Sunday, June 1, 2008

Combat Mission's Wrong Left Turn into the Uncanny Valley

This is a mirror of an article original appearing at this link:
Between 1968 and the year 2000, over 100 separate board wargames were released depicting tactical level (meaning individual units depict platoons, squads, or individual men) combat in the 20th Century. Since then, several dozen more titles have appeared - some new ones, mostly modular in concept with multiple add-ons for the same game system. A review of these games gives one an opportunity to see how different designers have tackled design problems.

From 1985 to present, there have been several dozen different PC titles to portray the same thing - i.e. 20th Century tactical combat.

The history of the Combat Mission franchise is well known, from release of Combat Mission: Beyond Overlord in 2000 through its two successful sequels to its latest release with a new game engine, Combat Mission: Shock Force. Some points to note - it was certainly not the first tactical, squad-based game to graduate to the computer. Under Fire was on PCs as early as 1985. It was not the first 3D tactical squad-based game, Muzzle Velocity had that honour. It was, though, the first game to combine squad based tactics, turn-based planning, 3D graphics, and a WEGO system that was at that time revolutionary - and from what I understand, is still unique. Other games of the genre seemed inferior by comparison - Steel Panthers was far too "gamey" - you could draw fire by taking advantage of the turn-based system unrealistically since everything was sequential and strictly IGO-UGO, and Close Combat was too closed-ended - making maps was not something casually done, and on-map forces were restricted to over-strength platoons confined to a few hundred metres of terrain.

Designer John Hill - later famous for Squad Leader - published "Ten Rules for Playability" in the early 1970s. Part and parcel to playability was the concept of scale. You can't simulate Napoleon's invasion of Russia using 40 metre hex maps. Games were designed to be played in one sitting (for the computer, this is debatable, since we can now save games without having to worry about cats and kids, but I think a good rule of thumb remains that a playing time of a couple of hours for a single game is about right). His inappropriately named Squad Leader, released in 1977, had the player control company-sized groups of men, divided into individual pieces representing 10-15 men apiece. The system was highly abstract - "Design for Effect" was his mantra - and it worked. It was simple and it was fun, and it provided the framework for a dizzying amount of more complex rules later, which have again been simplified down to the three Starter Kits - ASL Lite, if you prefer.

When Squad Leader came out in 1977, it abstracted a whole lot of processes that another squad level game released a year earlier had rendered in painstaking detail. It went on to become Advanced Squad Leader, which now has 13 modules, 2 deluxe modules, 7 historical modules/studies, a rash of TPP imitators, over 4000 printed scenarios and possibly the most commercially successful board wargame to date.

The game that came a year before it was Tobruk. Aside from being an ugly little game with flat terrain boards and an uninspiring choice of theatre, it tracked every single tank's shot in painful detail. And it was so labour intensive that almost no one played it. Publisher Avalon Hill thought so little of it, they actually gave the rights back to its inventor, the late Hal Hock, and said "thanks, but no thanks." The game sat around for 20 years until Raymond J. Tapio rewrote it heavily and made it playable. In other words, more like Squad Leader. It is now on the market as Advanced Tobruk System, but it seems to me that Tapio still makes money selling TPP ASL stuff too because ATS won't do it on its own.

Combat Mission
Wargamers gagged for Squad Leader for the computer for years (and when Hasbro abused the rights to the franchise name with an absurd man-to-man game by that title, the gagging turned to retching, but I digress), and Combat Mission delivered.

What did Combat Mission really do?

Well, it provided the same painful detail of tank penetrating hits and ammunition that both Tobruk and Squad Leader did (not to mention other games I haven't mentioned here, like Yaquinto's trilogy of 88/Panzer/Armor who had great fidelity in the armour modelling realm also), but with the advantage of a computer doing the calculations so that the player didn't have to. The newcomer was a bit lost, but it didn't take much intuition to figure out Tiger=good, Sherman=bad. And learning was part of the fun.

But the whining about the graphics! The blocky 3D cavemen running around like refugees from a Jack Kirby comic book were positively grotesque. The twitch crowd whined to high heaven about camera controls and wanting to see every man on the map.


This is how a squad of 10 men is represented in the original Combat Mission trilogy.

So what does BFC do?
They listened, of course. For some reason, tracking individual bullets became a matter of priority. But let's stop and look at the decision for a second. A single Mimimi LMG fires 1100 rounds per minute of 5.56mm ammunition. An M4 carbine can fire 500 rpm. That's a lot of firepower for a squad. More important - that's a lot of trajectories for the engine to trace.

