Why is math so underrated?

General questions, debates, and rants about RPGs

Moderator: Moderators

Username17
Serious Badass
Posts: 29894
Joined: Fri Mar 07, 2008 7:54 pm

Post by Username17 »

maglag wrote:You didn't even need to burn spells, just buy some super cheap boots of speed you noob. And yes casters could dominate an encounter with a well aimed spell thanks to all the elegant mathematic, but haste meant casters were never in real danger of failing anything because anything that managed to survive that first spell probably wouldn't be able to survive the next. Whereas the fighter got to stab again at -5 and -10, yay for elegant math!
You still haven't mentioned anything at all that is remotely relevant in the games most people played. Even activating boots of speed wasn't really something you wanted to do very often if what you were going to do was cast spells. Spending any resources to be able to dump spells faster is only remotely relevant if you actually have enough spell slots that you have extra slots for battle spells that you otherwise won't be able to cast because battles are going to end before you can cast them all. That's just not the kind of first world problem that Wizards actually had before about level 10.

Yes, the game broke down in various ways in the double digit levels. You haven't done a very good job of describing the ways in which that happened, but we'll all grant that it did. But your characterization of that as "half the game" is fucking absurd. People stop playing any particular campaign after some finite amount of time. Some people stop after level 2. Some people stop after level 5. The people actually playing at level 13 who give a single shit what you can do with a 7th level spell slot is a minority of a minority of a minority of a minority of a minority.
maglag wrote:Also notice how the ogre has only a +1 Will save, so still easy caster fodder.
I still don't know what you think this statement is supposed to demonstrate or prove. Ogres were a good example of a well balanced challenge at the level they were presented in 3e. Dangerous, but handlable at the level described (2nd) by a standard party. There will of course be some monsters that are more dangerous and in melee and some monsters that are less dangerous in melee. And obviously, Ogres fall into the 1st category.

At 2nd level, the Fighter gets plenty chances to shine by having a longsword and cleave. That makes him MVP against a lot of enemies. It's totally OK that there are also some encounters at 2nd level where the Wizard casts sleep and then the PCs just win. Like, why wouldn't it work like that?

It's just very weird that you keep pointing to things that legitimately were not problems, that actually were very good examples of the game working like a very finely tuned machine, and acting like they somehow prove the opposite point.

There are things about 3e that don't work. But they weren't important for hardly any groups until they'd been playing for quite a while. None of 3e's problems are actually a problem at 1st level and few of them matter much at 4th.

-Username17
User avatar
saithorthepyro
Master
Posts: 265
Joined: Sun Jan 08, 2017 10:39 pm

Post by saithorthepyro »

3rd edition does have better math than most. It's certainly very late on the list of games I would say has a bad and unbalanced math component. Does anyone have any other games on 3rd editions level?
User avatar
Judging__Eagle
Prince
Posts: 4671
Joined: Fri Mar 07, 2008 7:54 pm
Location: Lake Ontario is in my backyard; Canada

Post by Judging__Eagle »

saithorthepyro wrote:3rd edition does have better math than most. It's certainly very late on the list of games I would say has a bad and unbalanced math component. Does anyone have any other games on 3rd editions level?
I'd be hesitant to put forth any that I'd consider as comprehensive and effective, as wargamey as the d20 mechanic is.

When I went and read through about 100 "major" tabletop RPGs a few years ago, only a few had ideas that I thought might be worth adapting. None of them convinced me they were better at emulating the sort of cinematic dramatic narrative that is able to gain traction with players that was smoother to operate than After Sundown. Even [Tome] D&D games are saddled with the fact that they're piggybacking off of the d20 engine, which is at it's core a wargame, not an RPG.
The Gaming Den; where Mathematics are rigorously applied to Mythology.

While everyone's Philosophy is not in accord, that doesn't mean we're not on board.
User avatar
saithorthepyro
Master
Posts: 265
Joined: Sun Jan 08, 2017 10:39 pm

Post by saithorthepyro »

Just out of curiosity, what about the d20 mechanics and it's wargaming routes make it inferior to the dice pool system used by AS/SR?
User avatar
deaddmwalking
Prince
Posts: 3459
Joined: Mon May 21, 2012 11:33 am

Post by deaddmwalking »

Here's the thing - I probably prefer d20 but what I perceive as a problem is that it produces a flat probability curve.

