Two naval classics, and the line between them

Years ago I set out to do something that sounds simple and turns out to be almost impossible. I wanted to build a database of Pacific air-to-sea actions every time aircraft went after ships so I could work out how effective the platforms actually were and, maybe, build a better combat table than the games I owned.

I gave up. Not because the work was hard, but because of what the work told me:

 

“there just is not enough data to support a robust combat set of mechanics, nor enough data to build a normal distribution.”

And underneath that, a second problem I couldn’t effectively work or design around: luck.

The Pacific is rife with it, luck that is. A hundred Japanese bomber crewmen die attacking the Lexington at Coral Sea and score nothing; a single lucky spark on an ordnance-cluttered deck at Midway kills a fleet carrier in an afternoon. I filed that away as a frustration.

It turns out it’s a diagnosis to an extent: that two old games are the clearest proof of it. I just had to put them next to each other to notice.

The game that breaks on the rock

Fast Carriers is James Dunnigan’s 1975 attempt to simulate exactly the war I was cataloguing.  SPI ran the game and a companion history article in the same issue of Strategy & Tactics and the article describes a war the game cannot produce even with errata! Doah…

The history is built on catastrophic single hits: three Japanese carriers at Midway set ablaze by a handful of bombs, finished not by the bombs but by the fires raging through the planes fueling & rearming on their decks. Run that same moment through Fast Carriers’ own combat tables and, by the designers’ own bound-in analysis, it has about a .05% chance of happening.

They knew.

They admit it in the rules booklet and hand you an optional fire table to bolt on if you want the history back. Crazy days at SPI I guess, as volume and publishing dates trumped quality control? Maybe?

However I think it’s not a design failure. That’s my article [link], proven on a specific game. Fast Carriers didn’t fail to model air-to-ship combat because Dunnigan was careless it failed because air-to-ship combat, in 1942, isn’t a thing you can build a fair table for. Not enough data. Too much luck. The war the article promised is the war I surmised couldn’t be simulated, and the CRT in the same magazine is the receipt.

Figure 1 Art by ギャン

.

The game that sails around it

Now put Wooden Ships & Iron Men beside it. Same hobby, one year apart, and the exact opposite outcome a game so trusted that Avalon Hill’s own magazine ran reader-built scenarios for it continuously, for twenty years. Built on a 12 page rulebook. A rare achievement.

Why did it hold up when Fast Carriers needed a patch on arrival and arguably was never refined, nor fixed? Look at what those twenty years of fans of Wooden Ships &Iron Men actually did: they added scenario after scenario, and every one of them aimed at the combat.
Nobody ever tried to fix the less than opitmal movement. Because age-of-sail gunnery is the thing air-to-ship combat isn’t, highly physically bounded, highly repeatable. Wind you can measure. Range you can measure. A broadside of roundshot into a hull at a known distance does a knowable thing. There’s a fair table to be built there, and 100’s if not thousands of engagements and Wooden Ships built it, and the community spent two decades confirming it was worth building on.

This era of conflict lasted a long time and the short story it tells us is all in the data we can accumulate from these battles. Easily allowing authentic combat resolution systems and tables to be built.

The thing my article says can’t be modeled the luck, the missing distribution is the one thing Wooden Ships never has to touch. It’s fighting in the era before the variable that broke Fast Carriers existed.

.

The line

So here’s the argument the two games make together, and it’s the age old challenge of every designer.

A wargame can accurately simulate the parts of war that are governable by physics, geometry, the knowable exchange but it starts stretching the truth the moment it reaches the parts that aren’t.

Age-of-sail ship-to-ship sits on the governable side of that line. Pacific air-to-ship sits on the other, in the fog of too little data and too much luck, where the most real thing a designer can do is what SPI did: admit the table can’t carry it, and offer you a house rule instead of a fudge. If only other systems I’ve played did the same. Maybe that is a topic I will explore – look at all of the other Pacific WWII Air to ship games…hmm.

Other game systems have tried and to my mind failed to give us the elusive playable historicity. Now. No doubt there are solutions to be had, to bring this luck element under some level of chaotic good control. Hermann Luttmann does a masterful job of layering in chaos [which includes luck] into his Civil War titles. Once again, he has the benefit of a significant number of engagements, and deep documented history to enable him to present us with ‘reasonable’ levels of chaos, built on a solid foundation of fact.

That’s not a knock on Fast Carriers. If anything, it’s the opposite it’s the game being honest about the exact wall I ran into with a spreadsheet a stack references and web sites. Some wars you can put in a box. Some you can only gesture at. Knowing which is which is most of what judging a wargame actually is.

But maybe I’m wrong?

Can luck be modeled? Of course it can, that’s what the dice are for. Randomness isn’t the thing wargames can’t handle. Fast Carriers’ problem was never that it had luck. It’s that its odds were calibrated wrong which made the historical Midway a 0.05% event. That’s a bad distribution, not an unmodelable one. Hand the same designer better data, or just a fatter tail on the damage table, and the catastrophe becomes a rare-but-real outcome instead of a statistical impossibility. “Not enough data for a normal distribution” quietly assumes you need a normal distribution. You don’t. War is full of fat tails, and you can model a fat tail.

.

And here’s the one that really breaks my supposition: morale. Morale isn’t physical. You can’t measure it with a range-finder or a protractor. And wargames model it constantly cohesion checks in GBoH, morale rolls in ASL, the whole nerve-and-friction engine of Combat Commander. More than that: morale is often the best part of a good design, the thing that makes the game feel like command instead of arithmetic. So my neat little line — physical is governable, non-physical is where the lying starts doesn’t survive contact. The single most non-physical thing on the battlefield is one of the most successfully modeled things in the hobby. So WTF am I talking about you may wonder?.

Morale models well and air-to-ship model badly, if it isn’t the physical/non-physical split, what is it? Its because morale, for all that it lives in men’s heads, has regularities; units break at fairly findable thresholds, across thousands of engagements, and you can calibrate to that. Pacific carrier sinkings in 1942 don’t give you thousands of anything. They give you a handful of freak afternoons.

Which means the straight talk version of my thesis isn’t “some war can’t be simulated.” It’s narrower and less romantic: some war can’t be captured by a probability table alone, this stufft has to be modeled as a mechanism instead. And SPI knew it. That optional Fire Results Table bolted into the back of the rulebook is the mechanism to enable that ‘fire that spreads’ because of what’s on the deck, not a die roll for “sunk.” They could model the thing. They just couldn’t do it on the base combat table, and to their credit they said so. Flat Top and CV, a few years on, did more of it in the base rules. The problem was solvable. It got solved. Just not particularly elegantly I think.

So maybe the line I’m drawing isn’t between the governable and the ungovernable at all. Maybe it’s between the parts of war with enough repetition to fit a table, and the parts rare and lopsided enough that you have to build them a little machine or set of mechanics all of their own. Age of sail gets a table. The carrier deck gets a specific mechanic. Both are modeling. One is just harder, and more real about when the dice aren’t enough.

.