1:1 modelling became a buzzword on the BFC forums after the release of Combat Mission: Afrika Korps, the final title in their original engine trilogy, as if it was some kind of Holy Grail or something. But look what it's resulted in.
  • reams of data for the computer to sift through. So much data, that WEGO TCP play was ruled out by the developers as technologically impossible, who had threatened to leave out PBEM, too.
  • individual bullet tracking is in, but what about the brain behind the gun? There is no data on personalities; every soldier in every squad is a cookie cutter with the same physical - and mental - attributes as his mates.
  • area fire and movement commands still seem to snap to an 8 metre grid, called "Action Spots" by the developer - there are no visual cues in the game as to where these Action Spots actually lie.
  • map scale is reduced; where one could play comfortably on 4 square kilometres in Combat Mission: Barbarossa to Berlin, high end systems chug away on maps smaller than that in CM:SF. Real time control of anything more than a reinforced platoon on a "medium" map is probably impossible in most situations anyway if anything like actual tactics are going to be employed against a human opponent using a dynamic defence.

Whose idea was this anyway?
Design for Effect was elegant, and it worked. It worked in CM:BO, CM:BB and CM:AK, too. There was no great "need" for 1:1 modelling except among those with no real understanding of what a wargame is supposed to be. BFC understood when they made Beyond Overlord. People scoffed at those ugly 3D models and didn't "get" why 3 soldiers really stood in for 10. But the fans did - and hey, so did the reviewers. They LOVED CM:BO and the next two titles.

The Uncanny Valley
Is 1:1 the wrong way to go? I don't really believe so. Eventually, and soon, hardware issues won't be a problem with this level of modelling. But look at any "company-based" wargame in the field of board wargaming, and none of them have 1:1 modelling. Not one. Firepower is grouped at the squad or platoon level; that trend started with the very first tactical board wargame in 1969 and continues to the present day. What is gained in CM by tracking individual bullets? Nothing - in fact - it may even be a design blunder. Can't imagine counting up individual men on a Squad Leader map to count firepower factors. Just because we have a computer to do the computations for us, it doesn't make the design philosophy any more "correct."

I do know I still play the original CM titles and enjoy them. It doesn't take 1:1 modelling to engage you - it's the gameplay. In fact, I think we've reached the Uncanny Valley in CM, where it is so close to real life as to heighten expectations about game performance unrealistically, but not close enough to be convincing in actual play. Seeing squads stack up outside a door is great - but why did they run around the building to do it, and why didn't half of them just jump through the window?

This is a squad in CM:SF; sure, there are the right number of bodies on the ground, but how come they can't shoot in different directions?

Of course, I don't know if a simple update to the CM:BB graphics would have been satisfying, either (i.e. three man representations with the new 3D models).

But what about stuff like convoy driving (we still don't have a "follow me" command)? CM:SF has hugely powerful artillery modelling now - but the philosophy in CM:BO was that the games all started AFTER the big artillery barrages. Again, this is a step in the other direction from the original philosophy of abstraction - a philosophy that worked well.

CM:SF is necessarily in the middle ground where it will please no one. The level of abstraction that was such anathema to the Operation Flashpoint fans (I am one) is gone, but it isn't replaced with high fidelity yet. Squad members are nameless, faceless droids who are largely stupid. That every bullet they fire can be tracked doesn't mean much if every man aims at the same 8 metre grid.

The scenario editor/Tac AI has been revitalized - to the point that we are incapable of surprising ourselves. No random maps or forces, and the soldiers do whatever they're told to - and no more. The great appeal of random battles is gone. Again, there was abstraction in terrain and force mix, but because BFC is now saying that the map editor is more "realistic", we can't have "fun" in the form of random maps.

Whoever said "realistic" was the way to go? It sure wasn't BFC when the community posted for six months about the Sten Guns in the British OOB for CM:AK during the patch releases. They're still there.

Whither Next
It may be too late to turn back, but I think embracing some of the limitations of the past may be a good thing. Abstraction is good.

When Firepower came out in the 1980s, it was a man-to-man wargame that modelled every individual small arm from 1945 to the present, and had rules so detailed that you had to use arrow counters to mark which side of a tree your dude was standing on. And it wasn't nearly as much fun as Sniper! which was rereleased by TSR at about the same time, a reprint of the earlier SPI game from 15 years earlier, which had abstract values for weapons - and no little arrow counters.

A PBEM correspondent of mine wrote to tell me that he felt BFC's adoption of a RT option meant that instead of being the industry leader in WEGO, they now specialized in nothing, and were mediocre at both WEGO and RT. I think the 1:1 modelling has had a huge impact not just on computer hardware performance, but on perceptions of the game, design philosophy, you name it. You see that many guys on the screen you expect different things from the engine - stuff that has been explicitly stated aren't being delivered yet (like bullets not going through stone walls because of abstractions in the terrain mesh). The official forums talk of "gamey" tactics such as driving Strykers into buildings in order to let infantry dismount in safety. These are abstractions that one wouldn't expect to find in a 1:1 model - and speak to the heightened perception that comes with this level of modelling that the new engine does not in fact pay off. (Of course, flooding a position with armoured halftracks was also a problem in Squad Leader, allowing a "gamey" player to offload infantry in the vicinity of the enemy in safety - something that was addressed in the ASL rules.)

My Final Word
Panzer Command by Matrix Games may be the up and coming series to watch. If nothing else, the developers are actively engaged with the community and working on putting in features as requested. The developers actually play their own game, also, which may be a key difference between their approach and that taken by their competition. Time will tell.