In real life, you're almost certainly consistently able to do routine tasks. This is mitigated somewhat by the ability to take 10. If you liken any task to throwing darts you'd expect to generally cluster your results toward the center. While there might be outliers (and you might miss the board) you're not going to have 'equal probability'. Experience indicates that most performance is going to create some type of 'bell curve'. You're much more likely to get a middling result than to get your worst ever or best ever performance.

The following are real results in meters for Emiliana Lasa at the 2016 Olympic Games (he came in 6th) for Long Jump:

7.93 7.84 8.04 8.10 7.92 7.95

The difference between the best result (8.10) and the worst result (7.84) is 0.26 meters. As a percentage of his best result, we're talking about 3% variance (ie, his worst result is 97% of his best result). That type of consistency isn't represented well by equal probability distribution. With a d20 you're equally likely to get a 1 (lowest result) as the median result and the highest result (20).

If you use 2d10 instead, the median result (11) is 10x more likely than either the minimum result (2) or the maximum result (20). On 3d8 the minimum result (3) and the maximum result is only 0.20%, while a 13 or 14 are both 9.38% (47x more likely) and your odds of getting between a 10-17 are better than 67%. As you add more dice, you ensure that 'outliers' are reduced (though still possible) and results will cluster further toward the middle.

If you want a highly variable result, you want a small number of large dice. If you want a relatively consistent result, you want a large number of small dice. It's a matter of taste which one is more 'realistic', but I think the failures that exist with the d20 are even more apparent with d100 roll under - you end up failing a lot at things that you simply shouldn't.
-This space intentionally left blank
User avatar
Judging__Eagle
Prince
Posts: 4671
Joined: Fri Mar 07, 2008 7:54 pm
Location: Lake Ontario is in my backyard; Canada

Post by Judging__Eagle »

saithorthepyro wrote:Just out of curiosity, what about the d20 mechanics and it's wargaming routes make it inferior to the dice pool system used by AS/SR?
It's far too binary in how it handles success/failure, heavily incentivizing "success", but you never feel like you care if you succeed by a lot or just hit the DC. While there are some examples where "fail by 5 = worse than not succeeding"; there are just as few examples where "succeed by 5, 10, etc.+ = great success". Low modifier checks are more dependant on the RNG than the modifier. While high modifier checks don't even give a shit about the RNG. Finally, "what" a Difficulty Class X check ever means is not consistent; sure a "roll of 20" is translated into "you win" for many home games, it's unsatisfying to know the degree by which you win by.

Frank described it as not having sufficient granularity, as the RNG is by 5% steps, and lots of things can have >5% chance of happening, but still have a chance of happening.

Whereas with dice pools, there's always a chance of total failure; or great success. Additionally, AS uses an action resolution table that's universal for the system, 2 hits at any task means the same thing all the time.

I'm not sure if it's a reasonable or logical position, but it's how I feel about it.

Also it's my impression that players in AS games tend to remember the degrees of their successes/failures a lot better than d20 players do.

As an semi-related side note, while AS has TN5 on d6's, and 6's and 5's are both "one success/hit"; Frank has noted that 6's should equal two successes, not one. Which makes rolling for effectiveness, over simpy buying hits at 4 dice/hit (or 3dice/hit with armour), slightly more appealing.

Dmdeadwalking's post also clarifies part of it. More, smaller dice, lead to more consistent narratives.
Last edited by Judging__Eagle on Fri Dec 22, 2017 9:45 am, edited 1 time in total.
The Gaming Den; where Mathematics are rigorously applied to Mythology.

While everyone's Philosophy is not in accord, that doesn't mean we're not on board.
zugschef
Knight-Baron
Posts: 821
Joined: Sat Feb 02, 2013 1:53 pm

Post by zugschef »

I think the take 10/20 mechanic is infinitely better design than stupid bell curves which only lessen a symptomatic pain in the ass by introducing needlessly complicated math.

