Friday, March 22, 2013

Deceptive yak shaving

You're not stealing it if you say "Martin asked me to post this to see what people thought."
If you say "Look at this awesome thing I made (unless it sucks, then Martin made it)" and I link to it from the forums, it's also not stealing, and pretty funny.
If you actually  claim ownership, then I sic Katz on you.  :)   - Martin, a player in my game

 As an outgrowth from the tactical discussions that happened in previous posts (and will continue to happen, though I haven't gotten back around to them yet), I asked Martin to come up with a short sheet of fill-in-the-blank options for players to have in front of them. He did, and I think it looks fairly nice, though next session I need to pause to make sure people actually fill it out with their own damage and skill levels. (Otherwise it's useless of course.)

While completing that project, he got side-tracked with what he calls "Deceptive yak shaving." Specifically:

So while digging around learnin' about combat strategy for the cheat sheets I idly wondered what the best deceptive attack strategies were for various attack skill/defense roll combos.

A few hours later and a lot of Yaks ended up shaved.  Deceptively shaved.  Boy are they gonna be surprised when they find out.  - Martin again
 And after a quick side trip to a thread on the forums where he tried to pin down exactly how much more "damage" a critical hit was worth than a regular hit, after accounting for the fact that you will hit, and including relative effects like crippling and dropping of weapons, he came up with a really cool chart that lays out your optimal Defensive Attack strategy based on relative offenses and defenses:

For best results, use a yak-hair brush and lather with Mama Bear's handmade soap pucks.

I like this chart for many reasons. First, to defend its relevance, especially at higher defense levels: your base-line opponent in Dungeon Fantasy has Skill 14. With a Medium Shield, that means a defense of 14/2 = 7 + 3 for defense + 2 for shield + 1 for retreat = 13. If you make it a Large Shield and give the opponent Combat Reflexes, that's a 15, which is already pushing the right edge of the chart. (Also: yowza. That shows you right there that "eating up a defense so the Knight can hit" is actually quite useful, not a waste of time and talent at all.)

Second, it broadly validates the first-pass recommendations by +Douglas Cole that lowering your skill to 14 or 16 is the way to go if you want to maximize your chances. There are certainly exceptions, like when you have Skill 22 and your opponent has Defense 15, but they're exceptions around the edges, and you're not too far off from optimal if you stick to the recommended course.

Third, it is supporting evidence that Defensive Attack is doing its job. With the exception of that 22 vs. 18 slot, when your skill goes up and your opponent's skill goes up, you're better off taking a higher defensive attack. (That exception is sensible if you think about it: 22-6 = 16, maximizing your chances for a crit.)

Of course this chart isn't the end-all and be-all of GURPS combat recommendations. It assumes a man-to-man fight on level ground with absolutely no interfering effects. It's a fun little thing to noodle around, though.


  1. That chart is great work.

    It's also fascinating to me that the intuition (critical hits are worth more than a -1 to defense) is not true in most cases. Against a decent defense, you're better off reducing skill to 14 or 12 and forcing the opponent to defend than you are with critical hits.

    That pleases me twice, because it means that I've been maximizing my chances for my monsters by making them attack at a 14 and that crit-fishing isn't usually that good a strategy.

    1. It's good to point out that right down the middle of the chart your best strategy is never crit-fishing. DA is doing its job!

    2. Yes - I had worked on a dynamic version of this spreadsheet myself, and found that there were a remarkable number of times when being Deceptive to 12- was entirely effective, sometimes doubling your odds of getting through defenses.

      The converse is also true - there are definitely times, especially at the low net-skill levels, where Telegraphic attacks are absolutely the way to go.

  2. Interesting.

    One thing I never reveal to my players is the skill or defenses of their opponents, so it's less useful for them - they need to judge with one axis covered up. I can't see it helping them.

    Sadly for them, it does make it easier for me to decide what the baseline tactic of their opponents should be, if I want to make them potentially get past the very high defenses of the PCs.

