The Tao of Gaming

Boardgames and lesser pursuits

Variance in Games

Based on Larry’s comment, some musings…

Whether I’m pro-variance or anti- depends on the game, to a certain extent. In a large scale business game, I want a tight spread. For example, I may sell 900,000 cars or 1,1M cars in country X over the year. Not “I may sell 0 or 2 million.” {I’m looking at you, Automania). Combat, to my mind, is a highly variable event. How many people have had bullets whiz by their head, bounce of a flask, strike their helmet, etc? An inch difference? Dead.

However, I take Larry’s point … at a certain level of war (say, the Eastern Front) Armys could (and did) estimate troop losses due to combat, exposure, illness, accidents, etc. To my mind, that makes Axis and Allies combat system feel roughly right. A good player can glance at a combat and have a rough feel for the losses. But the system does allow for bizarre outcomes, which:

  1. Allow weaker players to sometimes win.

  2. Make for good stories (Frunk and possibly some others may recall us figuring out the odds of him losing a Memoir ’44 tournament game were not a “Million to one” as originally stated, but somewhere on the order of “500-1000 to 1”). I did have one argument a decade ago that a specific Titan battle was a “Million to one” event. Our math (and we did math) suggested that was roughly correct. But take enough events, and you’ll see things like that.

  3. Prevent exact calculation at the table and hopefully reduce analysis paralysis (although I’m firm in blaming the slow player, not the system; still, you shouldn’t encourage them).

Now, systems should be well designed (cough Vampire: The Masquerade cough), but there’s nothing inherently wrong with it, for a chaotic theme … and fighting counts. I rather like the Pokemon miniature system, where each miniature spins, and has a pie chart on the base showing results. So each figures hit/miss can be very finely tuned (in extra few degrees of “Miss”), and quite elegant.

But I’ve got nothing against WOW’s “Six dice may be 0-6 hits.” Also Nit picking, since each character usually throws ~2 defense dice, the curve that Dennis alluded too is even more compressed. 0 Hits becomes reasonably likely (since some of the 1s and 2s will get reduced) but six (and even five) are rare. Although the special powers on criticals can drastically increase variance, which Larry will dislike. (One of them is “If you critical on this melee attack, roll a magical attack of four dice.” Yeouch. Again, my decision to be affected by that was poor.]

One other point … even if the combat sub-system was exactly as Larry described (2-4 hits instead of 0-6), the overall system varies dramatically, due to critical breakpoints. Imagine a character with 6 wounds. If you get 3 hits, then 2, they get another turn and heal, kill you etc. If you get 3 and 3, they die. This is insanely important (and one of the key elements of luck in Titan, which usually throws enough dice in a battle to make it more like the East Front analogy than Axis and Allies). If you have the 0-6, again, you expect 3 and 3, but now also have the (very rare) single shot kill or “don’t kill in three attacks”.

The “variable” system is more variable, but in the aggregate either game will likely be decided by the “Just missed killing him” results.

So I don’t think you can just look at the combat subsystem and decry its variability.

Update: Based on Larry’s comment, I created a poll on BGG.


Written by taogaming

August 10, 2008 at 8:43 pm

Posted in Ramblings

Tagged with

6 Responses

Subscribe to comments with RSS.

  1. Well, while I understand the importance of realism to many gamers, I’ll take playability every time. And in the games I’m likely to play (say Axis &Allies or a Wallace military game), I’m not sure that level of realism is all that appropriate anyway. So just because the more chaotic system is more accurate doesn’t mean I want to see it in the games I play. But that’s just my personal viewpoint.

    Actually, WOW Mini’s system is probably not too problematic because hits occur 70% of the time (which leads to a flatter curve than a 50% probability, which seems more typical). I still think the player getting 2 hits when the average is 4.2 can say he wuz robbed, but I’ve seen far worse.

    My big problem is the Risk/A&A syndrome, where you set up an important battle and terrible things happen. In Risk, for example, each battle has the results of 2-0, 1-1, and 0-2 (attacker wins-defender wins). The problem is that the defender can rip off four or five 2-0 rolls in a row and destroy the best strategy. If the possible results were 2-1, 1-1, and 1-2, it would be far better and more calculable.

    By the way, critical breakpoints don’t necessarily bother me in a less chaotic combat system. If 3 hits is essential, you should (if possible) line up enough firepower to make 3 hits very likely. But this can be much tougher to manage in a system with high variability.

    Larry Levy

    August 10, 2008 at 11:20 pm

  2. To a certain extent, it is a matter of taste, so we’re talking past each other. The 2-1, 1-1, 1-2 system you give means that having 2n+1 units guarantees a win. Nothing wrong with that … many wargames do just that (moreso, in that reaching a critical point usually guarantees a win without losses).

    So the issue, boiled down to it’s lowest level.

    “In a combat game with generic units, where a player typically gets 5 units a turn. How often should the smaller side win in an 11 vs 5 battle?”


    August 11, 2008 at 10:32 am

  3. (Frunk and possibly some others may recall us figuring out the odds of him losing a Memoir ’44 tournament game were not a “Million to one” as originally stated, but somewhere on the order of “500-1000 to 1”)

    Actually it was odds that I would win, based on my opponent getting unlucky enough in his attacks to not kill one of my two strength units with ~24 1/3 chances. It was a little more complicated than that due to retreat possibilities, but I think it ended up being around 2000 to 1.


    August 11, 2008 at 12:55 pm

  4. When thinking about this sort of thing, I’ve started to think about it less as variance and more as risk. Combat, and many other real-world conflicts, are calculated risks. I think the question of whether there is “too much variance” is not really the right question; I think it’s more interesting to ask whether the player is given tools to manage his or her risks. If a game is decided on a 50-50 die roll, that’s one thing; but if it’s decided after a string of decisions that had risk management implications, and which at various points a player made a choice to minimize risk at one point which resulted in a high-risk conflict later or whatever, and the end result of these choices was a 50-50 die roll, that’s another thing entirely.

    This is why I like the “buckets of dice” combat system in, say, EastFront, but detest it in The Napoleonic Wars. In TNW combat is very risky because of the whole hit/route mechanic, but the players simply don’t have good risk management tools. They can’t take a risk now in the hopes of reducing risks later, or take a more measured approach now to reduce immediate risks … they just have to fight a battle and have to roll the dice and maybe if they’re lucky they have a card. Same thing with Sword of Rome; the players have little control over the risks they have to take, and neither do they have any way to mitigate them. The fact that the Sword of Rome CRT is brutally random isn’t really the problem, although it doesn’t help; the problem is that the players don’t have real, interesting choices about the risks they take. On the other hand, in EastFront, players are making resource allocation decisions (how many Panzers to which army group? How many HQ steps to burn?) that allows them to decide which sector is important and where to take risks and how much risk to take. Hannibal also has much more nuanced choices about risks. Actually, I’d argue that even Risk has more interesting risk management decisions than TNW.

    Chris Farrell

    August 12, 2008 at 7:29 pm

  5. Managing risk is a good way to think of it, so just imagine that I said that before you did….


    August 13, 2008 at 8:49 pm

  6. I did have one argument a decade ago that a specific Titan battle was a “Million to one” event. Our math (and we did math) suggested that was roughly correct. But take enough events, and you’ll see things like that.

    I haven’t thought about Chris Upson in several years. Have to track him down . . . .

    Steve Nicewarner

    August 24, 2008 at 3:11 pm

Comments are closed.

%d bloggers like this: