A possible systematic approach to balancing the fleet ship types

[Suzera]

If it makes anyone feel any better, professional game developers (AAA or indie) aren't any more agreed on this issue than the players are.

They really do different things in my experience. Math balanced games tend to have more variety in viable optimal things to do with games changing during play. Throw and see what sticks tends to lead to shortcuts, best things, exploits, but a wild variety of weird things to try, until you distill it down to the best or just what you feel like if you stop caring about winning or keep refining it forever with endless amounts of effort or keep the scope very narrow. AI War certainly has room for both. Ships like the MRLS, Tank, Raider, Triangles, Bulletproof, Cutlass, Flagships, Boosters and many others lend themselves well to a math model for balance. Things like Etherjets, FFBs, Autobombs, Maws and Parasites tend to entirely defy mathematical models unless you have a fully mathematically simmed game, which AI war most certainly is not. This is partly due to the fact that their biggest power variable is ALL THE OTHER SHIPS IN THE GAME. Then you have some ships like Lightning Shuttles, Shredders or Vampires whose effectiveness varies by difficulty or other player selected variables you do not want to close off due to their special abilities.

[x4000]

And then you have ships that open up new player tactics, which can't be mathematically quantified in any concise sense.  In general: yeah.

[Suzera]

You could probably bring them all into a unified model, but it's probably not worth it for the more gimmicky ships. The only one that would probably be nice to bring under the umbrella somehow is Shield Bearers, since it is a kind of coreish mechanic, and maybe make another type or two of them. I can't think of a nice way to do that really quick though, unlike the more conventional ships in the game.

[Red Cossack]

Use more math to balance but keep an eye out for fun and the heart of the game and who can lose?  People tend to react in a very binary way when they see something like this, either loving or hating it instead of saying "Hmm, after some fine tuning it could indeed be very useful, but I have the following reasonable reservations..."  Unfortunately, the moderate is often the one who has the hand grenade land in their lap from either side. 

[Sizzle]

An interesting aside:

Perhaps a GOTM (game of the month), or weekly game where players are presented with a common save starting point - that has their starting unit selected for them.

Essentially a vehicle for "Everyone try this" this week.  A lot of the unlocks suffer from never being tried because everyone already has their favorite unlocks already picked out.  Feedback we do get on certain units is usually an outlier until a big topic on "unit x needs a buff / nerf" arises.  Another thing that many units lack is a "what the heck is this useful for?" description. This also helps with that.

- bonus points for posting zany new ways to use the unit of the week

[x4000]

An interesting aside:

Perhaps a GOTM (game of the month), or weekly game where players are presented with a common save starting point - that has their starting unit selected for them.

Essentially a vehicle for "Everyone try this" this week.  A lot of the unlocks suffer from never being tried because everyone already has their favorite unlocks already picked out.  Feedback we do get on certain units is usually an outlier until a big topic on "unit x needs a buff / nerf" arises.  Another thing that many units lack is a "what the heck is this useful for?" description. This also helps with that.

- bonus points for posting zany new ways to use the unit of the week

I think that's an excellent idea.  It's not the sort of thing that we have time to manage directly here from the Arcen staff, but having that be community-organized would be great.  We had something like that way in the past, but it was short-lived due to bickering about who got to decide on an "official" GOTM or GOTW or whatever it was.

At some point I'd like to get some succession games going, to -- which I would be more part of -- but again for the moment time is the limiting factor.  Any volunteers for getting some GOTM/GOTY organization going?

[RCIX]

Well, the issue i have with a mathematical model is that there are soooo many exceptions and sooo many unexpected things that you're just as likely to spend more time coming up with a decent enough model as you are just manually working on it. For instance, in Supreme Commander 2, you have to deal with shot travel time. This makes several units less useful than might be expected (with micro).

Edit: we tried that GOTM thing a while back, but IIRC it just kinda petered out. :/

[Suzera]

The big things that matter are polycrystal, artillery, light, structural and ultra-heavy. Those are your biggest foes that are in every game. If it isn't one of those, it is chancey. Ultralight is useful for raid starships if the damage * multiplier is above 150k or so though.

You can add in a microing fudge factor if needed, but ideally you balance it and add features so that microing isn't really that useful. The farther apart the various speed and ranges are (on a logarithmic scale not a linear one) the less harsh any potential microing becomes.

Ideally the game would just do it for you, and it's kind of there to some extent on some things like Raptors or Bombardment Ships. I wouldn't mind seeing ALL ships do that to their various ranges. I have other things I want to talk about instead though than start a riot about removing ship microing inefficiency from the game.

It is one thing I REALLY liked about R.U.S.E. All the trivial things like making sure troops that are just going to needlessly die in one shot to tanks run away the game does for you. All that's left is to make the correct decisions, counterdecisions, counter counter decisions and so on. It's more like chess than Warcraft 3. It was very different and a breath of relief from micro-heavy games.

[Wingflier]

An interesting aside:

Perhaps a GOTM (game of the month), or weekly game where players are presented with a common save starting point - that has their starting unit selected for them.

