I notice that Pathfinder moves away from opposed rolls in favor of rolling against a target number - something I'm in favor of, as it homogenizes things pretty easily to be more in line with the core mechanic (rolling against a target number).
Which brings up that trying to sneak by someone or pick their pockets is still an opposed roll - which adds a step to resolving such issues (as well as making it more difficult to steal from your friends).
OD&D had a flat percentile chance for doing these things, and the subject wasn't even entitled to a roll - but people seemed to get along fine just then.
Do you think it's a good idea to give characters and NPCs a "Perception Target Number" equal to 10 plus their Perception modifier, and treat that as a target number to beat with Stealth and Sleight of Hand?
OD&D had a flat percentile chance for doing these things, and the subject wasn't even entitled to a roll - but people seemed to get along fine just then.
OD&D had it's flaws.
One of which was that it's just as easy to sneak up on a wide-awake, alert predator with keen predatory senses as it was to sneak up on a sleeping old man.
Seems to me there should be some input from the observer in this process. There should be some way to distingquish between stuff that's easy to sneak up on and stuff that's hard.
We just didn't know better back in the OD&D days.
Do you think it's a good idea to give characters and NPCs a "Perception Target Number" equal to 10 plus their Perception modifier, and treat that as a target number to beat with Stealth and Sleight of Hand?
This would make it way way way too easy for any lowbie (level 1) rogue to sneak around town freely, in and out of the king's treasure vaults as he pleases.
How many people in ordinary life have Perception as a class skill? None. Just a couple adventuring classes. I guess an Expert might have Perception on his class skill list, but they are rarely found among guards and watchmen.
How many people in ordinary life have exceptional WIS ability scores? Almost none.
So a rogue with a decent DEX, 1 rank in Stealth, +3 for class skill, could simply take-10 all over town and pretty much never have to worry about beating anyone's DC 10, 11, or maybe 12 passive Perception score.
Nobody in ordinary life would ever ever ever have a chance to perceive any rogue ever.
Not a good idea. Not by a longshot.
Now, maybe, if you forbid taking-10 on Stealth checks, so that even the best rogues would occasionally step on a twig or bump a vase or scuff their shoe, well, then yeah, passive Perception might be a workable concept.
Do you think it's a good idea to give characters and NPCs a "Perception Target Number" equal to 10 plus their Perception modifier, and treat that as a target number to beat with Stealth and Sleight of Hand?
Especially with Stealth. It makes sneaking past a crowd so much less punishing.
If your definition of "less punishing" is "automatic success".
I daresay, Mr. aMiB, and I don't mean this to sound like an attack, but your bias is showing again. It wouldn't have that effect if you seemed more willing to discuss Stealth from a balanced viewpoint.
But, to me at least, it seems your posts in other threads are more interested in turning Stealth into the perfect, unstoppable weapon rather than a balanced game mechanic.
It would be easeir to take your posts more seriously if they didn't come across this way.
Note that you cannot take 10 on Stealth checks, as there are consequences for failure.
Setting aside the fact that RAW there are lots of places you can't hide, there's not one guard with a decent perception check?
This would make it way way way too easy for any lowbie (level 1) rogue to sneak around town freely, in and out of the king's treasure vaults as he pleases.
A rogue automatically gets +3 to Stealth because it's a class skill. A guard dos not get +3 to Perception because that is not a class skill, unless the king is hiring rogues to guard his vault...
Further, sneaky rogues always have fairly good DEX scores. It's almost always their highest ability score. Guards usually do not have very good WIS scores. It's usually their 3rd or 4th (in some cases worse than that) ability score.
So the guard would have to be about 5 levels, or more, above the rogue, and would need to put half of his available skill ranks into Perception, to be able to see him on a simple take-10.
Yeah, maybe the king's vault guards have those kinds of levels and that kind of Perception skill, but it's not likely any guards anywhere else have those levels and also have put so many points into Perception when they have several other skill they likely don't want to neglect.
Or a lock on the vault?
That's a different discussion about disable device.
