over several engagements (or a long enough single engagement, where damage values vs armor values play a bigger role), or if you have a very large battle near the end of the game, the effect is pretty negligible. But it’s the smaller medium-sized cutthroat battles where randomness is such a big factor.

Now, if the probability factor is high enough (like fighters vs an HC) or low enough (like an HC vs fighters), it isn’t a big issue. It’s the middle numbers that make it more random than not.

If formations and HW1-style tactics had more of a role, it would be much more based upon how the player controls their ships, taking some of the emphasis off of the RNG (though it would still be a factor).

Yeah there is a big issue with the “Damage” numbers on a ship’s card not being accurate. It doesn’t take into account accuracy.

A corvette could have 300 dps, but it’s only 10% accurate against something really making it 30. But it’ll show 300 on that card.

Not to mention, those stats don’t show up before you build things.

In HW1 is was much more straight forward. As far as I know, there were no armor and damage types.
What made a ship good or bad was muchly to do with its flight modeling and the arrangement of its guns/turrets.
Which was great in that you didn’t need to know thousands of hidden stats to know what makes which units good or bad against which.

If it were up to me, the whole “typed accuracy”(as in, a human has typed in what accuracy each weapon does against each subset of ship types), and almost as largely varying damage bonuses would be done away with.
Everything would do the same damage regardless of what the unit type was, except for very clear exceptions such as bombers doing a bonus against subsystems.

I could achieve more than 95% the same balance the game has now with more clear, straight forward numbers, but it’s a ton of work going through every ship and adjusting them, not to mention all the math beforehand.
But to change the whole accuracy thing, that can’t be done via modding as it has to do with a hard coded chance to hit algorithm.