Btw, if you really want the common situation of players failing 3% of the time, simply use a d100 and let them fail on exactly 3 equally possible outcomes. Most people get their head around that; most people don't understand bell curves and variances, however.

It's really nice that you can calculate variances and probabilities in your head, but as a matter of fact, most people don't or don't want to (even if they could).
User avatar
deaddmwalking
Prince
Posts: 3459
Joined: Mon May 21, 2012 11:33 am

Post by deaddmwalking »

Most people don't calculate odds like a computer - I've never thought to myself 'I have a 97% chance of success. There's not really a problem with 'likely success' and 'unlikely success'. Dice pools are hard to grok on the fly about the specific benefit of an additional die, bit they're easy in that more dice means more chance of success (as long as failures don't negate a success). If a 5 or 6 is a 'hit' and you need 5 hits, you know rolling 10 dice is probably not enough but 15 is). If you're rolling 2d10 and adding them together instead of a d20 the fact that you're slightly less likely to get a 15 or better will quickly be understood and ultimately it's not much differebt than the current probability (likely fail). The big difference is that you can more comfortably target things toward a 'median' result on the design side. We often 'assume average rolls' but we know that in a single simulation it is virtually certain that you won't get average rolls on a d20.
-This space intentionally left blank
Mord
Knight-Baron
Posts: 565
Joined: Thu Apr 24, 2014 12:25 am

Post by Mord »

Using a curved RNG does not introduce any additional math on the player side. The statistical properties of the RNG are something the designer and MC have to account for, but from a player's perspective there's really no difference - roll dice and add bonuses vs target number.

Assuming the designer has provided the MC with suitable guidance on how to set reasonable DCs for tasks, there should be no difference in execution for anyone at the table.

As deaddm pointed out, the reason a designer would choose to use a curved RNG has everything to do with the increased central tendency of a normal over a uniform distribution. A consequence of this is that you can still allow for "critical" results on specific natural rolls without having them come up on 10% of all outcomes (this property of the d20 being why D&D as written is slapstick).

Increased central tendency isn't universally desirable, but there are reasons you might prefer to have it in any specific case. F'rinstance, the D&D use of 3d6 to roll starting stats will produce the expected statline more often than you could get rolling a 1d18.
Last edited by Mord on Sun Dec 24, 2017 6:04 pm, edited 1 time in total.
User avatar
erik
King
Posts: 5861
Joined: Fri Mar 07, 2008 7:54 pm

Post by erik »

Judging__Eagle wrote:As an semi-related side note, while AS has TN5 on d6's, and 6's and 5's are both "one success/hit"; Frank has noted that 6's should equal two successes, not one. Which makes rolling for effectiveness, over simpy buying hits at 4 dice/hit (or 3dice/hit with armour), slightly more appealing.
I hadn't seen that proposal. It sounded attractive at first as it introduces a bit more unpredictability and I pondered reworking the math on Nexus to accommodate it (was gonna call 6's "slams" instead of regular "hits"), but it's gonna really jack up difficulty thresholds, and I liked where I had em. I figure it's about the same for AS. Not sure if that juice is worth the squeeze.
zugschef
Knight-Baron
Posts: 821
Joined: Sat Feb 02, 2013 1:53 pm

Post by zugschef »

Mord wrote:Using a curved RNG does not introduce any additional math on the player side. The statistical properties of the RNG are something the designer and MC have to account for, but from a player's perspective there's really no difference - roll dice and add bonuses vs target number.
But the MC is a player and all players want to know how likely it is to succeed on a task or how beneficial it is to get a +1 bonus or double your threat range or whatever.

