Perception and Stealth

[Earthwalker]

I was playing some RuneQuest 4 this weekend. As we were walking in the wilderness the GM calls for perception tests and we all roll. Now we are “low level” characters so have about 40 / 50 % in perception. With all the characters rolling there is a real good chance one of us will make the perception test. We spot the trouble and work to avoid it.

Later on in the same game the GM wanted to see how quietly we were moving. So he calls for a stealth test. Again we all have about 40% in this. A few of us made it but one guy failed. As we failed we made too much noise and the elk we could have bagged for dinner ran off into the forest.

This got me thinking of the perception / stealth dynamic. Perception is general one pass all pass. Stealth is generally one fail all fail. With 4 players and about 40% skill you are looking at perception = pass. Stealth = fail.

Now there are ways to solve this as player, like sending a scout out up front who has good stealth. Of course then his pass and fail is on him. The first encounter we avoided was a dark troll. If in that situation you had failed a perception test and ended up in combat with a thing a single character would be dead before the rest arrived.

I have no real question just interested in peoples thoughts on this.

Also wondering how people would handle this or if they even see a problem at all.

[Jormengand]

Yes, that's called reality. It's why assassins usually work alone - yes, two people are more likely to be able to kill someone in a fight, but one person is more likely to be able to get there. You cannot and should not be able to sneak an entire army of ninjas into a castle no matter how stealthy they are.

[Yora]

My approach is to have one or two scouts sneak on ahead, but not very far. Just so much that the rest could come rushing to their aid within a round. If possible maintain line of sight with the scout, so he can signal the others to stop while he comes back to them to tell them what he saw.

[Thrawn4]

I agree with the others, it's just common sense that an army is less sneaky than a single person. Of course that can pose a problem in an RPG, but that's why you usually just send a scout. A clever DM can also give an incentive for a single person to sneak past: Get a key, find the information and so on. Being caught by a powerful enemy is a huge risk for that person, but in most systems a rogue/whatever can fend for themselves for a combat or two. Also, not every encounter has to be deadly - maybe they want to interrogate or cook properly.

[Earthwalker]

I can see what people are saying and I agree it is easier to sneak one person than an army. Not that we are an army but in fact three characters.

Even with 3 it was a case of

Perception = succeed.
Stealth = Fail.

I do think its harder to sneak with three people as opposed to one. I just found it was so much harder.

I could see handling both these tests as a team work test in fate for example.

So that you only ever roll one set of dice. In stead of rolling three sets of dice and checking pass / fail on each.

The same could be done in RQ as well I guess. Roll one perception test. +10 % for each other person there. Roll one Stealth test -10% for each person there.

[Lord Torath]

I think the Expert or Companion D&D rulebook had rules for evading searchers, with smaller groups being better able to hide, and larger groups being better able to find hiders, up to a point of about 10 or so searchers in one group. After that, more searchers made it harder to find the hiding group.

I'm not sure how to apply this to specific individuals with differing Stealth skills, though. Maybe use one check at the level of the least-proficient PC in the group?

Or have the sneaking PCs each make their stealth test. Then each success penalizes the perception check of the group looking for them. That way, failing a stealth check doesn't mean you're automatically detected, just that you're more likely to be detected.

[goto124]

Why is it annoying? Becauses it forces the party rogue to go work alone, leaving everyone else behind...

So we're trying to solve that problem. Someone suggested a while back for the rogue improving the entire party's stealth, fluffed as the rogue 'clearing' or 'finding' a suitable sneaky path for her friends.

[Maglubiyet]

Regarding the Perception part -- this is why many prey animals travel in herds. More eyes means more Spot checks, making it harder for predators to stay hidden.

When predators are also hunting in packs they can game on this behavior. One or more individuals intentionally expose themselves to the prey. The prey is then Distracted and takes a penalty on Perception checks to detect the other hidden predators in the group. The predators use this to close distance or set up an ambush.

Animal behavior as explained through the medium of RPG's.

[Mr. Mask]

There are a couple of elements which might help.

Bell curve dice rolls may be useful, particularly if you're trying to beat a certain score.

