D&D 5E 5E skills and the Perception vs Stealth imbalance

[Shiroiken]

The fact is, sneaking (Stealth) and detection (Perception) aren't born equal.

A +6, say, to Perception helps the party much more than a +6 to Stealth helps the monsters.

This means that 5E's idea to drop racial bonuses to stealth and not even grant most monsters proficiency does. not. work.

If five monsters sneak up on five adventurers, the math makes it exceedingly unlikely all five monsters will beat the best passive perception score in the group, and thus pull off an ambush.

This needs to change.

I disagree with much of your premise, but I can solve your problem right now. You are interpreting Stealth, a set of rules deliberately designed to be vague (in order to allow DMs to interpret it as they desire), in a way that makes it bad for your game. All you have to do is simply change your interpretation! Don't call for the check while the creatures are moving into position (actually better if they were already there beforehand), but only when Initiative is called for. Remembering that most groups have light sources (which can be seen for a LONG way off) and a ton of monsters have darkvision, monsters should be able to ambush traveling PCs on a pretty regular basis. If you want to ambush a camping party, remember that only awake characters get checks, and you can decide when to have the attack occur. If you need to roll for the monsters moving into position or if you want to allow the chance to warn the party, only roll once for each group of monsters with the same modifier (this will also speed up play).

A for your suggestions:

a) set passive perception at only 5 + prof + stat. If you are a designed guard that does absolutely nothing else, you gain the regular 10 + prof + stat, but only for a maximum of 1 hour per long rest. You simply can't maintain such vigilance for a whole guard shift. The idea is to force the players to guess when an ambush happens; if they're right they get to benefit from the generous RAW; but if not, well...
b) success means only that you gain awareness in the last second, preventing yourself from being surprised. You need to beat the Sneak check result by 5 or more to expose the ambush before it can be pulled off.
c) When I say you gain awareness, this includes a designated ward - a character you have declared yourself to protect, assuming you stay within 5 ft at all times. This way, you can still use your Perception to save someone from getting surprised; you can push the ward to the ground to avoid the assassin's poisoned arrow (assuming you win initiative too). But you can't put the entire group on full alert, unless the sneakees roll very low.
d) be generous with racial sneak bonuses. Most animals should be given +4 to Sneak. Renowed sneakers should gain +8. This means an Owl will not have simply +3 Stealth for +2 proficiency and +1 Dexterity; it will have +11.

a) this could work, but I would drop the ability to gain normal passive perception. PP is already assuming that you're paying attention the best you can, as long as you can.
b) many DMs already do this, except for the part about warning the party. I only allow an Active check against a predetermined DC to warn of an ambush if someone takes the "Watching for Danger" Travel Action. When dealing with a room to room dungeon setup, you have no chance to warn the party in my game.
c) I consider this very generous
d) I agree with granting more skills to monsters (including the occasional Expertise), but this is WAY too much IME! Remember, an owl isn't trying to hide from humans, but from small animals that are probably less perceptive. The problem with is is that you fall into the trap of 3E skills, where ONLY characters that are trained really have a chance.

 

[Jester David]

I think racial modifiers solve this neatly for animals or other creatures whose CR and HD doesn't merit a very high proficiency bonus, and where a very high Wisdom isn't appropriate.

Of course you could call it "Expertise", but:
1) "double proficiency" would only add +2 or +3 for the vast majority of critters this is about
2) I am not sure what would be gained by suggesting these critters are Rogues, that somehow isn't more directly adressed by using the concept of racial skill modifiers.

I guess we could create a feat called Natural Sneak that grants, say, a +5 modifier to your Stealth checks, but... *shrug*

I kind of understand that since feats are optional this can't work universally.

Besides, again, not sure what would be gained by having only one level of "racial sneakery" when two is better. I mean, it's still a very simple concept to grasp and use.

You don't have to call it anything: it's just in their statblocks. A few monsters already add double their proficiency bonus to some skills, as described on page 8 of the Monster Manual. Goblins and bugbears for example.

It wouldn't be very hard to create a variant monster that's stealth based for an ambush encounter.

 

[robus]

Before starting: Zapp could you please use the multiquote function? Of 18 (well, 19 with mine) posts of this thread, half of them are yours and many have with no rime or reason to be stand alone. To me this is borderline unreadable. I know there's no rule against multiposting and i'm no mod. I'm just asking this as a favour, no more no less. Ty.

While this makes the thread perhaps a bit cleaner, it actually hinders conversation. Responding to one part of a multi quote post is quite an editing exercise. Keep responses short and to the point to keep the conversation going and don't be redundant IMHO.