Btw, bell curves don't prevent stupid outcomes such as players falling from a tree and dying. It just happens less often. Just get rid of the source of stupid which is the die roll.
User avatar
nockermensch
Duke
Posts: 1898
Joined: Fri Jan 06, 2012 1:11 pm
Location: Rio: the Janeiro

Post by nockermensch »

zugschef wrote:Btw, bell curves don't prevent stupid outcomes such as players falling from a tree and dying. It just happens less often. Just get rid of the source of stupid which is the die roll.
This is only stupid if you think a RPG is supposed to be simulating something like "a heroic quest". Ulysses doesn't fall from a tree and die, but this is totally something that happens to a lot of real people in the real world.

I'm saying this because one of the reasons Math in RPGs is more than often is a mess is because nobody is quite sure of what we're even trying to accomplish using probabilities.

Is the game supposed to be:
a) a simulation of the lives of people who happen to do heroic shit?
b) a simulation of heroic quests by a group of heroes?
c) the shared telling of heroic stories about a group of people?
d) a wargame linked by story vignettes
... (there are certainly more paradigms)

This may look like it's descending to GNS shit, but I'm describing a problem that happens: 4E is firmly on (d) camp, while AD&D started by doing (a), while promissing (b) on its covers/press releases. If the game has things like "hero points" or tagging some enemies as "boss", people will think you're doing something like (b) or (c). But if the game has real chances of people falling from trees and dying, that's a big concession to (a) and the results of failing a climbing test and dying will look absolutely jarring to people who were sold on "playing hero".

GNS is shit, but the RPG genre absolutely needs an useful nomenclature to describe the paradigms of play, because it's them that inform which kinds of tests people should be rolling.
@ @ Nockermensch
Koumei wrote:After all, in Firefox you keep tabs in your browser, but in SovietPutin's Russia, browser keeps tabs on you.
Mord wrote:Chromatic Wolves are massively under-CRed. Its "Dood to stone" spell-like is a TPK waiting to happen if you run into it before anyone in the party has Dance of Sack or Shield of Farts.
User avatar
deaddmwalking
Prince
Posts: 3459
Joined: Mon May 21, 2012 11:33 am

Post by deaddmwalking »

zugschef wrote: But the MC is a player and all players want to know how likely it is to succeed on a task or how beneficial it is to get a +1 bonus or double your threat range or whatever.
I agree only so far as it can be expressed in plain language. Most people don't care if it makes something 3% or 5% more likely - in fact, they don't even care about the numbers - they care about the story. Knowing that the boss missed them because they had a ring of deflection is going to matter much more to them then 'enemies are 5% less likely to hit'. This editorial published yesterday in the New York Times addresses this pretty well.
zugschef wrote: Btw, bell curves don't prevent stupid outcomes such as players falling from a tree and dying. It just happens less often. Just get rid of the source of stupid which is the die roll.
It's possible that you may want to include 'failure' by design. I quoted some jump distances for the long jump up above, but some jumpers had 'scratches'. Maybe they stepped past the jump point or maybe they literally tripped and fell flat on their face, but the score wasn't recorded. You may want a chance of failure that is less than 5%. Using a bell-curve RNG helps make that possible.

But there's nothing that prevents you from using a curved RNG and still allow 'take 10' or similar mechanics. In our homebrew, we have a resource that you can use for a variety of things including taking 10 on a roll that you otherwise couldn't (after you've already rolled). You can use it to turn a failure into a success for a task that you should usually succeed on, but not one that you'd usually fail. Even with that (or maybe partly because of that) we don't want to eliminate dice rolls completely. There is still tension and because it uses a limited resource we avoid a 'fail fest' but we still are invested in rolling well.
Username17
Serious Badass
Posts: 29894
Joined: Fri Mar 07, 2008 7:54 pm

Post by Username17 »

Die rolls are not inherently problematic. Indeed, die rolls are necessary if there is any disagreement possible as to the outcome of an action.

What is problematic is die rolls whose output is success and failure.The binary output leads to bad answers to most questions. When I drive to work, I do not fail to get to work. I maybe get there early or late. When I cook dinner I do not fail to produce food, it just sometimes tastes better than other times.

A random number generator should produce results that are tiered, so that it can be used to determine how an action ends up. But it should only output "failure" if the penumbra of potentially goodness includes amounts that are insufficient for your needs.

