Balance in 40k

Hellions look pretty badass, but without the Baron...?

40k isn't perfectly balanced. That much is common knowledge; hence the amount of verbiage on forums across the internet about 'which unit is better, hive guard or pyrovores', 'how do I make my list into a super killy obscenity that Stalin would think was overpowered?', etcetera. We've all been there.

But does it have to be that way?

The common belief among many people is that 40k is deliberately unbalanced, so that Games Workshop can sell more of whatever cool new model that they have just released. This is probably being exacerbated by the current practice of releasing a random 'huge' unit for every codex, and making them generally pretty competitive.

To be completely honest, if this is their strategy, then I have no problem with it. They need to make money, like any other company, and it hasn't made the game unplayable by any stretch of the imagination. So if this is the plan, cool. Just don't try to pretend it isn't.

But is it really? I don't know that the Nephilim would be exactly flying out the door. Centurions aren't exactly an automatic take, either. So either it is their strategy and they're not all that great at it, its a smokescreen, or option three, GW aren't really sure how to balance the game after all.

There are those who will say, "of course you can't balance it, it's just so big and complex", but I disagree. With the wealth of tools available for generating and processing data, I think its definitely possible to balance every unit in the game against every other.

Now, I'm not advocating for units to get blander. Obviously there would be perfect balance if the only unit available was bolter armed tactical marines, but it would also be pretty fucking boring. Rather, I believe that the wealth of variety that exists can be balanced by careful analysis.

We have the tools. Look at business intelligence solutions, like Cognos or SAS. Do we have the guts to do the work?

There are a few dimensions to consider. First, obviously, is 'killiness'. This is probably the easiest to compute individually, but also requires a computation across every unit type in the game, so not a simple job. But for the majority, whose special rules don't impact their raw killing power, it is straightforward enough to be automated. So, we can assess against some arbitrary scale the relative killiness of different units for their points, and also their propensity to be killed, or resilience.

Clearly, there will be units that are more effective against some targets and almost useless against others. Incubi will be super killy against Space Marines, and useless against Dreadnoughts. That's cool, its part of the variety we like. But there's no reason that we can't boil the killiness statistic down to three broad categories: vehicles, monstrous creatures, and everyone else.

So after that, we have stats on how killy and how tough a model is. We can impose a basic points cost on it, relative to every other model in the game. Well, after we assess the relative worth of each contribution. I'd suggest plotting the relationship of the median unit against each of the three categories, and units that perform significantly better than expectation against one class can have their points adjusted up, and worse, down. Again, careful calculation of a model could produce some very accurate results.

Next are the other dimensions: mobility, scoring ability, buffs, range, and other unusual factors. These would probably require a bit more work, as there value is more about the structure of the game and the way it is played, than a raw probabilistic output, but there is no reason why they couldn't be calculated. I've done some mucking around with that previously, for example.

Probably, there would need to be some monitoring of games to assess the calculated values of mobility and scoring. Imagine if Vassal, or a tool like it gave the ability to track the movement of every unit, relative to the position on the board, and assess its utility to winning the game. That would give enormous power to test models and observations. Failing that, run a couple of GTs with video cameras installed above each table. Get about a thousand games, analyse each, and test and validate the dimensional models. Not cheap, I grant, but not difficult either. Amoritise the cost with some entry fees and you're good to go.

So after we have the value of mobility range, etc, we apply it has a multiplier for every unit. We now have a points system that ranks every model in the game according to its relative utility. Admittedly, this might turn out things like 13.28 points per space marine, but this can be addressed with a bulk price, eg 10 space marines = 133 points, or 5 space marines = 61 points, +14 points per extra. The rounding errors will even out in the wash.

It'd have to work better than the Studio Eyeball method that currently seems to be in vogue.