Plus, you can't take 10 in a threatening situation, such as around a bunch of guys who will toss you in a cell or stab you to death.
Not true. You can't take 10 when threatened, such as in combat. The mere threat of being seen so that you might have to run away or maybe face a combat is not threatening enough. By RAW.
Distractions or threats (such as combat) make it impossible for a character to take 10.
So you roll Stealth against a DC of 10 + the highest Perception mod,
No, you just take 10.
and if you succeed
You will. Almost always.
Super Ninja Rogue Uberness will have been achieved.
you sneak past the group. Less dice rolling, and it actually lets sneaky characters get to abuse being sneaky.
Fixed.
I don't really see the problem with letting the rogue sneak around busy city streets or sneak around at night without being seen save by a nearly-impossibly-perceptive foe,
Fixed again.
though. What else are you supposed to do with Stealth?
How about rather than sneaking around with guaranteed freedom to go wherever you want to, we discuss ways to make Stealth actually fair and balanced as a game mechanic.
That's what I'd like to do with Stealth.
Accusations of bias can be forwarded to the WOW forums, especially since you're accusing me of possibly wanting sneaking to work effectively and efficiently at the table, rather than being complicated, game-stopping, and really weak.
Wow, you totally caught me there.
I know!
Now, they say admitting it is the first step...
What baffles me is that taking 10 is being simultaneously described as guaranteed failure and guaranteed success.
Something that should be done before this is argued more is to calculate the odds of success of rolling against the target number of 10 + Perception mod versus opposed Stealth vs. Perception.
Ah, geez, math...
Let's assume that the guard has a Wisdom of 12 (+1 mod), and one skill point in Perception (because seriously, who hires guards that are terrible at spotting things?), bringing his Perception DC to 12. For a level 1 NPC, this seems pretty reasonable.
Now let's assume a level 1 Rogue PC with a Dex of 16 (+3 mod), a skill point in Stealth (+1), which is also a class skill (+3), bringing his total modifier up to +7. Not bad!
Against the target number, the Rogue has to roll at least a 6 to succeed, meaning the chance for success is 75%. Sneaking past one guard? Easy.
Now we make that an opposed roll - 12 becomes an average result for a guard, whose range can go as low as 3 and as high as 22. If the guard rolls 1 to 5, failure is automatic for him - and that's already 25% of the rolls he can make! Already there's a flat 25% chance for the Rogue to succeed before he even makes his own roll.
And when the Rogue rolls - if he gets a roll of at least 16, bringing his modified result up to 23, the guard can't possibly roll high enough to see him - another flat 25% for the Rogue to succeed.
Unfortunately, I'm not entirely sure how to calculate the exact odds for opposed rolls - art major, and all. Could someone a little more statistics savvy help with that?
(By the way, for sneaking past a group, if you normally allow Perception checks for every member of the group, just roll a separate Stealth check against every member's Perception target number to have the same effect.)
(Also, DM_Blake, it's possible to argue your point without sounding super condescending. Just saying.)
|
Ah, geez, math...
..snip..
Unfortunately, I'm not entirely sure how to calculate the exact odds for opposed rolls - art major, and all. Could someone a little more statistics savvy help with that?
Looks good so far. I'm not going to start on the maths at this time of night, I'm just going to point out that multiple guards should increase the DC of the check, probably along the lines of 2-4 +2 5-10 +4 >10 +6. Something to model the binomial chance of getting 1 or more success from N attempts.
Actually, hmm... Aid Another is DC 10.
So how's this: use the perception DC method for passive checks, but don't check against each and every guard in a group. Rather, take the highest DC and add 2 for each guard who doesn't have a net penalty to their Perception checks. (Normally you can't take 10 on Aid Another, this would be a special exception.)
So yeah, a level 1 rogue with +7 Stealth will have no problem sneaking around a single watchman with his piddly 12 DC. But if there are four of them in a group, that's up to 18... no more taking 10.