Usually the die roll should output an answer to the question "How well did you do?" with "success" and "failure" being in the discussion only if you are competing with someone else.

-Username17
User avatar
erik
King
Posts: 5861
Joined: Fri Mar 07, 2008 7:54 pm

Post by erik »

FrankTrollman wrote:Die rolls are not inherently problematic. Indeed, die rolls are necessary if there is any disagreement possible as to the outcome of an action.

What is problematic is die rolls whose output is success and failure.The binary output leads to bad answers to most questions. When I drive to work, I do not fail to get to work. I maybe get there early or late. When I cook dinner I do not fail to produce food, it just sometimes tastes better than other times.

A random number generator should produce results that are tiered, so that it can be used to determine how an action ends up. But it should only output "failure" if the penumbra of potentially goodness includes amounts that are insufficient for your needs.

Usually the die roll should output an answer to the question "How well did you do?" with "success" and "failure" being in the discussion only if you are competing with someone else.
*thumbs up*
A paraphrasing of this would be an excellent sidebar for the introduction to rpg mechanics section of most any rpg. I can think of binary non-competitive challenges where pass/fail applies, but the general argument is important. There’s a need for resolution mechanics to have tiered successes. I think that need is part of what fuels rosy support of *world products, such as it is.
User avatar
tussock
Prince
Posts: 2937
Joined: Sat Nov 07, 2009 4:28 am
Location: Online
Contact:

Post by tussock »

Failure works well in combat, in that it's not to say you didn't hit the monster, you just didn't hit it yet, you'll get another go next round, maybe after it hits you. Even a monster making it's save doesn't end things for the party Wizard, they'll just have to try something else.

Which is generalised into how failure works if something interesting is going to happen when you don't succeed now and have to try again later. You don't open the door in AD&D, there's a wandering monster check for making noise, then you can try again, only maybe after a fight (back when fights were bad for you, in that the real XP was for treasure). Having a time pressure mechanic which players naturally wanted to avoid, built into AD&D, was very helpful to those sort of time-use failure modes on searching places and breaking things.

But a lot of those gamist things got removed in 3e, on account of it being the "Yes" edition, arbitrary meta-rules like not rolling again, that made your pick lock rolls matter at all, were removed to say yes to players, but the failure rolls kept in the game out of a sense of tradition, and a bunch of "well, OK, don't even roll this skill please" rules added instead. Pick Locks in 3e is a fucking terrible skill design, literally you ignore all the rules in the skill for the other rule, which is you must be this tall to pick this lock, and if you're not just have the Barbarian break it open at no cost instead.


The real requirement on dice is that we are playing a game and there needs to be some fucking point to rolling them at all. If "failure", even temporary, doesn't suit, don't fucking roll dice, no one fucking cares how well you sung the song if failing to do it well doesn't fucking do anything.

From there, less adding and counting and shit is just better, real math is something the designer does to save anyone else having to do any, ever.

--

In an olympic jumping competition, there's totally failures, those X are mostly where the jumper walks out the front of the pit to force a red flag and stop them measuring a crap jump. The reason all the recorded jumps are so similar is the plentiful rubbish ones are never recorded.

The final, the guy in 2nd after round 5 beat the guy in 1st by 1cm on his final jump. The now 2nd place guy, last jump, had already set a PB in round 5 to take the lead, he gave it everything and stepped on the tape for an X. That's a failed Olympic long-jump, and as a dice game he'd be stacking the penalties up off other rolls to add more distance to a success result (no room for a short step, ignore the breeze, don't look at the board, extra effort, psych up, roar of the crowd, ...).

--

In D&D, we mostly want to know if you burnt some extra actions in combat (to stand up again, or clamber up the last bit with a climb check or whatever), or if you could reach something at all by jumping for it. In a game with grappling hooks and 10' poles and a fucking Levitate spell at low levels, of course you can reach it, so only the combat use even matters. Can you jump onto the dragon without wasting your whole round and maybe getting swatted off?