Different levels of success and failure would be helpful. If one person is spotted, are the others spotted or noticed? Perhaps the guy who is spotted still gets a surprise round because the spotters are surprised? Maybe the guy who was spotted doesn't even know he has been spotted, and is now "sneaking" into an ambush?


Perception rolls do need to be managed in some way. If you try to sneak through an army camp, do you roll perception 10,000 times? If you roll enough times, it won't represent the actual chances of the situation, but instead will be artificial chances. The main way to structure accurate perception rolls would be to compare real world chances, but that is difficult to estimate. You can to an extent by simplifying it. Like, how likely is a regular guy, a clumsy oaf, Batman, and Robin to infiltrate X scenario? You then work out reasonable %s, and arrange your perception and stealth system around that.

[erikun]

This is a fairly common problem in many RPGs, called the "Roll Until Fail" situation. Basically, any non-100% chance of success a player has at any one task gets progressively more difficult the more times it is required to be rolled.

For example, if a player succeeds with a 2 on a d20 when climbing a cliff, then that's a 95% chance of success. If they need to roll twice for the climb check, that's 90% chance. If they need to roll three times, then 85% chance. By twelve rolls, the only-fail-on-a-1 character has only a 50% chance of succeeding the entire climb. Needless to say, the odds are a fair bit worse for a character who needs a 11 on d20 (50% chance) on the initial success.

There are a couple of solutions I've seen that try to fix the problem. One is to not make failure cause the entire situation to fall apart. Using the climbing example from above, tying a rope around everyone and using pitons would (at least somewhat) ensure that any one failure only means the loss of progress on that single roll, rather than forcing a restart from the very beginning. On a stealth check, it would mean alerting only the group of guards the party ran across, as opposed to alerting the entire enemy camp. Another option is to make a single roll and have it hold for the entire scene. That is, if a player rolls a 35 on their stealth check, then that 35 is tested against every single perception check they come across - although such a solution would need a "roll once and take that value" rule, to prevent players from just trying to adjust something minor for a chance at forcing a re-roll. And other people have mentioned other options, including just not having the PCs attempt to sneak around in a group. (One game session I've played involved a dwarven cleric just casting Silence on himself, sticking with the heavy-armor characters, and using darkvision to follow around the stealthy rogue outside the Silence spell.)


[Edit] I should also note that some games develop a sort of shorthand for perception checks, just to avoid situations where it's one group of five rolling against a single stealthy target. Generally, it's something like "a group gets a bonus based off the number of people working together to scan the area" and puts it down to one roll, and either just using one roll for the highest group's bonus, or splitting stuff into several large groups which roll separately.

e.g. A group with 10 people, with three scanning the area, one casually looking around, and the others eating would probably just use the perception on one of the three scanning the area, with bonuses for the other two assisting. The one casually looking around isn't assisting the others, so they wouldn't grant a bonus or assist the others. Although you might use theirs if their perception bonus is higher than the others put together.

[Mr.Moron]

This is why I generally prefer group skill test dynamics. That is where either some form of group check with each player contributing dice/modifiers to a communal pool and making a single check is made for everyone, or where one player makes a check an additional people add/subtract depending on how good bad they are.
h
It's really not that hard to bolt on to systems that don't do it by default, which is what I typically I do.

[Satinavian]

If this is a problem or not depends on the dice system used. When you use a system with somewhat consistent results and degrees of success/failure to compare, then it's not a problem. Yes larger groups will still spot more and be louder, but competence is the main factor and the differences are pretty realistic.

If you use other systems that try to bring uncertainty in all rolls, have a huge variance or similar, just use a single roll and modifiers for numbers of people.

[erikun]

If this is a problem or not depends on the dice system used. When you use a system with somewhat consistent results and degrees of success/failure to compare, then it's not a problem.

It becomes an issue whenever something with a set chance of success gets called upon repeatedly, where a single failure is the only one that matters. It doesn't matter if the 90% chance of success comes from a d20, or a 3d6, or a 4dF, or whatever combination of dice are being used. If the roll gets tested repeatedly, then failure will happen at some point.