Furthermore, make it fair with an inversion. If the guards have reason to believe there's someone skulking in the area (no longer just passively keeping an eye out, but actively looking [and spending actions to do so]), they roll their Perception checks against a DC equal to 10 + the rogue's Stealth modifier. Rogue doesn't get to benefit from a high roll in this situation, but they do. And even that guard with a -1 can get lucky and hit DC 17.
This streamlines basic stealth around the unwitting, but also significantly reduces the number of "don't bother rolling" situations when someone's actively suspicious. Bottom line: it's not your action, you're a fixed DC.
Furthermore, make it fair with an inversion. If the guards have reason to believe there's someone skulking in the area (no longer just passively keeping an eye out, but actively looking [and spending actions to do so]), they roll their Perception checks against a DC equal to 10 + the rogue's Stealth modifier. Rogue doesn't get to benefit from a high roll in this situation, but they do. And even that guard with a -1 can get lucky and hit DC 17.
I would disagree with this.
I don't mind looking at passive DCs for people using their skills passively. I like the idea of passive Perception DCs for guards on duty - my only objection is that the math works out very badly for game balance.
But having the rogue use a passive 10 + Stealth Mod when he is actively trying to be stealthy is, in my mind, dead wrong:
Player: My rogue tries to sneak past the guards.
DM: What's your Stealth mod?
Player: You mean my stealth roll? *picks up a d20*
DM: No, put down your die. Just tell me your modifier, I'll roll for the guards to spot you.
Player: But I'm sneaking! They're just playing cards. Why do they get to roll but I don't?
DM: It's more fair this way...
I just don't see that going over very well with most players.
I say leave the d20s in the hands (or paws, claws, teeth, tentacles, or other manipulative appendages) of whoever or whatever is actively using a skill. If it's an opposed roll, both sides are active, so both sides roll. If one side is just passively using their skill to oppose an active skill someone else is using, then the passive DC makes sense - but in that case we still need a mathematical solution to make it interesting.
I think setting passive perception to 15+Perception Mod would solve most of the problem. The guards mentioned upthread (12 WIS, 1 rank in Perception) would have a DC of 17, and the level 1 rogue would have +7 on his roll (still an automatic success if the rogue takes 10, but maybe there's a level 2 or 3 sergeant there that might beat the take-10 so the rogue better be really sure taking 10 is enough or he's guaranteeing his own failure; maybe rolling would be better).
Problem is, when someone comes along who is a ranger, rogue, expert with Perception as a class skill, etc., suddently that guy might be +5 or +6, or more, even at level 1, and his passive Perception would be 20 or higher, making take-10 impossible and even making rolling a losing possibility (60-70% chance of failure).
So 15+mod seems too high, 10+mod is too low, weird stuff like 13+mod is just strange given all the other DCs progressions tend to follow multiples of 5.
It's awkward all around.
Me, I'll settle for the opposed rolls.
Something that should be done before this is argued more is to calculate the odds of success of rolling against the target number of 10 + Perception mod versus opposed Stealth vs. Perception.
Ah, geez, math...
Let's assume that the guard has a Wisdom of 12 (+1 mod), and one skill point in Perception (because seriously, who hires guards that are terrible at spotting things?), bringing his Perception DC to 12. For a level 1 NPC, this seems pretty reasonable.
Now let's assume a level 1 Rogue PC with a Dex of 16 (+3 mod), a skill point in Stealth (+1), which is also a class skill (+3), bringing his total modifier up to +7. Not bad!
Here's the math, all condescention aside:
The rogue beats the guard in the opposed roll exactly 70% of the time. The guard equals or beats the rogue 30% of the time. I know, I was surprised when I got this result too.
I don't see anything in the RAW that says what happens on a tie, so I assume both are successful: the rogue manages to sneak well enough to be invisible and silent, and the guard manages to perceive so well that he can see/hear the stealthy rogue. Since this means the rogue is spotted, I give the ties to the guard.
So there you have it.
With opposed rolls, the rogue succeeds more than 2x as often as the guard. Fully 70% of the time.
That seems good enough for me.
