Quote from: TimYes, I always try to take my meals in the Gobi or Sahara. :raise:

I knew I should have confirmed the right spelling.

Quote from: TimAnyway, you're the one eating the highly decorated three layer cake that's ALL the same flavor and took four hours of prep to make, not me.

There is only one flavor for cake:  Chocolate.  Sorry.  Had to be said.

(My wedding cake was chocolate, too, by the way.)

QuoteNo, I would make one roll for each unit (for the lack of a better word)

Tim, same thing, what I mean is that you roll for each discrete chance of being noticed. Obviously if you like you can aggregate several individuals from one instance into a single roll--any questions over whether multiple individuals should have a better chance than just one, and how much better, is just details.

But note, this is iterative rolling by another name, it just brings up another issue with "sneak" rolls in other games, which is whether it should be done as a straight roll, opposed rolls (better level of success wins), single roll against a difficulty/penalty based on who might be listening, or a straight roll which, only if failed, triggers a "perception" roll by the "opponent".

Also I'd like to go back to John's objection to multiple rolls; I want to emphasize that since he's apparently not concerned about illusionism or "conflict resolution", it basically amounts to an aesthetic objection to rolling lots of dice. This is why I brought up the math; except for the question of decision/evaluation points, one roll can always substitute for many, in theory, if you're willing to do the calculation. (Again by decision points, I mean things like choosing whether to proceed or go back, or follow an unexpected opportunity, e.g. when sneaking around. And by evaluation points, I mean things like determining where you are, exactly, when the guards raise the alarm, or when you slip off the cliff wall.)

Many rolls can also substitute for one--again in theory. And John is right that there's nothing inherently more realistic about the curve that comes from one method or the other. What strikes me as a bigger issue with "let it ride" and related "scene-based conflict resolution" is that it changes the relationship between different but related curves. E.g., when you have a strict 1 roll/50' climb check, it scales differently going from a 50' wall to a 300' wall, than it does if you simply roll once for any given "climb attempt scene". In BW you can give a higher obstacle for a higher wall, sure, but just as the actual math for converting multiple checks into a single diceroll is complicated, so the use of Ob modifiers is going to be very hard to make come out "just right". The result, I think, is that there will be a bias aginast a plan which depends on a chained sequence of tasks--where failure of any one means total failure--compared to a plan that can be summarized in a single action, no matter how "extended", since the latter just means one roll which will "Ride".

Quote from: Elliot WilenTim, same thing, what I mean is that you roll for each discrete chance of being noticed. Obviously if you like you can aggregate several individuals from one instance into a single roll--any questions over whether multiple individuals should have a better chance than just one, and how much better, is just details.

Oh, ok. I understand now. I might have some minor quibbles regarding how I'd set up a dungeon or other set-piece encounter in BW (much more of a 'hit the highpoints' flowcharty style) vs. how I'd do it in D&D (traditional map with many individual encounters), but they're sort of beside the point.

QuoteThe result, I think, is that there will be a bias aginast a plan which depends on a chained sequence of tasks--where failure of any one means total failure--compared to a plan that can be summarized in a single action, no matter how "extended", since the latter just means one roll which will "Ride".

I think this paragraph is really perceptive. All I can say is that when GMing BW I make a special effort to make certain that failure at any given point is just as interesting as success and doesn't bring things to a grinding halt. By "special effort" I mainly mean improvise.

We also stop and discuss stakes about 50% of the time which can also help avoid stone wall dead ends.

Quote from: John MorrowThere is only one flavor for cake: Chocolate. Sorry. Had to be said.

We are in total agreement, here!

Tim

Quote from: jhkimSure. I don't like rolling Stealth every round either, for similar reasons to not liking to roll for finding traps. As for how I'd handle it...

No.  I understand you prefer Burning Wheel, Let it Ride,  Storytelling games, etc.  But you had asked:

Quote from: jhkimSeriously, could those in favor of repetitive rolls give some more concrete examples of when they think it is significantly better than a single roll?

Which I did with the example of the map, and specifically stealth checks for character moving down the long corridor between the entrance and room 51.



Let's consider the way it would work in play:

* Characters noisily marching down the hall would be heard by any monsters in the rooms with doors on the hall: 5, 15, 16, 17, 45, 30 as they pass their doors, and be seen by any monsters in the East-West halls as they cross them.  
* Any character can move quietly, although those not wearing metal armour are quieter than those in chain or plate mail.  If moving quietly, monsters  
in rooms they pass can make a hear-noise check.
* A thief that makes a successful move-silently check doesn't make ANY noise, and the monsters in the adjacent rooms get no hear-noise check.  If they fail their check, they're still moving quietly.

