FS3

[faraday]

@kitteh said in FS3:

I really didn't mean to reopen any of this and I'm not sure why this whole post is @me.

Oh I didn't mean for the whole post to be at you. Just the quirks of the bbs reply thing. I just wanted to make sure it was clear that no criticism was aimed at you.

@The-Sands said in FS3:

@faraday Actually, 2d6 isn't really a bell curve. It's a pyramid shape. ... The thing with the 'many dice' systems is that they aren't real bell curves and they have some terribly awkward spots.

Technically you're not wrong about 7 being most probable, but just about every textbook depiction of 2d6 probability shows it as a bell curve so I'm not sure the distinction is meaningful.

I know this is a crappy chart but I just wanted a quick illustration. It shows the chance of hitting (an opposed roll with ties going to the attacker) at varying skill levels in FS3 3rd ed. A 1 (dark blue bottom line) has virtually no chance against a 12, and a pretty decent chance against a fellow 1. I'm not saying this is a bell curve (obviously). I'm just saying I think it's a pretty straightforward progression with no weird peaks or valleys or jumps where 1 extra skill level gets you some tremendous advantage. But maybe I'm missing your point?

[mietze]

Honestly the "zomg I'm failing too much!!!" can also depend on what games the player has played on recently.

Once I went from TR (high high ridiculous dice pools for everyone) to a low powered/starting level game that was also WoD. People freaked the fuck out constantly because "I only got 1 success instead of 17!" Or "I'm not even going to try, since I have a dice pool of 5!" (Which really isn't bad particularly when everyone else has low pools, but people were so used to seeing successes and big ones every roll they'd have anxiety if they thought they wouldn't.)

Saw a similar thing with people transitioning from high power WoD to FS3 in the battle star game I played.

I loved combat rp in the FS3 system, I found it both freeing and more fun. Yeah I'm weird, but I find it easier to describe missing in more fun and varied ways than hitting. Luckily it was not super jarring for me because I wasn't coming directly from an all monster stats all the time environment though.

[The Sands]

@faraday Looking at your chart one irregularity would appear to be people of equal skill. The person with a skill of 1 shooting at a person with a skill of 1 has around an 80% chance to hit while a person with a skill of 12 shooting at a person with a skill of 12 has only about a 60% chance of hitting.

This is probably backwards since at very low levels a person with poor skill has more difficulty hitting a stationary target. At the higher end of skills you have a greater tendency toward 'the quick and the dead' where it is a big deal with who gets off the first shot.

However, even if we handwave a little bit to explain why the curve works the way it does you get a weird 'bump' at 8 where the odds are higher of an 8 hitting an 8 than a 7 hitting a 7 or a 9 hitting a 9.

Also, the shapes of the curves is a bit off. If you track simply the curve for the person with 1 die the change in probability seems to decrease at each step (draw a line between the dot for 1 to 1 and for 1 to 2 and you will see that 1 to 3 is above it. Likewise a line between 1 to 2 and 1 to 3 will have 1 to 4 above it).On the other hand the curve for someone with 12 dice seems to increase the change between each step. If you draw similar lines you will see that the subsequent points always fall below those lines.

This isn't meant to say that your probability curves are horrible or anything. Just that there are mathematical oddities that creep in there.

[faraday]

@The-Sands I think the dots are misleading you though because a 7-vs-7 is better than 8-vs-8 and 8-vs-8 is better than 9-vs-9. Basically the higher you are the better you are at dodging, so the attacker has a lower chance of success. I can see where you might want same-vs-same to work differently, but it is at least internally consistent. I'm fine with how it is.

[WTFE]

@faraday said in FS3:

And in large part I agree with it. It’s not fun to miss all the time. It’s not fun when you build a character around a concept of being good at X, and then the one time you have to use X in a scene, you fail (even if you would’ve succeeded at that roll 95% of the time overall). It's boring when combat rounds are just everyone whiffing at each other (in fact, I've been known to immediately trigger another turn when that happens).

I just don’t know how to fix that with dice.

The problem with FS3 is that the number of times you do actually have to roll is so small that any failure is magnified tremendously. If you only roll three times in a combat, even a single failure is going to make a "can't miss, hotshot badass" look "ho-hum". If the character is, by reading the sheet, supposed to be "ho-hum" that's not a problem. But when the character sheet reads "this guy is a veritable god of combat" failing even one out of three is pretty bad.

But that's not how perversity works. Perversity will have you failing two times out of three . Or even all three times.

The problem is that you're insisting on a (weighted) stochastic resolution system without realizing that those weights will only be visible after a whole lot of samples. That's how stochastic systems work: emergent properties emerge after a whole lot of samples. For any small group of samples the emergent properties are invisible or, worse, as seems to be the case with FS3, grossly distorted in perception.