And no, an actual skill chapter in an RPG should not start with "you should usually not use these rules". That is shit game design. Do not do that, even though 3e totally did.
PC, SJW, anti-fascist, not being a dick, or working on it, he/him.
Mord
Knight-Baron
Posts: 565
Joined: Thu Apr 24, 2014 12:25 am

Post by Mord »

FrankTrollman wrote:What is problematic is die rolls whose output is success and failure.The binary output leads to bad answers to most questions. When I drive to work, I do not fail to get to work. I maybe get there early or late. When I cook dinner I do not fail to produce food, it just sometimes tastes better than other times.

A random number generator should produce results that are tiered, so that it can be used to determine how an action ends up. But it should only output "failure" if the penumbra of potentially goodness includes amounts that are insufficient for your needs.

Usually the die roll should output an answer to the question "How well did you do?" with "success" and "failure" being in the discussion only if you are competing with someone else.
It's difficult to categorize degrees of success to any given task in a way that doesn't result in bears, because the classic RPG thinking about "checks" is couched in terms of tasks with binary outcomes. Did you pick the lock, leap the chasm, or lift the portcullis with this attempt? "I have an X% chance with each attempt to accomplish the stated goal and a Y% chance to kill myself or otherwise render the challenge unconquerable, now let's roll dice until I win or wreck myself."

In this mode of thinking, your degree of success is how many rounds it takes you to pick or jam the lock. In time-sensitive situations, though, you are truly reduced to a binary state - either you did the thing in a single roll or you didn't and whether the lock is jammed or not doesn't matter because the lava tide has risen past the platform the safe is on.

Modeling tasks that are not so cut and dry - negotiating with the kobold tribe, cooking dinner - using this same paradigm is going to result in the exact problem Frank is talking about. Applying the same thinking about success to these more open-ended tasks doesn't work - when I grill one hamburger, it might be burnt or raw or perfect, but as long as I didn't drop it in the cat box, what I did is going to be recognizable as dinner. My efforts are not being measured against some Platonic hamburger, so I don't have to keep grilling burgers until I get a perfect one; it's not useful to think of what I've done as a binary "success" or "failure" in the same manner as a picked lock.

This is where the bears come in - the MC wants to give the chef credit for doing a better job making the hamburger, but the MC doesn't have mechanics for modeling how satisfied the customer is with the chef's interpretation of "medium rare" because the system designer wasn't insane. Here's where the MC might make up dumb shit like "you rolled a nat 20 on your cooking roll, the person who eats your hamburger cries with delight and gives you $1,000 and a handjob." That's dumb, but there's no advice in the book for the MC as to what those degrees might look like for any given task. The best you the designer can do is provide the MC with shorthand like "exceptional success," "superhuman task," or "success at a cost..." :nonono:

I don't have the answer here. I don't know how you can produce a system that offers granular degrees of success that are meaningfully distinct from each other on a conceptual level. And don't tell me it's Shadowrun 4, because even something like "you got X hits on a task that required Y hits" requires an intepretive framework that simply isn't there in the majority of cases. What does it mean that I rolled 3 hits versus 4 hits on a Cooking check that only required 2 hits, and how do the rules help me to understand that meaning? What is the difference between a Cooking check that requires 2 hits and one with 3 hits?

A dicepool/hit system tells me that someone who gets 6 hits on a test with threshold 3 has exceeded the minimum expectations by the same amount as someone who got 4 hits on a threshold 1 test, but when the counted hits are applied to an open-ended type of check, that number of hits doesn't represent anything concrete that my brain can latch on to. In this regard it is no more useful than saying Person A rolled a 40 on a DC 10 check while Person B rolled a 60 on a DC 30 check.
Last edited by Mord on Fri Dec 29, 2017 5:52 am, edited 1 time in total.
User avatar
maglag
Duke
Posts: 1912
Joined: Thu Apr 02, 2015 10:17 am

Post by maglag »

FrankTrollman wrote:Die rolls are not inherently problematic. Indeed, die rolls are necessary if there is any disagreement possible as to the outcome of an action.