A thief moving ahead of the party should roll to move silently each turn.  Depending on where they are in the hallway, the GM knows which monsters to make hear-noise checks for if they fail their move-silently roll.  Keep in mind that there are some extra doors in the hallway to side-passages, and the players don't know that there aren't secret doors as well.

Multiple rolls are also in the player's best interest.  If they fail a check one round, it doesn't mean they've been discovered.  There could be no monsters in the adjacent room (or they failed their hear-noise check).

Running this specific dungeon, it is significantly better to ask for repetitive rolls each round than a single stealth check that applies to the entire hallway.

Now, you might prefer the story, story, story approach of other games, and not enjoy a traditional RPG as much as a storytelling game -- and honestly, that's totally cool.  But, like I said earlier:

Quote from: jhkimLet it Ride is great for a storytelling game. Not always so great for a traditional RPG where the focus is on exploration and adventure.

Quote from: Elliot WilenAlso I'd like to go back to John's objection to multiple rolls; I want to emphasize that since he's apparently not concerned about illusionism or "conflict resolution", it basically amounts to an aesthetic objection to rolling lots of dice.

Many rolls can also substitute for one--again in theory. And John is right that there's nothing inherently more realistic about the curve that comes from one method or the other.

Converting from many rolls to one roll and getting the odds right is non-trivial -- and by "right" I mean "right for the game's fictional universe."

If I have 17- skill in GURPS my odds of blowing an unmodified roll are very small. My odds of blowing any one of 10 rolls are still pretty small, but figuring out exactly what they are and then rolling that would (for me) require a computer.

I probably care way more about getting the odds right than most people... but it'd be worth it to me to make 10 rolls rather than one that was a poor approximation of what I'd paid for.

Cheers,
-E.

Quote from: -E.Converting from many rolls to one roll and getting the odds right is non-trivial -- and by "right" I mean "right for the game's fictional universe."

If I have 17- skill in GURPS my odds of blowing an unmodified roll are very small. My odds of blowing any one of 10 rolls are still pretty small, but figuring out exactly what they are and then rolling that would (for me) require a computer.

I probably care way more about getting the odds right than most people... but it'd be worth it to me to make 10 rolls rather than one that was a poor approximation of what I'd paid for.

Cheers,
-E.

If you need to make all x or fail, and your chance to succeed on one roll is y.  Then your chance to succeed on all is (y)^x.  For roll under 17 on 3D6, that is about 99.5% on one roll or about 95% to make all 10 (give or take a factor of .995 .  Yet if your base chance is 80%, then to make all 10 rolls you only have about a 10% chance (give or take another factor of 0.8 given my cheap calculator).

Basically with too many cascaded rolls, unless you really are in the 95% category, you are sure to fail.

Quote from: XantherIf you need to make all x or fail, and your chance to succeed on one roll is y.  Then your chance to succeed on all is (y)^x.  For roll under 17 on 3D6, that is about 99.5% on one roll or about 95% to make all 10 (give or take a factor of .995 .  Yet if your base chance is 80%, then to make all 10 rolls you only have about a 10% chance (give or take another factor of 0.8 given my cheap calculator).

Basically with too many cascaded rolls, unless you really are in the 95% category, you are sure to fail.

Yeah -- I can actually do the math; I don't typically play w/ a calculator or the exact odds sitting in front of me... my point was that I'd rather do the 10 rolls than do the math.

And I'm okay with a low chance of success for multiple attempts: I think sneaking through a dense enemy encampment should be hugely risky, even for someone w/ an 80% chance of getting past one guard post.

I'm sure Burning Wheel and any other 'Let it Ride' system has ways to represent this but I'm guessing they're either rough approximations of what multiple rolls would indicate or they require some work to figure out the odds.

Since I'm only in favor of calling for multiple rolls when it's particularly meaningful or interesting I'd prefer the rolls to a crude approximation or the calculation required to get a single, accurate number.

Cheers,
-E.

Quote from: StuartLet's consider the way it would work in play:

* Characters noisily marching down the hall would be heard by any monsters in the rooms with doors on the hall: 5, 15, 16, 17, 45, 30 as they pass their doors, and be seen by any monsters in the East-West halls as they cross them.  
* Any character can move quietly, although those not wearing metal armour are quieter than those in chain or plate mail.  If moving quietly, monsters  
in rooms they pass can make a hear-noise check.
* A thief that makes a successful move-silently check doesn't make ANY noise, and the monsters in the adjacent rooms get no hear-noise check.  If they fail their check, they're still moving quietly.