Part of this is that we tend to remember things that surprise us more than we remember things that go as expected. If I'm playing Joe Badass Hotshot and I make a shot, that's a "well, duh!" result that's not worthy of notice. If, however, I'm playing Joe Badass hotshot and I fail a shot ... that's unexpected and will be noticed and remembered. And if I go through a whole combat without ever making a shot -- something that can happen with far greater frequency when the number of rolls is small than when large -- that's going to leave the impression that the character sheet is writing cheques the system can't cover.

So there's two solutions to this:

1. Mitigate the stochastic nature of the system. This can be done by a variety of mechanisms. CORPS uses an auto-success if the (modified) difficulty is below a certain threshold against your skill. Other systems provide luck points (under a bewildering variety of names) that permit you to do controlled editing of the stochastic weirdness. (Some of those even allow you to use the luck points as a two-edged sword -- swapping an unimportant failure for success where it is important, for example.) Some provide a system of "level shifts" that monkey around with probabilities and skills in such a way as to let badasses be badasses where they're supposed to be (wading through mooks, say) while making real challenges, well, really challenging.
2. Make the system use its stochastic samples enough that emergent properties actually emerge in a single conflict. This means no more "a whole combat can be resolved in three rounds" naturally, but that's your trade-off if you're going to insist on a purely (weighted) stochastic resolution system.

[The Sands]

@faraday said in FS3:

@The-Sands I think the dots are misleading you though because a 7-vs-7 is better than 8-vs-8 and 8-vs-8 is better than 9-vs-9. Basically the higher you are the better you are at dodging, so the attacker has a lower chance of success. I can see where you might want same-vs-same to work differently, but it is at least internally consistent. I'm fine with how it is.

That's what I said by 'handwaving a little' to explain why the curve doesn't work the way I suggested it should. You can argue that you get better at dodging faster than you get better at targeting. However the real thing is that there is not a complete internal consistency. While XvX tends to decrease as X gets higher when you go from 7v7 to 8v8 that doesn't happen. At that point the odds increase, then when you go to 9v9 they decrease once more.

Again, this isn't a huge spike or anything, but the chart at least does show it occurring. Out of curiosity are these odds calculated odds or did you use Monte Carlo for determining them?

[Thenomain]

To complicate the statistical issue, a perception check: The "I'm Failing Too Much" is an ongoing issue, possibly due to confirmation bias. On pretty much every game I've coded dice for, I've had to show a brute-force test of the roller to prove to people that it was working as intended.

This conversation made me think of this, and I know people will say "that's not what we're talking about". I know this, thanks.

[WTFE]

@Thenomain I can't even begin to count the number of times I've cracked open the code to a die roller because I was convinced there was a flaw in it that was biasing the results. Or later how many times I just did a quick monte carlo of, say, a hundred thousand virtual rolls to check similar perceptions. True randomness sadly, given the flaws in the human brain's construction, never appears random. While that which is emphatically non-random can appear random.

[Ganymede]

@Thenomain said in FS3:

To complicate the statistical issue, a perception check: The "I'm Failing Too Much" is an ongoing issue, possibly due to confirmation bias.

See, @WTFE is going on about failing too often, and yet I have yet to fail a roll, even in PvP. So, I mean, something has to be wrong.

[WTFE]

@Ganymede Apparently I was playing earlier versions of the game that were more prone to failures.

[Ganymede]

@WTFE said in FS3:

Apparently I was playing earlier versions of the game that were more prone to failures.

Maybe. Might I suggest that you are protesting a bit much, then? Because, from all the evidence I can see -- anecdotal, experiential, and circumstantial -- everyone on BSG playing FS3's new edition seem to be having no problems at all.

Like, I literally kicked the crap out of someone in Pyramid over whom I had a 2-dice edge. Blanked him on opposed rolls, so the score ended up being 5-0.

[Thenomain]

@WTFE said in FS3:

True randomness sadly, given the flaws in the human brain's construction, never appears random. While that which is emphatically non-random can appear random.

[WTFE]

@Ganymede said in FS3:

Apparently I was playing earlier versions of the game that were more prone to failures.

Maybe. Might I suggest that you are protesting a bit much, then?

Nope. I am reporting on my experiences with the FS3 system and why it is that "we use FS3" is a mark against a game as a result. I'm also explaining exactly how it is that you can have "the numbers" say one thing while "perception" says something entirely different. I mean you can see faraday's frustration at how the numbers say that people are succeeding all the time while she has to keep pushing the odds up higher to the point that nobody fails anymore.

And I think that's bad too, incidentally, because while it does stop Assie McBad from coming across as Assie McRainbowButt, it also makes it so that Assie McBad ain't actually all that much better than Hum the Ho either when both are shooting at enemies. (Obviously, given the chart above, if they're shooting at each other Assie will wipe the floor.)

And I pretty much place the blame for these issues straight at the feet of insisting on a purely stochastic resolution system without giving it enough samples. One thing or another has to change if you want to both distinguish a badass from a plod and not make the badass occasionally get frustrated (which will be magnified in memory because human nature) by occasionally being a clown instead. Either more samples have to be used or you have to fudge the purity of the stochastic resolution.

