Wiggins dropped by, to see what we were chatting about in the 12 October H&I post where Ry linked to Opposed Systems Design.
He left this comment.
Thanks much for the link.I was being a bit harsh, so let me clarify myself.
Mathematical laws of chemistry describe immutable dynamics of the physical world. I question whether combat has any similar immutable dynamics that are waiting to be mathematically modeled. The difference lies in the thinking adversary inherent in war, as opposed to the stable relationships of chemistry.
Maybe the mathematics of dynamical systems offers a way to account for this - I think this is why chaos and complexity theory get folks so excited - but I haven't encountered it yet.
My second post didn't claim Lanchester wasn't useful; it argued that it was being used in situations where it wasn't appropriate. The question wasn't 'is Lanchester useful,' it was how often is it useful and in what cases? I was voicing my concern that Lanchester is rarely useful and sometimes used inappropriately. Not because it helps, but
because it's something concrete we can work with.... of course it would have helped my argument if I had, I dunno, had an actual example to back that up...
I started to answer in email, then decided - "Hey! I've not done a geeky wargaming post for a while! No time like the present..."
I was voicing my concern that Lanchester is rarely useful and sometimes used inappropriately. Not because it helps, but because it's something concrete we can work with.
... of course it would have helped my argument if I had, I dunno, had an actual example to back that up...
I can help, and give you an example.
I didn't weigh in too much on this discussion, as it cuts to the heart of what I do for a living, and there are issues of classification, employment, and intellectual property rights that make me stand mostly mute.
But I will say that Lanchester can be useful, at it's most macro and most micro levels, when you are comparing forces which can be generally assumed to be at parity on the issues in contention. Oddly enough, that's pretty much all that Lanchester was proposing.
Such as modeling force-on-force from a hardware perspective, to examine the effects of the hardware. It can also be useful for examining organizational structure and doctrine - again, essentially positing a peer opponent. If your hardware wins, and your doctrine wins, and your hardware *and* doctrine wins in a Lanchester world, you are going to probably fare well, within the confines of entropic events and effects.
Of course, that's a very narrow set of bounds. And the consequences of bad assumptions about parity (and *your* basic competence and morale) are huge.
So, when the Army designed it's first sims for that kind of thing, it used Lanchester because we were assuming that at most levels, the Soviets, from a morale and basic professionalism level, were peers. So, we wanted to know if our tanks, tied to our doctrine, would prevail on the North German Plain (i.e., Fulda Gap). And if they didn't, what tweaks or wholesale changes would make a difference? And that sort of thing was generally confined to analytical work on those levels. Examples are VIC at the macro level, and Janus at the micro.
Of course, problems surfaced. Trainers wanted sims too. And if you've ever sat down at a Janus workstation, it looks (especially back in the day) pretty clunky, and the graphics are nothing to write home about, but it sucks you in almost immediately, if you're a trainer or wargamer.
I was in on the leading wave (heck, I helped lead it, at my level) on integrating those sims into training.
And, of course, one of the first things we did was try to build scenarios of historic battles. And found you had to tweak the crap out of things (something that board game designers like Dunnigan and crowd have know for decades) to generate historical outcomes. Not only because two regiments of muskets is *not* the same thing as Chamberlain's 20th Maine and Oate's 15th Alabama, but the digital terrain is not the same as the historic terrain. The man matters, and terrain representation is critical.
It's *not* when you're trying to compare tank vs tank, doctrine vs doctrine, on common terrain (the digital battlefield).
Danger, Will Robinson! Danger!
So Wiggins is correct - you have to understand your tool. One of the things I did as the Chief of a simcenter was sell my facility as a C2 trainer. Because you can use a Lanchester sim to drive training where the focus of training is the military decision making process. But quickly, commanders seized on it as a wargaming tool for things like NTC rotations, and it was a never-ending fight to get them to understand that the digital terrain does not exactly match the NTC dirt, and micro environment matters in fights at that level. That a plan that works in sim at Fort Sill, or Fort Hood, or Fort Riley, isn't guaranteed to work at Fort Irwin, on the real dirt, against the real OPFOR. But that the practice of your staff planning processes would pay off. You'd still need to make your plan based on the actual ground when you got there.
I promised an example of using Lanchester inaptly.
Desert Storm. Remember the predictions of 10K casualties? Say hello to my little friend, Janus. The modeled Iraqi Army was using soviet-derived equipment, with a modified 1950's British doctrine, fighting a defense from prepared positions where they'd had time to really dig in. And they were being commanded by a competent Red force team. Against a competent Blue team. And while we won, we got hammered doing it, because Lanchester is a bloody algorithm. And it makes a huge difference, when your data for your tanks is conservative - and you can actually kill a couple hundred meters farther out than you were able to in sim (as we found out when they modeled the Battle of 73 Easting and had to tweak things to match the historical record).
If you're going to err, err on the side of caution, I say. But you have to understand the limits of your hardware, architecture, design intent, and algorithms as you make choices.
There is much work that has gone on since then in building the mission planning and rehearsal tools, and they're pretty mature and robust for naval purposes and for aviation, both of which can largely ignore the effects of micro-terrain, because the ocean is flat, and the ground is effectively flat to fast movers and many helo missions, and is platform-based, which is much easier to model from an AI perspective.
Ground combat, with micro terrain and micro people, is still an on-going and evolving challenge.
And when you try to model the Current Operating Environment, boy-o-boy is Lanchester not your friend. But it was never meant to be, either.
So yes, I agree with Wiggins - people with the best of intentions, especially under a "I have to have it now" imperative, can badly mis-use Lanchester-based models, to their grief.
Okay, time to put this to bed and head for the BDE TOC. Have fun chewing on this, I probably won't be able to respond to anything until tonight.
14 Comments