A thief moving ahead of the party should roll to move silently each turn.  Depending on where they are in the hallway, the GM knows which monsters to make hear-noise checks for if they fail their move-silently roll.  Keep in mind that there are some extra doors in the hallway to side-passages, and the players don't know that there aren't secret doors as well.

And if that is your ideal way of running things, that's great!!  That's your preference and I perfectly accept it.  Different strokes for different folks.  

However, hearing the exact methodology you describe, I can say that this doesn't sound like a method which I would enjoy as either a GM or a player.  In the case you describe, I would ideally prefer to condense down the repetitive rolling rather than rolling move silently each turn.  

Does this mean that your way is wrong and my way is right?  No, it doesn't.  You like the approach you describe, and that's fine.  

As a parallel, I love tactical combat.  I'm still a big Champions fan, for example.  However, there is one aspect of Champions that I don't like.  Endurance point tracking is excessive bookkeeping for me.  I realize that a simple approach where you don't track END phase by phase can't capture all the details of the system as written, but I still prefer it.  

You'll notice that this thread was Why I Like "Let It Ride" -- i.e. this is explaining my preference, not a claim about what approach is objectively better.  I don't think either of these has anything to do with storytelling, though.  I don't particularly prefer story games -- I've generally tended towards what I had called Simulationism (though not the mess that GNS made of that term).  It seems bizarre to me that someone would call making a stealth roll and then attacking a monster in a dungeon crawl "story story story".  

Quote from: -E.Converting from many rolls to one roll and getting the odds right is non-trivial -- and by "right" I mean "right for the game's fictional universe."

If I have 17- skill in GURPS my odds of blowing an unmodified roll are very small. My odds of blowing any one of 10 rolls are still pretty small, but figuring out exactly what they are and then rolling that would (for me) require a computer.

In my experience, I find that the "right" odds in terms of the system mechanics (i.e. what N rolls result in) are often broken in terms of the game's fictional universe.  That is, if you actually played through all characters rolling for their actions all the time, you'll often have broken odds.  For example, rolling out craft skill rolls by the system would result in a bizarre economy -- or the catastrophic highway deaths might result from making driving rolls -- or army actions might be broken.  Systems with critical misses often show this strongly.  I recall some estimate of how many people in an army would chop off their own heads in RuneQuest or Rolemaster.  

The above presumes that the repetitive roll odds are "right" and any way of condensing the rolls should emulate them as closely as possible.  I just think that the result of a Let It Ride roll have to be believable on its own terms.  For the most part, mechanics at best get the average close to believable and variance not too outrageous.  

By parallel, the Take 10 and Take 20 rules in D20 (along with the linear probability spread itself) aren't particularly realistic and don't match other options.  However, on the scale of a game they work well enough.

Quote from: jhkimYou'll notice that this thread was Why I Like "Let It Ride" -- i.e. this is explaining my preference, not a claim about what approach is objectively better.  I don't think either of these has anything to do with storytelling, though.  I don't particularly prefer story games -- I've generally tended towards what I had called Simulationism (though not the mess that GNS made of that term).  It seems bizarre to me that someone would call making a stealth roll and then attacking a monster in a dungeon crawl "story story story".  

You asked for a concrete example of when someone would use repetitive rolls.  In the specific situation I mentioned, I think a single "let it ride" check would be much inferior.  It's not as well suited to a turn-by-turn approach.  It's better suited to a "Conflict Resolution" / "Story" approach.

"Story" doesn't presuppose the genre -- so sneaking about and fighting monsters is no less a story than challenging moral questions.

Quote from: Stuart"Story" doesn't presuppose the genre -- so sneaking about and fighting monsters is no less a story than challenging moral questions.

Hey it worked for RE Howard, Fritz Leiber, the directors of the Die Hard movies, and... me!

Quote from: jhkimIn my experience, I find that the "right" odds in terms of the system mechanics (i.e. what N rolls result in) are often broken in terms of the game's fictional universe.  That is, if you actually played through all characters rolling for their actions all the time, you'll often have broken odds.  For example, rolling out craft skill rolls by the system would result in a bizarre economy -- or the catastrophic highway deaths might result from making driving rolls -- or army actions might be broken.  Systems with critical misses often show this strongly.  I recall some estimate of how many people in an army would chop off their own heads in RuneQuest or Rolemaster.  

