Algorithmic balancing |
While I don't know what is actually involved in such a calculation (I don't have anywhere near the higher math required), I have had an idea bouncing in my head that I thought I'd share.
I was wondering what possibilities there are in creating an algorithm to calculate the various aspects of components in a grouping (e.g. weapons, armors, facilities, etc.) and coming up with an equation that plugs in all effects and turns out a good balance of size, cost, effects, level changes, structure as it all pertains to the parts of the game it has an impact on.
I know it's a bit exorbitant for anyone but a mathmatician to attempt figuring out, but I thought I'd throw out the idea and maybe get a discussion going of ways to balance things instead of having facilities all handwritten to new sizes based on opinion only or weapon sizes for similar reasons. Most weapons being all 30kt seems too flat.
Thoughts?
Re: Algorithmic balancing
Agreed, the headache involve is way harder than you would think, even mathematicians would think twice about it. For example, even worse than range is how to value the damage type... how much should we value shield/armor piercing and shield/armor 2x/4x effectiveness, or how valuable are those engine/weapon destroying weapon... etc. All those are dependent upon yours and the other player's strategy, and since that can change, there is no real value that you can just easily assign to make them balanced.
The problem is, a lot of times, one player's definition of balance is another's definition of imbalance. If we all agree, then we could make a formula for everything, but getting everyone to agree is the hard part.
Re: Algorithmic balancing
I'm thinking about it et least for year (and i started thinking about this becose of different game) (and only for whole units (whatever it is swordsman or space ship), not on component level).
So far, my only result (and qute obvious one), is that if two units have same speed and range (and nothing like pre attack dellay, changing magazines ...) then you compute afective dmg and hitpoints of both, and if dmg*hp is same on both sides, they should be balanced.
If they don't then lets say dps1*hp1 = X*X * dps2*hp2 then first unit should cost about X times as second.
I say about, becose weaker units will die faster than strong one so their dps will go down, where dps of single strong unit does not go down.
But what makes things worse is, that players can use multiple unit types at once, so even if A>B>C>D>A A+C can be both > and
Re: Algorithmic balancing
You could just use fuzzy maths and assign it a max and min value for each property, and then randomly select from inside that(or if you want to get really scarily complex, assign it inside that range based on previous ship designs and how successful they have been for tonnage killed vs tonnage lost), that way you end up with a variable weighting system that will result in the AI periodically trying new weapon mixs and will prevent it from getting overly static in it's designs.
Re: Algorithmic balancing
Balancing requires, at the very start, a number of arbitrary values to be chosen. Range depends entirely on speed, for example. If you change one and change the other by a proportionate amount, nothing really changes.
Damage and firing rate requires an arbitrary value to be chosen for the base amount of dps, to which other weapons are compared. Armor at the very least needs to offer some advantage over taking more weapons, so there are some constraints based on that, and likewise it should not be so powerful that it is most beneficial to take only armor except for one weapon and merely the most efficient amount of ordnance storage (more weapons or more ordnance storage) and outlast the opponent.
Damage types are at most comparative. There is no difference between twice the base damage and half damage past shields and normal base damage and double damage to shields, for example.
Even if a "perfect" formula is found, one would still need to insert several arbitrary seed values, some of which will determine the effectiveness of different components depending on the desired flavor. That is, powerful shield depleting and armor skipping weapons will favor a rock-paper-scissors relationship, while making them effective but not overpowering (against their type) would make them more generally useful.
Re: Algorithmic balancing
Interesting topic, but i think going with several specific AI designs is better than trying to come with a ONE SIZE FIT ALL formula that the AI uses to make it's ships... Also, the way the AI is programmed, it doesn't know if one attack ship is better than another design of attack ship. When the time comes to upgrade, all obselete attack ships get upgraded, even if the newer design is worst than the previous ones... Also, in stock, the AI used the biggest hull availiable when designing new ships, older designs with smaller hulls are no longer upgraded.
Re: Algorithmic balancing
I don't think that's what the OP is trying to do. I think he wants to balance all the weapons, so that no matter what path you or the AI take, it will always be balanced because the weapons themselves will be balanced. The idea is great, but I'm not so sure if it's practical to pull off using a single formula. I'm not saying it's impossible, but you will probably have to do plenty of play testing to make sure they work as you want them. Which, is pretty much what most people do when they rebalance the game anyways, although we don't use a single formula, we do have a certain balance in mind, which is pretty similar to fuzzy math. In this case, trying to come up with a magical formula for all the variables would just end up wasting a lot of time that should be spent on play testing...