The only system I've really seen that tried to solve this was Burning Wheel, which takes one roll for the "scene" and uses that for all applications. I haven't seen a system yet that gives progressive bonuses upon each success (which would elimiate the problem) although with the climbing example above, there are situations where a single failure doesn't mean the complete loss of progress.

I am curious of what dice system you are talking about, though. Just in case there was something I missed.

[Satinavian]

It becomes an issue whenever something with a set chance of success gets called upon repeatedly, where a single failure is the only one that matters. It doesn't matter if the 90% chance of success comes from a d20, or a 3d6, or a 4dF, or whatever combination of dice are being used. If the roll gets tested repeatedly, then failure will happen at some point.

If you have to pass two rolls instead of one, a chance of 90% drops to 81%, but a chance of 60% drops to 36%, nearly half it's value. If you have a system, where a competent character gets a 95% chance and a less competent 5% on the very same task of moderate complexibility, an easy task leading to 98%/50% and a hard task to 50%/0%, the number of people rolling will not nearly matter as much as their abilities. But your average D%-system does't work this way because it uses a uniform distribution.
If you introduce dregree of successes and compared rolls, your statistic becomes even more complicated. For example a good scout will have a certain probability of achieving numbers that are impossible for the other side to overcome, rendering the number of opposed rolls unimportant while being subjected to them in the normal way, when he rolls low himself.

So yes, the system does matter a lot for this question.

One of the system i play atm uses a pretty atrocious and complicated four-parameter-three-dice probability distribution which is difficult to explain. But in many cases it can be approximated roughly as a Gaussian with the mean as on of its parameters (the others influence mostly higher momentums and yes, skewness is a thing). I found, that the above problem does not really arise in that system and i can let everyone roll. While it is still important to let everyone behind who can't sneak if you are scouting, an additional scout doesn't hurt much and that additional sentinals provide more benefits due to being able to distribute them to all important places, less because of multiple rolls. I am sure, you could have similar effects with e.g. a 5d6+x system or even some xd6+y, if you account for degree of success.

I really dislike uniform distribution based systems.

[nedz]

I have played / run games with an all high stealth party. This had a different mechanic in that the chance of individual failure was tiny and if someone did screw up then they would just pull out and cause a distraction elsewhere. It is possibly to sneak an entire party into position, albeit with the occasional hilarious result.

[Mr.Moron]

It becomes an issue whenever something with a set chance of success gets called upon repeatedly, where a single failure is the only one that matters. It doesn't matter if the 90% chance of success comes from a d20, or a 3d6, or a 4dF, or whatever combination of dice are being used. If the roll gets tested repeatedly, then failure will happen at some point.

The way we handle this in my IK game which is more house rule than RAW in that system, but kind of uses a framework found elsewhere in the engine is this:

For any given check to which people might contribute I first establish exactly how many people it's plausible would contribute. Then I'll tell the group "Hey make a detection check against 15, with one primary and 3 assists". They'll talk among themselves about who should be primary (generally, but not always the person with the highest modifier).

The secondary characters make the check against the DC -2 to -4, and if they pass the primary gets +1 for each assistant that passed.The primary's result is then the only one that counts. Note that the system is 2d6+mod, so a -2 actually makes the secondary checks much easier.

The idea here is that the dice represent a combination of luck, local conditions, timing and other factors that aren't encapsulate perfectly in an ability score. If players are all doing mostly the exact same thing, in mostly the same way, at the same time, in the same place opposed to the same target much of the "Dice" factors are going to be similar or the same for most of them. This means that on the abstraction layer we really only want a single die-roll "Counting".

In practice it means:

Primary: 14 vs 15
Secondary: 13 vs 13 (Pass: +1)
Secondary: 14 vs 13 (Pass: +1)
Secondary: 12 vs 13 (Fail)

Is a success, even if nobody in the group got 15. And

Primary: 6 vs 15
Secondary: 18 vs 13 (Pass +1)
Secondary: 12 vs 13 (Fail)
Secondary: 11 vs 13 (Fail)

Is a failure for the group, even though making four rolls managed to fish an 18 by rolling double sixes at some point.




However having only the highest stat player roll is kind of boring and doesn't consider that having more people is an advantage (just not as big a one as independent rolls would make it). So other players still get to make rolls but only as part of a contribution to the total effort.