 

[ClearlyTough69]

Just as I make a single initiative roll for several monsters of one type, so do I make a single Stealth check for several monsters of one type (not to be confused with a group check). By contrast, when making checks for monsters individually, each extra check adds to the chances of a low roll, making it more likely that even a PC with poor passive Perception will notice a threat and so not be surprised. It also keeps down the number of dice to be rolled: a good thing in my book.

So in the example of the wolves, I would make a single Stealth check for the whole pack, giving them a better chance of remaining hidden from at least one or two PCs. If they can achieve this, then those PCs are surprised - a big advantage for the wolves.

If I do roll Stealth for monsters individually, say because they are a mixed group, then there's a good chance that at least some of the monsters remain hidden, even if none of the PCs is surprised because one monster rolled poorly. This in itself is a good thing for the monsters, especially if they have ranged weapons, as they may be able to benefit from advantage on their attack rolls for being hidden - not so good for wolves, mind: they're not renowned for their archery!

TL;DR the power of an ambush is twofold: it may render some creatures surprised, and it may allow hidden ambushers to attack with advantage. To stack the odds in the ambushers' favour, roll as few Stealth checks as you can by making one roll for groups of similar monsters.

 

[Horwath]

The fact is, sneaking (Stealth) and detection (Perception) aren't born equal.

A +6, say, to Perception helps the party much more than a +6 to Stealth helps the monsters.

This means that 5E's idea to drop racial bonuses to stealth and not even grant most monsters proficiency does. not. work.

If five monsters sneak up on five adventurers, the math makes it exceedingly unlikely all five monsters will beat the best passive perception score in the group, and thus pull off an ambush.

This needs to change.

I want the following:
*) a group should not be able to rely on a single good Perception score
*) good Perception should chiefly benefit yourself and possibly a ward of yours, not an entire group
*) the rules need to favor ambushers much more, so that ambushes actually happen with some regularity.
*) the game is much more fun if dark forests and caverns actually ARE dark, foreboding and scary. Simply having a character with max Wisdom and Perception proficiency should not shortcircuit all that. Sure diligent defenders should be able to take precautions such as setting up perimeters (light torches that illuminate intruders etc), but that's an exception and not the rule.
*) monsters that are supposed to be sneaky should actually be sneaky, even level-appropriate heroes should have a hard time noticing these critters.

The basic notion of passive scores is good, since we can't have both Stealth and Perception rolled - it creates too large swings. So we'll keep that, but we'll considerably rein in what Perception can do for you.

Therefore I suggest:
a) set passive perception at only 5 + prof + stat. If you are a designed guard that does absolutely nothing else, you gain the regular 10 + prof + stat, but only for a maximum of 1 hour per long rest. You simply can't maintain such vigilance for a whole guard shift. The idea is to force the players to guess when an ambush happens; if they're right they get to benefit from the generous RAW; but if not, well...
b) success means only that you gain awareness in the last second, preventing yourself from being surprised. You need to beat the Sneak check result by 5 or more to expose the ambush before it can be pulled off.
c) When I say you gain awareness, this includes a designated ward - a character you have declared yourself to protect, assuming you stay within 5 ft at all times. This way, you can still use your Perception to save someone from getting surprised; you can push the ward to the ground to avoid the assassin's poisoned arrow (assuming you win initiative too). But you can't put the entire group on full alert, unless the sneakees roll very low.
d) be generous with racial sneak bonuses. Most animals should be given +4 to Sneak. Renowed sneakers should gain +8. This means an Owl will not have simply +3 Stealth for +2 proficiency and +1 Dexterity; it will have +11.

There isn't an actual problem in this.

In darkness even with darkvision you have -5 to passive perception if the check is relyning on sight, and most ambushes are silent as you are already in place and waiting for suckers to walk into it.

If it is good ambush site you can give advantage to stealth check for performing an ambush.


also why should an alert character have hard time with stealthy character? arent they best at their job? even with same bonuses ambusher wins in 55% without any modification to the roll or passive perception DC. That is good enough for me.

 

[CapnZapp]

Before starting: Zapp could you please use the multiquote function? Of 18 (well, 19 with mine) posts of this thread, half of them are yours and many have with no rime or reason to be stand alone. To me this is borderline unreadable. I know there's no rule against multiposting and i'm no mod. I'm just asking this as a favour, no more no less. Ty.

Reply by PM sent.

 

[CapnZapp]

Don't call for the check while the creatures are moving into position (actually better if they were already there beforehand), but only when Initiative is called for.

Sure.

This is also a gameplay mode I detest when playing myself.

I fully understand that people aren't having issues with Stealth if they simply don't use the skill rules until the monsters are within 30 ft or 60 ft and it's time for Initiative.

But as a player I can't stand the notion that no matter what you do, monsters that doesn't possess any remarkable abilities or any noteworthy skill bonuses, can sneak up so close to you. Especially since your character's skills might beat them all.