... on the other hand, if the community can come up with some ideal balance for the game, it might make things a little easier for everyone trying to do so.
Re: Algorithmic balancing
But then you have racial variables that throw away any technological balance the game may have. And last you have bugs that can change everything until they are fixed. With IRM, i did a lot of balancing with my tons of weapons and i believe i have found a somewhat acceptable balance. But players are great at doing min-maxing, combining racial advantages and abusing of the numbers in a way i never though of. Sooner or later i end up changing the balance again...

Re: Algorithmic balancing
You can usually use some sort of formula as a guideline to get most items in line and then refine them with feedback from other players.
-----
[LP] Re: Algorithmic balancing and suggestions
Here is my line of though on weapons from my moding experience from SEIII and SEIV:
Initially it would seem that this is a simple calculation between damage and Rof (rate of fire) i.e.
Consider:
10kt pea shooter I 20 dmg @ 2sec RoF
It is simple to come up with the variant:
10kt fast pea shooter I 10dmg @ 1sec RoF
The problem comes from the first shot. Since all weapons are charged(i.e. fire immediately when contact with the enemy is established) The first shot does all of its damage immediately. The 10 extra damage might destroy components, So it's a full 1sec duration of 10kt worth of components that is lost on the enemy ship. So it comes out that our fast pea shooter is weaker
You might think that if we had a weapon charge period (i.e. enemy comes in range then the weapon charges for 2 sec then it fires) the components would be balanced. However in that case the fast pea shooter is more powerful since it will subtract 1sec worth of 10kt components on the enemy ship.
You could balance that by making "pea shooter I" deal say 17dmg, that's the reason the wave motion gun deals far less damage that it's tonnage would suggest in SEIV.
One would also say:
"But wait, you shouldn't complain about them being the same size and having the same RoF. The reason is simple: weapon mounts. It could be very simple if we could alter all the aspects of a weapon with mounts."
Think about it: using mounts you can change (among other things):
Adjust the Rof
Adjust Range
Adjust the damage dependency on range
Adjust ordnance
Adjust supplies used
Adjust tonnage
Adjust structure
Why not going one step further?
Adjusting damage type
Adjusting weapon type
Adjusting Damage area: (I would pay $$$ for this)
single target,
AoE(Area of effect) at single target,
AoE centered on Ship,
AoE full sector, (in and out of combat?)
AoE entire system, (in and out of combat?)
Arc Blast AoE (macross cannons anyone?!?!)
If we had mounts like that then even a simple "Laser I" weapon could become any weapon by applying a mount (and what if we could apply multiple mounts? (weapon designs as ships?)) There would be no need for any other weapon beyond a more powerful version of "Laser I". or you would go like this:
laser is damage 1-20
plasma is damage 5-30
neutron is damage 10-40
antimater is damage 15-50
You would create the following weapons ingame:
"Laser Blast cannon I" AoE at target mount
"Armor piercing Laser I" Damage type mount
"Homing plasma Missile I" weapon type mount using missiles and plasma tech
"Laser storm I" Arc Blast mount using laser tech
"Mass neutron Emission I" AoE centered on Ship mount using neutron tech
and so forth
In the end all that maters is what I consider to be the identity of a weapon in my mod, not the balancing, I will get that eventually. All mods need convincing and intuitive weapons and components with good sci-fi physics explanations and backstory. That's the hard part, and we shouldn't expect much inspiration from the stock game because it has to follow the Space Empires theme of weapons, looks and lore.
Anyway that's all and sory for the long post.
Gnosis.
Re: Algorithmic balancing
Another obvious thought from me:
You should spend same (measured in most critical resource) for offense and defense. Why?
Lets say you spent 100 in both (so you have 100A and 100D (1A does not need to be equal to 1D)).
So 200R worth balanced ship have 10.000 AD
You decide to spend another 200. If you spend it all on attack, you will have 300A 100D.
300/100 ship have 30.000 AD and cost 400R
but
200/200 ship have 40.000 AD and cost 400R
so its better to balance spending in offense and in defense




Re: Algorithmic balancing
Its possible, but it is very difficult. The problem is some effects aren't easy to calculate properly. What value do you put on range, for example?