[nedz]

The way we handle this in my IK game which is more house rule than RAW in that system, but kind of uses a framework found elsewhere in the engine is this:

For any given check to which people might contribute I first establish exactly how many people it's plausible would contribute. Then I'll tell the group "Hey make a detection check against 15, with one primary and 3 assists". They'll talk among themselves about who should be primary (generally, but not always the person with the highest modifier).

The secondary characters make the check against the DC -2, and if they pass the primary gets +1 for each assistant that passed.The primary's result is then the only one that counts. Note that the system is 2d6+mod, so a -2 actually makes the secondary checks much easier.

The idea here is that the dice represent a combination of luck, local conditions, timing and other factors that aren't encapsulate perfectly in an ability score. If players are all doing mostly the exact same thing, in mostly the same way, at the same time, in the same place opposed to the same target much of the "Dice" factors are going to be similar or the same for most of them. This means that on the abstraction layer we really only want a single die-roll "Counting".

However having only the highest stat player roll is kind of boring and doesn't consider that having more people is an advantage (just not as big a one as independent rolls would make it). So other players still get to make rolls but only as part of a contribution to the total effort.

So 1 stealthy character and 3 clutzes can sneak into Mordor ?

For characters helping each other do something your system is fine, but for one where the question is: "does anyone make a mistake" I'm not so sure.

[Mr.Moron]

So 1 stealthy character and 3 clutzes can sneak into Mordor ?

For characters helping each other do something your system is fine, but for one where the question is: "does anyone make a mistake" I'm not so sure.

In cases where I need to account for "Idiot ****s it up for everyone" I'll generally say something like this:

"This is an opposed stealth check. Whoever is leading you is primary and rolls first. Everyone else is secondary. If you get within 2 of the primary's score you manage to help keep things even stealthier and give the primary check +1. If you get 4 or more below the primary, you mess things up and reduce groups score:"

So in the stealthy guy and three clutz scenario you might get.

Stealth-Man: 15
Klutz A: 8 (Bad Check, -1 for primary)
Klutz B: 10 (Bad check, -1 for primary)
Kltuz C: 12 (Failure, but not enough to penalize)


In this case the groups final stealth score would be: 13, much easier to find than 15. For a regular mook guard (2d6+6) this means the difference between 9+(27%) and 7+(58%).


I could also go "If anyone rolls below 8, even on a secondary the group's cover is blown". However I'd be afraid that approach would just lead to party-splitting and more feel-bads than is needed (simply providing a penalty already feels pretty bad).

[Mr. Mask]

Mr. Moron's description fits what I've heard from guerilla teams. Even without training, it's possible to follow the other's lead, and do fairly well.

[goto124]

So 1 stealthy character and 3 clutzes can sneak into Mordor

When the alternative is splitting the party?

[Calen]

Why is it annoying? Becauses it forces the party rogue to go work alone, leaving everyone else behind...

So we're trying to solve that problem. Someone suggested a while back for the rogue improving the entire party's stealth, fluffed as the rogue 'clearing' or 'finding' a suitable sneaky path for her friends.

I often let my players, stealthy characters apply the excess of their rolls to help out their teammates. (4e game If the player exceeds the stealth check by a significant margin and they are trained in stealth then they give an ally a +2 on their check.)
- Rogue quietly points out loose stones and dry branches so that Sir Clanksalot doesn't blunder on to them. The party successfully avoids attracting the trolls notice.

By extension you can apply this to a wide variety of "All must Pass" checks.
Endurance - The barbarian takes part of the mages pack and helps him over rocks as they climb the forbidding mountain pass.
Diplomacy - The bard quickly turns her companions social blunders into jokes or the lead-in to a clever story.

[goto124]

Ooo, nice, never thought it could be extended to Diplomancy. That should help other party members to speak up, instead of having fears of failing checks and messing everything up.

[Earthwalker]

In cases where I need to account for "Idiot ****s it up for everyone" I'll generally say something like this:

"This is an opposed stealth check. Whoever is leading you is primary and rolls first. Everyone else is secondary. If you get within 2 of the primary's score you manage to help keep things even stealthier and give the primary check +1. If you get 4 or more below the primary, you mess things up and reduce groups score:"

So in the stealthy guy and three clutz scenario you might get.