The above presumes that the repetitive roll odds are "right" and any way of condensing the rolls should emulate them as closely as possible.  I just think that the result of a Let It Ride roll have to be believable on its own terms.  For the most part, mechanics at best get the average close to believable and variance not too outrageous.  

By parallel, the Take 10 and Take 20 rules in D20 (along with the linear probability spread itself) aren't particularly realistic and don't match other options.  However, on the scale of a game they work well enough.

This is a good point; I suspect that I'm incorporating a good deal discretion, "common sense" and my own perspective in what I'm asking for rolls on. My preference for representing stealthing past several guards as a series of rolls versus one roll is probably aesthetic at its core (I was going to write something about character reliability, but I'll leave it alone for now).

And I'll step carefully away from any claims to realistic -- on the game's terms or any others.

I quite like the Take 10 / Take 20 rules (in case that wasn't clear).

Cheers,
-E.

Quote from: -E.Converting from many rolls to one roll and getting the odds right is non-trivial -- and by "right" I mean "right for the game's fictional universe."

If I have 17- skill in GURPS my odds of blowing an unmodified roll are very small. My odds of blowing any one of 10 rolls are still pretty small, but figuring out exactly what they are and then rolling that would (for me) require a computer.

I think I addressed, perhaps a bit obliquely, elsewhere in my post. My point was that there's nothing about many vs. one roll that necessarily produces better or different odds. John really only seems to care about having to physically roll dice more often. Personally that's not a great concern for me, as long as we aren't talking about utterly meaningless dicerolls. The example he gave in his livejournal, where the party was exploring a bunch of pits, would be meaningless: the party members tied themselves together, there was apparently no chance of the rope breaking; the only thing the dice told the group was how many times somebody had to be hauled up after slipping.

As I mentioned in my posts, one-vs.-many does matter when you have internal decision points in the process represented by the dice. It's also hard to get important detail out of a single binary outcome. Though as a side note you could use normal/special/critical results to squeeze more detail out of a single roll. (In climbing: normal failure=you can't climb; critical failure=you fall. In sneaking: normal failure=you're seen at the gate; critical failure=you're caught in the middle of the camp.)

I do agree with him that the aggregate probabilities that arise from the multiple use of simple chances in many systems are often silly because they compound small errors in the basic chances. (It's easy to overlook a 1/1000 chance of dismembering yourself in a skirmish, hard not to notice when 50 guys do it per minute in a battle.)

Another objection to multiple rolls that hasn't been mentioned is that even if you get the basic probabilities right, difficulties may not scale the way that would be implied by multiple rolls, because e.g. your ability to climb the first 25' of sheer slippery rock isn't independent from your ability to climb the next 25'. I think a good argument could be made that many situations are easier to model "plausibly" by looking at an overall difficulty with a single roll instead of making multiple rolls.

That said, once we're in the real world instead of a hypothetical game where your GM is a mathematical savant who can calculate aggregate probabilities faster than you can roll dice, I think that a rigid "single-roll" approach would be likely to distort decisionmaking in a way that'd feel artificial. And this is really my objection: the rule is presented as something that needs to be rigidly applied as a protection against GM abuse; however, a player who expects that the rule is going to be applied rigidly can easily find ways to exploit it--again as I wrote above, the trick is to approach any given obstacle so that it only involves a single skill, therefore only a single die roll. The result I think would be to "flatten" and "discretize" the narrative in, well, a "story-like" fashion. The scale of a task matters less than the dimension. Of course this is all solvable to some extent if the player expects that the GM will exercise some discretion but this is exactly the same solution to the use of iterative die rolls.

As a side-note in this, besides the "Take 10" and "Take 20" rules in D&D3, one of the innovations in the rules was to include a "Re-roll" section in many of the skill descriptions that tended to strictly defined when you were allowed to reroll it.  i.e. Rerolls are defined in the rules, rather than by arbitrary GM decision.   Despite this, I don't recall the same sort of complaints that this was disempowering the GM.  

Quote from: Elliot WilenI think that a rigid "single-roll" approach would be likely to distort decisionmaking in a way that'd feel artificial. And this is really my objection: the rule is presented as something that needs to be rigidly applied as a protection against GM abuse; however, a player who expects that the rule is going to be applied rigidly can easily find ways to exploit it--again as I wrote above, the trick is to approach any given obstacle so that it only involves a single skill, therefore only a single die roll. The result I think would be to "flatten" and "discretize" the narrative in, well, a "story-like" fashion. The scale of a task matters less than the dimension. Of course this is all solvable to some extent if the player expects that the GM will exercise some discretion but this is exactly the same solution to the use of iterative die rolls.