Essentially a vehicle for "Everyone try this" this week.  A lot of the unlocks suffer from never being tried because everyone already has their favorite unlocks already picked out.  Feedback we do get on certain units is usually an outlier until a big topic on "unit x needs a buff / nerf" arises.  Another thing that many units lack is a "what the heck is this useful for?" description. This also helps with that.

- bonus points for posting zany new ways to use the unit of the week

I think that's an excellent idea.  It's not the sort of thing that we have time to manage directly here from the Arcen staff, but having that be community-organized would be great.  We had something like that way in the past, but it was short-lived due to bickering about who got to decide on an "official" GOTM or GOTW or whatever it was.

At some point I'd like to get some succession games going, to -- which I would be more part of -- but again for the moment time is the limiting factor.  Any volunteers for getting some GOTM/GOTY organization going?

I wouldn't mind doing something like this, I'm sure there would be plenty of volunteers willing to make games for the community.  Should I make the thread right now?

[Sizzle]

An interesting aside:

Perhaps a GOTM (game of the month), or weekly game where players are presented with a common save starting point - that has their starting unit selected for them.

Essentially a vehicle for "Everyone try this" this week.  A lot of the unlocks suffer from never being tried because everyone already has their favorite unlocks already picked out.  Feedback we do get on certain units is usually an outlier until a big topic on "unit x needs a buff / nerf" arises.  Another thing that many units lack is a "what the heck is this useful for?" description. This also helps with that.

- bonus points for posting zany new ways to use the unit of the week

I think that's an excellent idea.  It's not the sort of thing that we have time to manage directly here from the Arcen staff, but having that be community-organized would be great.  We had something like that way in the past, but it was short-lived due to bickering about who got to decide on an "official" GOTM or GOTW or whatever it was.

At some point I'd like to get some succession games going, to -- which I would be more part of -- but again for the moment time is the limiting factor.  Any volunteers for getting some GOTM/GOTY organization going?

Is it something you'd want to select which ship is the starting one and leave the other stuff to the community?   You could get more of the ships you want looked at by editing the ARS unlocks (if such is possible).

[x4000]

It's not possible to edit that sort of thing at the moment.  Generally I'd suggest setting up a specific savegame that everyone starts from, since seeds, etc, have such a wide-ranging effect.  But that's a good question for what the community prefers to do for this.

[Suzera]

Well, the issue i have with a mathematical model is that there are soooo many exceptions and sooo many unexpected things that you're just as likely to spend more time coming up with a decent enough model as you are just manually working on it. For instance, in Supreme Commander 2, you have to deal with shot travel time. This makes several units less useful than might be expected (with micro).

I missed this post somehow. You can certainly factor in bullet travel time in SupCom2 into a math model without too much trouble. It's a type of accuracy that is a function of distance, shot speed and target speed of things you want it shooting at.

The biggest fudge factors quicky off the top of my head in AI War are the randomized formations (there are ideal ways to form up the ships and it is a really big pain to do in AIW), difficulty level changes, ship cap changes (armor being outside the damage multiplier changes things as does AE unit effectiveness), map planet count changes, the gimmick ships and the randomized units the AI gets every game. Two of those are removeable without removing options. The ship cap scaling can be fixed by changing the formula to be the same for all ships no matter ship cap EXCEPT for the AE unit damage amounts (and only the raw damage amounts) which are increased inversely proportional to the ship cap count change. Half ships = double AE damage. Armor will still screw with that a little though I guess, but it would be generally close, particularly for bombers. Lightning stuff is all super armor pierce anyway so at least those are taken care of. I don't think any AE units are aimed for high armor targets either. Proper formations and perfect automicro are more work, but still doable.

[keith.lamothe]

Well, the issue i have with a mathematical model is that there are soooo many exceptions and sooo many unexpected things that you're just as likely to spend more time coming up with a decent enough model as you are just manually working on it. For instance, in Supreme Commander 2, you have to deal with shot travel time. This makes several units less useful than might be expected (with micro).

Right, I'm not actually trying to achieve the goal "balance the game" with this model, just get to the point where very few (preferably none) of the bonus types cause you to slam your head into your desk if you get them from an ARS

We might even be able to get a bit further than that, but that's the current goal.

[Wingflier]

I'm pretty close to a mathematical equation that establishes a cost:effectiveness ratio based on the fighter.  Out of curiosity, how many hull types exist in the game?

[keith.lamothe]

I'm pretty close to a mathematical equation that establishes a cost:effectiveness ratio based on the fighter.

I initially started with that, and it's interesting info, but the fighter isn't necessarily quite where we want to balance, and extrapolations from it could be pretty brittle.  But thanks for working on that, it could be pretty useful for some stuff  

Out of curiosity, how many hull types exist in the game?

Hull type as in "UltraHeavy", "Polycrystal", etc?  They're all listed in the OP; I'm too lazy to count atm as I need to run

Or do you mean how many bonus types (plus 3 for the triangle)?  Or just how many types of ships at all?