    1. Oh, one question - does this take into account the "value" of a critical miss? One reason I always aim for a 16 to hit is not to maximize criticals but to minimize critical misses. I tend to play risk-averse, and I'm curious if that is factored in or not.

    2. That it does not. However, if they're on the same order as the value of a critical hit, I don't see it shifting the graph much, considering the difference is just whether or not 17 counts.

    3. Yeah, probably not by much. But "drop your weapon" is very common on the table, and that can be fatal if it's your main defense (two-handed weapon fighters especially need to worry about that one).

      I might have to make a quickie dumbed-down version - one axis is skill, the other is three broad columns of "low defenses" "medium defenses" "high defenses" - and pick the median of each. Just as a broad guide for people.

      (as if it matters - the main fighters in my group have Trademark Moves and skill in the mid-20s. Everything is highly deceptive and targeted and a net 16.

    4. Do you give your players any info at all about relative levels? "This guy is fast, maybe even faster than you" or "you've never seen anyone move like that" or "he may be strong, but he's not light on his feet"?

    5. I don't, or at least not in a systematic fashion. If the monster's main schtick is that he's strong and slow, it'll either be obvious - he's an ogre - or I'll tell just from how he acts - hey, that golem is ponderous - or a Hidden Lore roll might give it if they ask.

    6. Information about your foe's relative skill level would be a nice result of an IQ or Per-based skill roll, vs the lowest, maybe, of Acting and your foe's weapon skill. Or something like that - basically you can size up your foe and tell if he's a lot better or a lot worse than you.

      If you've seen The 13th Warrior, this kinds of acting and deception can be pretty fun.

    7. I'll sometimes give that information, and I'd happily give out a Per-based skill roll on an Evaluate. But no one does that. They tend to gauge skill by approach and by results - lots of successful defenses = probably skilled. Lots of targeted attacks and/or Deceptive Attacks = probably skilled.

      But generally, I give out only some vague information. But the dual-weilding deceptive-attacking guys they fought last session, everyone knows they were good. The hobgoblins they butched a while back? Not very good. That's about all.

  3. I've been working on the same thing, and although my chart is formatted completely differently, it looks like we're getting the same results.

    So far, considering only DA and ignoring all other options, I've found that I can reduce the results to four rules of thumb:

    1. Don't use DA unless you can get your opponent's defense below 16.

    2. Dont use DA unless your opponent's defense is at least 5.

    3. Otherwise, try to get the sum of your skill and your opponent's defense into the range of 22 to 24.

    4. But if rule 3 would drop your skill to 10, then instead don't drop your skill below 16.

    Rule 3 gives you sub-optimal results in a few cases, but not by much. It's better than annoying everyone at the table by fiddling around with charts when you're supposed to be rolling dice.

    I found that the value you place on a critical hit changes the results by at most 1 level of DA, and only in a minority of cases. I haven't included the negative value of critical misses on the attack or critical success on defense by your opponent yet, but I don't expect those to change the picture much.

    1. "try to get the sum of your skill and your opponent's defense into the range of 22 to 24."

      That, right there, was the sort of "rule of thumb" I was looking for when I started the exercise. I wanted it to be a little less quantitative than qualitative, though, but I've not figured out how to make that happen.

  4. Three quick comments:

    1. I agree with Peter that the actual use case needs to bin defenses into broad categories, although I'd go with 5 bins: "None or Neg"/5-8/9-12/13-16/17+

    2. Rapid strike, if allowed, wrecks this table. I am not yet done evaluating it, but the -3 per extra strike WMs get makes RS very attractive. Flurry of blows likewise. Both of them together (-1.5, round down) are obscene and strongly encourage blender emulation. Even at the default -6 low defenses should not be lowered, they should be saturated.

    3. This sort of evaluation is really deceptive for anything that's fairly tough compared to your damage roll that has sensitive hit locations - which will be quite a few things, especially in DF. If a beastie has eyes, a lot of those DA4 and DA5 columns and/or RS options are not as effective as "stab it in the eye." Many of the lower DAs are not as good as vitals shots. Etc. This is moderately quantifiable, and I may get around to checking some of them.