But as I have shown, that guard would have to be at least 6th level to have any chance to spot that same level 1 rogue taking-10. And that's assuming he never spent more than half of his skill ranks on other guard skills, like Intimidate, Profession, or Sense Motive. Or heaven forbit if he has a life that involves a hobby or two, like music or whittling or making furniture for his home. Even a guard dedicated to a lifetime career of guard-duty probably only puts ranks into Perception every other level, or maybe 2 levels out of 3, so now we're talking about 9th-12th level guards to spot a level 1 rogue taking-10.
Exceptional super-guards might be the, uh, exception. Sure, some occasional guard might have a rank in rogue (so a +3 class-skill bonus) or might have a really good wisdom, or might be so fanataical about Perception that he really does put half of his skill ranks into it, or maybe is just very bright so his INT bonus gives him extra skills. But those are all fairly exceptional guardsmen.
A rogue automatically gets +3 to Stealth because it's a class skill. A guard dos not get +3 to Perception because that is not a class skill, unless the king is hiring rogues to guard his vault...Wait. You're worried that a player character won't completely own a group of level 1 human warriors most of the time, when engaged in his specialty?
When, exactly, do skills get to be good, then?
If by good you mean that the rogue might succeed, or will probably succeed, then I grant you that a dedicated level 1 rogue should have a reasonable chance of success against a group of level 1 warriors.
But it should not be an automatic 100% chance.
That's a different discussion about disable device.You're making up extreme situations then pretending like they're proof that this house rule is broken. The only thing the rogue gets to do automatically is sneak around city streets.
Rich. I made up no extreme situation. I simply pointed out the rules.
Implement the take-10 rule and that rogue will get to do a lot more than sneak around city streets. Maybe the lock on the vault door will stop him, or maybe his take-20 on Disable Device will pick the lock. Who's to say?
But the mere fact that he can sneak right up to that lock and pick it in secrecy by taking 10 on his Stealth checks would be a bit much, in my mind, for a level 1 rogue to automatically accomplish.
Not true. You can't take 10 when threatened, such as in combat. The mere threat of being seen so that you might have to run away or maybe face a combat is not threatening enough. By RAW.You just said it was a threat. Now, where's the rule about what sort of threats constitute preventing taking 10?
You got me there. A bad choice of words. Pretend I said "risk" instead of "mere threat".
Being seen is not a threat. Getting hit in the head by this orc right here brandishing his axe is a threat. Especially since in game terms, that orc "threatens" the space I'm standing in. That's a threat.
Getting spotted is an inconvenience. Well, maybe if the spotter is a medusa. Or a beholder (extinct in Golarion, I presume).
How about rather than sneaking around with guaranteed freedom to go wherever you want to, we discuss ways to make Stealth actually fair and balanced as a game mechanic.
That's what I'd like to do with Stealth.So the rogue can steal something kept in a dark room, laying in plain sight with no lock on it, guarded by nothing but level 1 NPC classed mooks. 3/4 of the time.
Yeah. That's totally guaranteed freedom to do whatever you want.
You'll note that I said "go wherever they want". What they can or can't do when they get there is a separate issue.
But since they can disable many ordinary locks, and since some of them can see in the dark and the rest can bring a candle to many places where the rooms are unoccupied and the guards are outside, then yeah, I'd say it's pretty much freedom to go wherever they want and to usually do almost whatever they want when they get there.
At level 1.
Too much.
Or... you can just use tejon's method, which is clean, simple, and efficient.
(Although I personally would be hesitant to add +2 for each guard, I would go with +2 for the first and +1 for each after the first)
You could. I liked his idea about adding to the DC for extra observers. It even has a precedent in the tumble application of the Acrobatics skill where you add to the DC for additional defenders threatening the space you move through.
Still, that means 4 guards at every door that needs guarding, just to be sure to eliminate the automatic success of a level 1 rogue taking-10.
That's a lot of guards.
Still, it's a good compromise for anyone wanting to implement passive perception DCs.
In the stealth vs perception game taking 10 is done to reduce the odds that you'll roll low vs the other person's high. If the NPC's are automatically taking 10 as a rule, it would be wise to houserule that the PC can't (unless he takes skill mastery stealth of course)
Now that hardly seems fair.