What is problematic is die rolls whose output is success and failure.The binary output leads to bad answers to most questions. When I drive to work, I do not fail to get to work. I maybe get there early or late. When I cook dinner I do not fail to produce food, it just sometimes tastes better than other times.
You may be some godlike entity, but we mortals sometimes burn food or crash cars. There's whole business around that like insurance companies.

And hospitals. Ever heard of medics? They are the real world healbots, patching up people who critically failed their rolls. But of course you would've never met a medic since you never burned anything in accident or bumped into something hard or got sick. :wink:
FrankTrollman wrote: Usually the die roll should output an answer to the question "How well did you do?" with "success" and "failure" being in the discussion only if you are competing with someone else.


"How well did you do?" is still a pretty open space. Walking out of a car accident or house fire on your own legs still breathing is considered doing pretty well by most real world people (in contrast to ending up crippled/dead).

But in D&D even crippled/death can be easily fixed. Your character losing an arm or a head is a minor incovenience in the grand scene of things and you just take a bit longer to win even if you need to retreat and ressurect before having another go. Heck, one of the cornerstones of mage domination is how good they're at running away from trouble.

Plus hey, plenty of groups consider it pretty hilarious when a party member royally screws up. You don't really hear game stories about when everything went right, you hear game stories about when half the party died and the other half scrambled to not lose.
FrankTrollman wrote: Actually, our blood banking system is set up exactly the way you'd want it to be if you were a secret vampire conspiracy.
Username17
Serious Badass
Posts: 29894
Joined: Fri Mar 07, 2008 7:54 pm

Post by Username17 »

Mord wrote:It's difficult to categorize degrees of success to any given task in a way that doesn't result in bears, because the classic RPG thinking about "checks" is couched in terms of tasks with binary outcomes. Did you pick the lock, leap the chasm, or lift the portcullis with this attempt? "I have an X% chance with each attempt to accomplish the stated goal and a Y% chance to kill myself or otherwise render the challenge unconquerable, now let's roll dice until I win or wreck myself."
If you're just rolling until you succeed the question isn't "Do I succeed?" it's "How long until I succeed?" And when you are asking a question of how long or how many or how well then binary answers don't even make sense.

When it comes to stabbing fools in the eye, a success/failure mechanic is good enough. Especially if you have a damage roll to determine answers to questions of how deadly that stabbing ends up being. But when it comes to any of the classic Rogue skills, that kind of mechanic is essentially worthless. AD&D turned forty this year, and in that entire time not one person has satisfactorily answered what it actually means to have a 37% chance to Hide in Shadows.

The question on sneaking is "How well are you sneaking?" That is, how much can you get away with before you are seen by the enemy? The question on lockpicking is "How long does it take you to get through the door?" And so on.

None of those things are amenable to binary success and failure because they are not yes/no questions to begin with.

-Username17
User avatar
Cervantes
Journeyman
Posts: 129
Joined: Mon Jul 28, 2014 10:27 pm

Post by Cervantes »

Are there mechanics for "how long until I succeed?" for 3.5e, because that's been an issue in games I've been in. To the point where our DM has unabashedly suggested the 2e rule of "two strikes and you're out" for bashing open locks.
User avatar
RobbyPants
King
Posts: 5201
Joined: Wed Aug 06, 2008 6:11 pm

Post by RobbyPants »

The only two things I know of are
  • Take 20 - takes 20 times as long, but you auto-succeed, assuming you could succeed on a natural 20.
  • Keep rolling, keeping track of how many rolls, until you make it.
I don't know of any simple rule, though. I mean, you could figure out your chance to make it in 20, take the reciprocal, and multiply by 6 seconds (or however long the action takes). That's kind of annoying, though. It takes out all the randomness, which could be good or bad.
User avatar
Cervantes
Journeyman
Posts: 129
Joined: Mon Jul 28, 2014 10:27 pm

Post by Cervantes »

Take 15, takes 10 times as long, makes sense since there's a 97%~ chance of getting a 15 with 10 rolls.