I far prefer an alternate way of playing where, if four critters do manage to come within charging distance, that if the numbers are examined, it actually makes statistical sense.

This thread is therefore meant to explore ways to reconcile what happens around the table with actual numbers and die rolls.

For instance, I've gotten the suggestion "only roll for the character(s) in first rank" (or "only the first rank character gets a chance of spotting the threat"). I'm taking this to mean "only involve the characters nearest to the threat". But I would much rather the game explains exactly how this works. If you stand just behind the first character, do your chances really drop from outstanding to non-existant.

And by giving monsters racial bonuses, it gets much easier to swallow a group of them all beating my Perception score. I know individual DMs can (and should fudge), but I don't like it when the rules themselves essentially fudge it.

I guess it boils down to me wanting stealth to be simulationist and not narrative.

 

[CapnZapp]

The problem with is is that you fall into the trap of 3E skills, where ONLY characters that are trained really have a chance.

If sneaky critters can't even surprise untrained oafs in plate helmets that are as blind and deaf as random peasants (without the DM skipping the sneaking part and moving right up to the ambush part) that game has Stealth rules worse than non-existing.

This doesn't mean I don't see the value in bounded accuracy.

But any rules for Stealth must have a clear vision and takes into account the fundamental mathematical underpinnings.

I mean, if the designers added an "ambush bonus" that told me they were aware of the inherent advantage of the perceptors over the ambushers, it'd be okay.

One suggestion has been "have ambushers make a group check". This might not be my personal favorite solution, but it does address the math. By allowing the group to succeed even if 2 out of 5 fail, you have shown a basic awareness of the statistics.

 

[Shiroiken]

Sure.

This is also a gameplay mode I detest when playing myself.

I fully understand that people aren't having issues with Stealth if they simply don't use the skill rules until the monsters are within 30 ft or 60 ft and it's time for Initiative.

But as a player I can't stand the notion that no matter what you do, monsters that doesn't possess any remarkable abilities or any noteworthy skill bonuses, can sneak up so close to you. Especially since your character's skills might beat them all.

why are the monsters moving close? The best ambushes have the target come to them. The monsters then aren't moving, so it makes sense that you only roll once. As far as distance, that's up to you again. You can wait until only 30', or you can call for it at 100', or a 1000' if you wanted. The farther away you call for the check, the less important the result is.

I far prefer an alternate way of playing where, if four critters do manage to come within charging distance, that if the numbers are examined, it actually makes statistical sense.

This thread is therefore meant to explore ways to reconcile what happens around the table with actual numbers and die rolls.

Houserules are a great way to fix this. As I noted, some of your houserules seem reasonable. I just feel that you might inadvertently make Steath too strong. Remember, whatever benefit you give to stealth, you give to your players too. I currently have a 12th level rogue that almost never has to roll Stealth because none of the monsters can see him (minimum Stealth check of 23, not considering -5 for Dim Light). Also, again you can also have a single roll for all monsters, which removes the statistical deviation.

 

[CapnZapp]

even with same bonuses ambusher wins in 55% without any modification to the roll or passive perception DC. That is good enough for me.

The probability of one monster with +5 passing a DC 15 check is indeed 55%.

However, the probability of two monsters both passing is only 55% x 55% or 30,25%.

More generally, the probability of N monsters all passing is 55%^N. For N=5 the probability is only about 5%.

If you only make the check when the monsters are in position, thus turning the check into the answer to the question "who's surprised?" this might be okay.

But if you use this check to determine if the heroes spot the monsters already when they try to sneak up on the party, when there is still time to do something about the ambush (such as fleeing etc), this doesn't work at all, since it means only one in twenty ambushes actually happens.

---

If I agree that your number 55% is good enough for me too, I want the math to result in 55% of all ambush attempts to actually succeed. What happens on the "surprise round" is an entirely secondary concern to me.

Now, using group checks would get us there. The probability of five out of five checks succeeding might only be 5%, but the probability of at least three out of five checks succeeding is 59%.

---

Now the statistics of group checks swings rather heavility based on the number in the group. I found an old thread complaining about how the probabilities actually decrease if a group of two unskilled characters are joined by a third, even if that character is highly skilled. The reason, of course, is that the addition of the third character means that now 66% of the group needs to succeed compared to only 50% of the original two-man group. That is to say, one out of the two unskilled characters STILL needs to succeed: all the addition of the expert has done is added the risk of that expert failing (however small).

To illustrate this, compare the above scenario to when one of the five monsters decide to hang back. The probability of setting up the ambush drops from 59% to 39%!

Probability of 3 to 5 checks out of 5 making it: 59%
Probability of 3 to 4 checks out of 4 making it: 39%

So while group checks should probably be standard, they do add a measure of unpredictability that might catch DMs unawares.