Player: I'd like to take-10 on my Stealth roll.
DM: You can't.
Player: Why not? I did last week and you had no problem with it.
DM: Yeah, but those guards let you. These guards aren't letting you take-10.
Player: Letting me?
DM: Yeah. Since these guys are not actively looking for you, you are required to actively roll your Stealth check.
Player: Wait a minute. Lemme get this straight. If the guards are on their feet, peering into every shadow, alert for danger, I can take-10, but if they're sitting around, bored, waiting for their shift to end, not looking for me at all, they're preventing me from taking-10?
DM: Yep, that pretty much sums it up.
Player: ...
ROFL, no my friend, this wouldn't be a variable rule, it would be made a constant part of the game.
If you implemented it, stealth would become a no take ten skill. (Without the appropriate specific ability that allows it)
So you mean the guards could never actively try to stay alert and watch the shadows and pay close attention to their surroundings (e.g. actually try to perceive danger by making their own d20 rolls)?
Well, then in that case, it would be easier to convince the players.
DM: No, put down your die. Just tell me your modifier, I'll roll for the guards to spot you.
Player: But I'm sneaking! They're just playing cards. Why do they get to roll but I don't?
...but that's not what I said. :)
I say leave the d20s in the hands (or paws, claws, teeth, tentacles, or other manipulative appendages) of whoever or whatever is actively using a skill.
I thoroughly agree. I, too, am advocating this.
If it's an opposed roll, both sides are active, so both sides roll.
And that's how it works RAW, which is incidentally how I run my games. But the OP proposes turning passive Perception vs. Stealth into a flat DC. If this is done, I think it needs to be symmetrical: if someone takes an action to specifically look for you, the tables are turned and your Stealth modifier becomes a flat DC. This will only happen if they have some reason to believe you're there (e.g. you already flubbed a roll), or if they're particularly attentive guards who stare unflincing into the darkness for hours on end (i.e. your DM is a jackass). Sitting around playing cards, they're only passively perceptive (and perhaps have a circumstance penalty for distraction), making them only a 10+x DC to be beaten.
All of this applies inversely when it's the players taking watches, or actively searching for someone. And seriously, if you've ever participated in such skulduggery, this makes a lot of sense. It's easy to hide from someone who doesn't already know you're there; much, much harder if they're looking.
And if you really want to guard that door from rogues, you don't place four blind warrior guards. You place one half-elven expert with Skill Focus: Perception, who provides a DC 20 all by his lonesome, and then you give him a bell to ring. :)
But the OP proposes turning passive Perception vs. Stealth into a flat DC. If this is done, I think it needs to be symmetrical: if someone takes an action to specifically look for you, the tables are turned and your Stealth modifier becomes a flat DC. This will only happen if they have some reason to believe you're there (e.g. you already flubbed a roll), or if they're particularly attentive guards who stare unflincing into the darkness for hours on end (i.e. your DM is a jackass). Sitting around playing cards, they're only passively perceptive (and perhaps have a circumstance penalty for distraction), making them only a 10+x DC to be beaten.
Fair enough.
But don't forget, if you are not already using your Stealth, then you're just standing there, or walking there, or whatever. In plain sight. DC 0 for the guard to perceive you.
So since we're actually talking about actively perceiving someone who is stealthy, then that means the stealthy guy alredy rolled his stealth check (or took 10) on his own round - and that's the value the guards should be trying to spot, both via their automatic opposed perception (or passive DC) and via their active perception checks if they take move actions to look for the stealthy guy.
Or put it another way:
By RAW, if "if someone takes an action to specifically look for you", you don't get to reroll Stealth. Stealth requires an action on your part - the action you took on your turn when you rolled your Stealth check. So if you don't get to roll Stealth at this moment, why would you get a passive take-10 Stealth DC when you have already determined the DC on your last turn?
All of this applies inversely when it's the players taking watches, or actively searching for someone. And seriously, if you've ever participated in such skulduggery, this makes a lot of sense. It's easy to hide from someone who doesn't already know you're there; much, much harder if they're looking.