Stealth-Man: 15
Klutz A: 8 (Bad Check, -1 for primary)
Klutz B: 10 (Bad check, -1 for primary)
Kltuz C: 12 (Failure, but not enough to penalize)


In this case the groups final stealth score would be: 13, much easier to find than 15. For a regular mook guard (2d6+6) this means the difference between 9+(27%) and 7+(58%).


I could also go "If anyone rolls below 8, even on a secondary the group's cover is blown". However I'd be afraid that approach would just lead to party-splitting and more feel-bads than is needed (simply providing a penalty already feels pretty bad).

I am thinking when I am running things I am going to use something similar to this for alot of group tasks. Make it so one poor roll out of a group doesnt have to cause the whole group to fail.

Thanks Mr Moron.

[Jay R]

I agree with the observation, but disagree that there is a problem to fix.

A group of ten people are far more likely to spot something. A group of ten people do have an extremely difficult time sneaking.

That's why the sneak role (2E Thief) is crucial, and why he always gets a bit more solo action, which helps balance the fact that he's worth less in melee.

[King of Casuals]

I always liked the idea of group stealth rolls, with one roll for the entire party. It usually happens when a stealthy rogue character gives the party advice on how to sneak, and they all follow his lead.

[Mr.Moron]

I agree with the observation, but disagree that there is a problem to fix.

A group of ten people are far more likely to spot something. A group of ten people do have an extremely difficult time sneaking.

EDIT (I'll be using d20 numbers in the examples in this post, just because that is what people are likely most familiar with)
The issue is the matter of a degree. A group of 10 people are generally going to have more eyes open (assuming everyone is being equally vigilant) but the dice abstract out a lot of things, and not all of the things in that abstraction are going to be different for every observer in group at the exact same moment.

For the sake of argument, lets assume we have a group of 10 folks resting after a long day on the trail. They're sitting around the campfire in a circle, enjoying dinner. They're all equally observant (+5). The skillful spy "Señor SneakPantys" sneaks up on them sneakily with his superior +10 and rolls well: 12. For a total of 22.

Taken as a whole the group has a 90% to spot him. This seems extreme to me. They're all in the same location, with the same lighting conditions, at mostly the same angle to the approaching Señor SneakPantys. Virtually all the variables that the die is used to abstract are the same but collectively spotting someone nearly invisible to any one of them is a sure-fire thing!

What I'm trying to say here is that I like to be able to represent the fact there are diminishing returns on additional participants for some types of group efforts.

it works similarly for stealth.

Imagine you have "Rebdar, The Supreme Commando" with +20 Stealth, leading a group of civilians past an enemy encampment, he's also got a solid +15 in his commanding people skill (which is sadly irrelevant if we're playing things straight).

The civilians (4) are untrained and have a +0. There is only one guard on duty, "Chip, the Incompetent" who has managed a whopping -4 to his perception rolls. He has also rolled poorly, getting an 8 for a total of 4.

My basic expectations would be that for Rebdar and his group this is a happy day. They've got a crack leader and the only one looking for them can barely spot his own nose.

Rebdars group has a 87% chance of being found by Chip who promptly signals (he manages to avoid fumbling is perform[Horn] check, depsite a -10) the more component soldiers in the fort to turn them into swiss cheese.

[goto124]

the only one looking for them can barely spot his own nose

I tried to look at my own nose. Pretty sure it's a DC 17 check.

[Grinner]

There's one element to this search/hide dynamic that hasn't been addressed. Unless you have some kind of telepathic connection in the group, one individual may spot a band of forest goblins waiting in ambush, but communicating anything more complex than their presence may prove difficult. ("Look! A goblin! Behind that tree!" "Wait...which tree?") Furthermore, when a guard hears Sir Clanksalot, that doesn't necessarily mean he automatically notices Sir Clanksalot's companions.

[goto124]

But when Sir Clanksalot gets attacked, the entire party must reveal themselves to save him.

[Grinner]

But when Sir Clanksalot gets attacked, the entire party must reveal themselves to save him.

Do they?