Is there the distinction between your claim of artificial sounding tactics and the earlier believability claims?  Personally, I often find the results of tactics depending on many rerolls to feel artificial.  (e.g. "Let's stay here a roll X ability twenty times")

In general, any rule can be "exploited" if it is applied rigidly.  This applies just as much to rules like "Roll Climb every round" as to Let It Ride.  It's always possible to try to dodge such rules exploitation by dispensing with "rules" and instead having only guidelines that depend on GM judgement.  The eventual result of this is Amber Diceless play and similar systems.  While some people may prefer it, purely depending on GM judgement isn't necessarily objectively better -- since GMs can be exploited by playing on their weaknesses and preferences.  

An old observation of mine was that power gamers tend to specialize as either the "Rules Lawyer" or "Wheedler".  A Rules Lawyer will manipulate whatever hard rules you give out to maximize what comes out.  A Wheedler will socially manipulate to get permission for "GM's option" things which bend or break the rules -- often taking things which the GM wants in the campaign like plot hooks and so forth.  

If anything, I think that the GM discretion route is more likely to lead to story bias, since GM's are more often swayed by what would make a cool story -- at least in my experience.

John, the same thing applies either way--if you "rigidly" allow multiple rolls, you get dumb results and uninteresting decisions, just as if you "rigidly" apply Let it Ride.

There's plenty of evidence that Let it Ride and other elements of BW...not to mention other Forge-inspired maxims...are a corrective against the GMing practics that e.g. Luke & (to take another example) Judd Karlman* took for granted. It's like if your steering is out of alignment, maybe you need to remind yourself to pull the wheel to the right. The problem is that these "rules" only make sense relative to the baseline of "bad" practices: if you apply them to existing functional practices, you'll drive your car into a ditch, so to speak.

So anyway, my approach to any rules is to keep them relevant. A GM who allows multiple rolls with no consequence--but doesn't let you just declare automatic success--is doing a bad job of interpreting the rules. The same applies e.g. to a GM who doesn't enforce a rule to keep it meaningful. A classic example is a game where you can spend points on social skills or charisma, but the GM doesn't ever call for rolls (or at least take the ability into account) in social situations.

Where I'm going is "let's roll X ability 20 times" is only artificial as a tactic if the GM isn't doing the job the GM is supposed to do. It might feel weird to you since you don't like to roll dice a lot but now you seem to be saying something different in reaction to my criticism of Let it Ride.

So here we come to BW: the rule in BW isn't presented as "use common sense and make the rules relevant". It's presented as "stop GMing the way you're used to, and Let it Ride". In the context of Forge ideology, and let's be clear, that's where this aspect of the game comes from, you're supposed to follow the rules so that every group is playing the same. However, if the faults of LiR are supposed to be tempered by discretion, then the rules aren't really written that well: they say one thing to GMs who are used to guiding a story, and quite another to GMs who are used to applying common sense and enforcing the intent of the mechanics.

On the other hand if no discretion is to be applied--if an "empowered" player is going to feel cheated because the GM adjusts Obstacles a bit and/or breaks scenes down into a few rolls--then the rule really is taking us into a different player-stance relative to the game-world, something that can be good or bad depending on what you want. I can really go either way, but if I'm playing in a game that rigidly enforces a player's right to frame and then mechanically resolve conflicts, I think it'll take me farther from immersive roleplay and closer to storytelling crossed with boardgame mechanics. E.g., if you're trying to infiltrate the fortress, then it doesn't really matter mechanically whether you Sneak or Bribe a guard, as long as your skill is the same. The further you go in this direction, the more the actual skill becomes nothing more than a bit of descriptive color. And this is where the dissonance sets in, some players will want to see greater restraints based in internal world-logic, others will either see things differently or simply play for maximum advantage, logic be damned. The need to self-police combined with the temptation to "push it" (especially if other players are getting away with it) seems to me to require a more third-person, collaborative approach than the sense of first-person character-in-world immersion that I get with a GM who enforces the rules fairly and consistently.

*Somewhere I think I remember reading how Judd used to carry around a box of dice he'd shake up so he could roll "secretly", except that he'd just shake it and then make up the results. Apologies if it was someone else; the point is that the frame of reference from which you're applying a corrective matters a lot.