And if you really want to guard that door from rogues, you don't place four blind warrior guards. You place one half-elven expert with Skill Focus: Perception, who provides a DC 20 all by his lonesome, and then you give him a bell to ring. :)
Sure - but he costs 50x as much salary as all 4 guards put together. Kings are frugal, you know.
Besides, you have those soldiers for many reasons. Defense of your castle, suppressing rebellion, defending the royal personage while away, keeping the princess out of trouble, etc. You need lots of them, and you can't replace that small army of soldiers with an elf or two - that's just not enough firepower to keep the royal grounds safe.
But when your army of soldiers is not battling invading orcs, they get bored and idle. So you assign them duties, like guard duty, to keep them busy and to instill a sense of discipline. Two to a door should do it - any more and you have to hire extra men.
Just one way of looking at it.
Besides, even if all that junk I just said doesn't apply, how silly would the world be if anyone with a month of training could be virtually invsible, so invisible in fact, that the only way to keep them out of, well, everywhere, were to hire nearly superhuman (elves are quite superhuman, for the most part) guards with a great deal of specialized training, and then place these super-guards absolutely everywhere that is worth guarding just to spot these invisible scofflaws? Such a world wouldn't even have human guards at all.
I will say my intention of this mechanic is primarily to save time and make secret checks easier without modifying the chance for success dramatically. I can't imagine that taking 10 against a target number should be allowed.
Sure - but he costs 50x as much salary as all 4 guards put together. Kings are frugal, you know.
I'm pretty sure 1st level expert and 1st level warrior are both "trained hirelings," 3sp per day?
So, what level do you want PCs to be before they can sneak past two guard dogs and a handler? This sets the DC at 30.
What's your math on that? I get DC 22: 18 for one dog (base 10, dog's Perception +8), and +4 for two assists. A small PC with Stealth as a class skill, with the elite array and no feats spent, can take 10 for that at 2nd level.
Dogs get an extra untyped +8 on perception for things they can smell. It's not in their statblock; it's under the skill description.
Odd that this isn't itemized like the adjustments to Acrobatics and Track.
Well, I'll just go back to what I typed before I actually looked at the dog's statblock: I'd like to see you sneak past an awake, alert dog. Ever. :)
Well, I'll just go back to what I typed before I actually looked at the dog's statblock: I'd like to see you sneak past an awake, alert dog. Ever. :)Well, the point is that a ridiculous amount of CR-appropriate opposition is on par with dogs, only with Perception ranks equal to their HD. Unless you're small and blowing multiple feats and much money on magic, you can kill a pride of lions long before you can sneak past them, with your rules. Satyrs, salamanders, dogs, pretty much anything with scent, pretty much all animals, spirit naga for some reason, ropers...
I didn't have to look hard to find lots of things with Perception mods more than 10 better than the CR.
It is supposed to work that way.
Do you think a dog's nose works just like your nose? Maybe that's why you find it unreasonable that you can't sneak past a dog?
Take a look at this article.
Dogs can even find people buried under earth or snow.
A dog’s nose also contains infrared receptors that are sensitive to temperature.
If unfolded and spread out, the surface area of the dog’s nasal membranes would be about the size of a handkerchief. The surface area of a human’s nasal membranes comparison would be about the size of a postage stamp.
In fact, it’s estimated that dogs can identify smells somewhere between 1,000 to 10,000 times better than humans can.
Possibly 10,000x stronger than our noses. 10,000x.
And I know when I'm sitting in a chair in a conference room next to my co-workers, all of whom are clean and hygienically competent, but yet I can often smell them next to me (and that's not counting the smokers or the perfume wearers who I can smell from across the table).
By tht reasoning, my nose can often detect a human in an adjacent spae (game terminology), but sometimes I can't - I guess I miss my perception roll some of the time.
But a dog's nose, even at the lower 1,000x figure. Wow. 1,000x. He could smell each individual in the conference room, identfy them (if he's met them once before), and smell everyone in the office outside the conference room too. Even with distance penalties.