[faraday]

@Ganymede said in FS3:

Like, I literally kicked the crap out of someone in Pyramid over whom I had a 2-dice edge. Blanked him on opposed rolls, so the score ended up being 5-0.

But I think that's exactly the problem that some folks are complaining about. He "should have" beat you. He certainly "shouldn't have" lost 5-0. Just as @Auspice's char "should've" cleaned my char's clock in the marksmanship contest.

And yet... extreme upsets like that happen with reasonable frequency in real sports and real combats. How many Olympic underdogs have there been? How many medal winners have done something amazing?

@WTFE said the same thing I was saying: That the perception of failures is magnified over a small sample size.

Personally I have no problem with a Expert char missing one time out of three in a single combat, as long as he's not missing one time out of three in every combat. I remember one of the top pilots on BSGU commenting about how she couldn't hit crap in the first few combats, but now that we've played 40-some combat scenes, she's one of the top aces. Which is as it should be.

Edit to clarify: That's 40 scenes total, not 40 scenes where she was in

[kitteh]

It certainly seems like the dice work like they should on the long term (even if we're not strictly getting a 'scientific sample size'), and I think most of us are perfectly OK with that. The multi aces are people with the higher skills who've been in lots of combats. The schlubs perform schlubily. For all the talk earlier about fancy tactical choices or whatever, but it mostly boils down to that.

I'm not sure FS3 gets singled out here since no game system (that I know of) is performing 1000 rolls and averaging them to determine success/failure. Maybe in WoD you can min-max a whole lot more to have ginormous pools, but presumably those chars should be fighting equivalent challenges. Maybe the one thing in FS3 is that there's less differentiation in what you can do or what you can fight, but that goes with the setting, too. Only so many ways a Viper can pew-pew.

[Lotherio]

@kitteh said in FS3:

I'm not sure FS3 gets singled out here since no game system (that I know of) is performing 1000 rolls and averaging them to determine success/failure. Maybe in WoD you can min-max a whole lot more to have ginormous pools, but presumably those chars should be fighting equivalent challenges. Maybe the one thing in FS3 is that there's less differentiation in what you can do or what you can fight, but that goes with the setting, too. Only so many ways a Viper can pew-pew.

It does oddly get singled out more often, a new game comes on, people ask what system, the new game host says FS3. Then it starts again, people will avoid it because of FS3. The first post is negative usually because FS3 is being used it seems.

The stats are always as @faraday has said, more chance of success then other similar games (WoD/Fate/etc). No one attacks new games using those systems because thise games use those systems. But FS3 gets unfair guff. I think it's because it has popularity and a few places use it both because they recognize the basics (dice pool of attribute+skill vs # .. Each die over is one success) from systems like WoD and it's easy to install and get running.

I agree dice on the average game using FS3 aren't used enough.

Also the any target under a certain number is auto success. This has always applied on FS3, the shooting range is a given, normal mundane tasks aren't rolled, they're auto success, dice are fur high intensity situations when I've seen them used.

[kitteh]

I would imagine the bad name comes more from the out-of-the-box nature of the codebase, which encourages quick, sometimes low quality / protest games using it. Which obviously says nothing about FS3, but it causes people to associate those games with it, and then maybe look for anything to poke at?

[Three-Eyed Crow]

@WTFE said in FS3:

The problem with FS3 is that the number of times you do actually have to roll is so small that any failure is magnified tremendously. If you only roll three times in a combat, even a single failure is going to make a "can't miss, hotshot badass" look "ho-hum". If the character is, by reading the sheet, supposed to be "ho-hum" that's not a problem. But when the character sheet reads "this guy is a veritable god of combat" failing even one out of three is pretty bad.

I think this is a really good observation, though I'm not sure how best to address it. FS3 dice are very forgiving until they're...randomly not, and there's sometimes just this 'Whelp, that weirdly happened' feel to Failure, rather than any kind of opportunity to make it fun and interesting. Maybe everyone should build in an RP Hook 'flaw' they could exploit to earn a Luck point in Failure situations? I don't know. They're rare enough that they should be cool/exploitable for some drama, not just 'huh, shrug, that dun work.'

This is more an issue in straight-up rolls, not so much opposed/+combat situations.

[faraday]

@Three-Eyed-Crow You can always use a luck point to re-roll when that happens though in a straight-up roll. Luck works a little differently in combat of course.

[The Sands]

@faraday Running the probabilities it looks like you are using Monte Carlo method for calculating distribution. I went ahead and did the math and you can get the exact mathematical probabilities here..

It appears that the spike I saw at 8v8 was simply a statistical anomaly and not part of the actual probability curve though I still stand by my statement that the probability deltas are still just a little off, IMO (but again, this isn't a horrible thing. It is far better than you will see in a lot of other systems).