You'd probably want to have DC thresholds for time instead of trying to do a function from roll -> time spent. Something like:
  • DC-10: Succeed, takes four times as long
    DC-5: Succeed, takes twice as long
    DC: Succeed in baseline time.
    DC+5: Succeed, takes half as long.
    DC+10: You know what's up
Last edited by Cervantes on Fri Dec 29, 2017 9:56 pm, edited 1 time in total.
Mord
Knight-Baron
Posts: 565
Joined: Thu Apr 24, 2014 12:25 am

Post by Mord »

FrankTrollman wrote:The question on sneaking is "How well are you sneaking?" That is, how much can you get away with before you are seen by the enemy? The question on lockpicking is "How long does it take you to get through the door?" And so on.

None of those things are amenable to binary success and failure because they are not yes/no questions to begin with.
Right, but there is a binary component to them - on time-insensitive lockpicking challenge, the outcomes are "jammed lock, no longer possible to attempt" and "picked lock in X attempts" where X ranges from 1 to forever. There's not really any room for doubt on whether the lock ends up picked or not. Stuff like lifting, jumping, etc. follow the same mold.

On a cooking challenge, what are the degrees of success and what do they mean? The outcome, a grilled hamburger, could be anywhere on the blue to well done spectrum without being an inedible lump of carbon, so it's preposterous to say something like "after some initial failures, you took 18 rounds to grill a hamburger to perfection." The ideas that there are 17 grilled hamburgers you threw out or that you spent 18 times as long cooking one burger are both ridiculous. What does it mean that I rolled 4 hits against a threshold 1 Grill challenge?

Whether it's grilling or negotiating, open-ended questions where the ultimate outcome is somewhere in a really broad space that could still qualify as "success" are the ones that I don't know how to answer without introducing "success at a cost" bears.
User avatar
deaddmwalking
Prince
Posts: 3459
Joined: Mon May 21, 2012 11:33 am

Post by deaddmwalking »

Craft rules determine how quickly you create an item based on cumulative checks. A really good roll can allow you to create the item significantly faster than 'normal'. Something about converting the price into silver pieces etc.
-This space intentionally left blank
hyzmarca
Prince
Posts: 3909
Joined: Mon Mar 14, 2011 10:07 pm

Post by hyzmarca »

Mord wrote: Right, but there is a binary component to them - on time-insensitive lockpicking challenge, the outcomes are "jammed lock, no longer possible to attempt" and "picked lock in X attempts" where X ranges from 1 to forever. There's not really any room for doubt on whether the lock ends up picked or not. Stuff like lifting, jumping, etc. follow the same mold.
As someone who has opened locks as part of a job, I can say that this isn't true A jammed lock just means that you need to use more destructive means, like a power drill. It's fairly easy to create a success range which a high number of successes mean that there is no sign of picking and a low number means that the lock is obviously damaged or destroyed.


On a cooking challenge, what are the degrees of success and what do they mean? The outcome, a grilled hamburger, could be anywhere on the blue to well done spectrum without being an inedible lump of carbon, so it's preposterous to say something like "after some initial failures, you took 18 rounds to grill a hamburger to perfection." The ideas that there are 17 grilled hamburgers you threw out or that you spent 18 times as long cooking one burger are both ridiculous. What does it mean that I rolled 4 hits against a threshold 1 Grill challenge?
That's easy. 1 Hit = Edible, 2 Hits = fast food good, 3 hits = restaurant good, 4 Hits = people will comes from other cities to eat this burger.

Whether it's grilling or negotiating, open-ended questions where the ultimate outcome is somewhere in a really broad space that could still qualify as "success" are the ones that I don't know how to answer without introducing "success at a cost" bears.
Success at a cost fails when the cost has nothing to do with the actual test. If you're trying to sneak into a building without leaving any evidence, then "you obviously break the lock" isn't a bear, it's something that the forensic scientists can be expected to notice.

Likewise, cooking a hamburger good enough to appear on the food network isn't a bear, it's just cooking a really good hamburger. Because it's easy enough to create a cooking success table where the results range from "won't kill you" to "delicious."
Post Reply