And here's an intersting bit I found from Duke University: "With its larger olfactory membranes, a dog's nose does amazing things. Researchers at Duke University found that a randomly selected fox terrier could after three weeks detect the scent of a fingerprint on a glass slide when compared to four clean slides. When the researchers placed the slides outside in the rain and dust, the dog was still able to pick out the slide with the fingerprint after 24 hours of weathering."
A fingerprint. Fingerprints don't have nearly as much body odor as a whole person. And that was after three weeks in the lab then 24 hours outside in rain and dust.
That's quite a nose.
So no, I don't wonder why it takes a an incredibly difficult Stealth DC to sneak past a dog.
My thoughts.
Setting a stealth DC for NPC's is o.k. As for PC's they should have the chance to roll a 3 or a 19 and that makes the game more interesting.
As for opposed roles, because it was brough up, if they total roll + modifiers is a tie, benefit goes to the person with the highest modifier, similar to someone having improved initiative winning a tie.
As to dogs. I was going to sneak up on mine while he was sleeping outside on the stairs next to the back door. As I approached the door, up cam his ear, then his face, then he hopped up and started clawing at the windows thinking I had a treat.
As to guards. Urban rangers are the inn thing. They get weekly bonuses if they take prefered enemy(human) or are half elves with skill focus(perception) Every king should have a few for his valuables.
I think that covers it.
|
My thoughts.
Setting a stealth DC for NPC's is o.k. As for PC's they should have the chance to roll a 3 or a 19 and that makes the game more interesting.
As for opposed roles, because it was brough up, if they total roll + modifiers is a tie, benefit goes to the person with the highest modifier, similar to someone having improved initiative winning a tie.
As to dogs. I was going to sneak up on mine while he was sleeping outside on the stairs next to the back door. As I approached the door, up cam his ear, then his face, then he hopped up and started clawing at the windows thinking I had a treat.
As to guards. Urban rangers are the inn thing. They get weekly bonuses if they take prefered enemy(human) or are half elves with skill focus(perception) Every king should have a few for his valuables.
I think that covers it.
or are half elf urban rangers with favored enemy (Human) and skill focus (Perception). every king needs a few for thier humoungous treasure horde. they also always have an aasimaar cleric of Sarenrae and a Grey elf wizard amongst thier consorts. the grey elf wizard to identify random magic items delivered to the king. as well as to identify other peoples items to keep the king safer. the Aasimmar cleric of Sarenrae to constantly detect poisons, and to serve as the king's "Therapist" should something be wrong in the palace.
Relying on guards to watch for thieves in a world where invisibility is cheap and common is going to cause problems no matter what you do.
Ehh... buying a single invisibility spell costs over half a year's pay for the average skilled laborer, and more than an average guard's entire equipment budget (which is expected to last quite a while). That's not cheap and common. PCs are just rich and exceptional.
And yes, I know that's the point you're arguing re: should be able to make the DC against mundane opponents at a reasonably low level. And I don't entirely disagree with you. Unfortunately, we're modeling a complex process against a linear probability spread, and at one end or the other something's likely to go wrong. So: should the margin of error favor the sneak, or the guard? I strongly prefer the latter.
Letting +2's stack to infinity does go too far, of course. It's probably reasonable to limit that to one assist, who must have 3+ Intelligence (dogs don't get to assist each other) and at least a +0 Perception modifier. Just call it a circumstance bonus, even. Harder for the sneak to predict where they're all looking at any given time.
And yeah, a 1st-level rogue can still easily glide past a mediocre watchman. A smart king instructs them to watch the vault door (for being opened) and each other (for being shivved), things which have no stealth at all.
The problem is that favoring the guard falls apart when you are matching skill against skill, not skill against incompetence.
All I can say to this is, it depends on the balance of realism to theatrics at your table. Stealth is very hard when anyone is paying attention. Realistic stealth generally involves at least two of distractions, darkness and distance. And it's worth noting that all of these exist as mechanical concepts in-game.