The game mechanics for a stealth sneak-up and backstab?..

[Crakkerjakk]

Originally Posted by Ulzgoroth

I'd probably opt for the every turn approach. The longer you're generating suspicious noises and shaking the bushes for, the better the chances of someone noticing you.

Oddity: the noise table in HT has a person 'stalking' only -1 harder to hear than a person walking. A quick contest of stealth is much more variable.

Here's the problem. Guy with Stealth of 16? Normal chance of success ~98%? In order for him to successfully cross 20 yards, he needs to make a Stealth roll 20 times. Not even looking into it being opposed by someone else (I can do the math but I'd have to pull out my stats book) in order for him to silently cross 20 yards he needs to make 20 rolls, at one roll per second, unless he's willing to take that -5 penalty for moving faster. That means that there's about a 2% chance of failing each turn. Over 20 turns, rolling every turn causes that to snowball to 31% chance of failure. The guy with a Stealth of 16 only has a 68.8% chance of crossing that 20 yards silently. Thats the effective skill of 11-12 if we were just using one roll.

I'll try to do the math on what an opposed roll would look like against guys with differing perceptions tomorrow.

Originally Posted by Mailanka

I think one roll is sufficient to "get your attack." That seems consistent with the way DF handles it. If you want multiple turns behind someone getting ready (evaluate) or making faces or whatever, I think that should require multiple stealth rolls, but a highly skilled sneak is the kind that can linger right behind someone for minutes on end without discovery. The more mundane people have to take their advantage where they can get it.

Can't you take a step with an evaluate? How close do you have to be in order to be able to take an evaluate? Why does it require more rolls to just step up behind them as opposed to stepping while taking an evaluate maneuver, aside from perceived game balance? I don't see why if a sentry was staring off straight ahead for a half hour someone couldn't simply stand behind him for half an hour without making a noise. I would still make my players actually move towards the sentry, turn by turn, and if they were competent sentries they would be walking a post, moving instead of simply standing in one place and looking in one direction. They might have to be fairly imaginative to actually close distance with a moving sentry (since they're probably both moving the same speed) but if you have just a guy sitting at a desk reading a playboy, it doesn't take any special skill to sneak up behind him without him hearing, or stay there for an extensive period of time without making a noise.

[Ulzgoroth]

Originally Posted by Crakkerjakk

Here's the problem. Guy with Stealth of 16? Normal chance of success ~98%? In order for him to successfully cross 20 yards, he needs to make a Stealth roll 20 times. Not even looking into it being opposed by someone else (I can do the math but I'd have to pull out my stats book) in order for him to silently cross 20 yards he needs to make 20 rolls, at one roll per second, unless he's willing to take that -5 penalty for moving faster. That means that there's about a 2% chance of failing each turn. Over 20 turns, rolling every turn causes that to snowball to 31% chance of failure. The guy with a Stealth of 16 only has a 68.8% chance of crossing that 20 yards silently. Thats the effective skill of 11-12 if we were just using one roll.

I'll try to do the math on what an opposed roll would look like against guys with differing perceptions tomorrow.

I think I see your point...and it becomes more dramatic if you're using a less overwhelming base skill. On the other hand, I have to say I'd think it a pretty impressive feat to approach inaudibly to within 2 yards of someone (after which you can step and attack safely). So I don't know that I have a problem with the odds you're getting.

It seems like distance ought to provide benefits to inaudible movement. Listen modifiers have a clear enough scale, but I don't know what the baseline distance is.

[vicky_molokh]

Originally Posted by Ulzgoroth

I think I see your point...and it becomes more dramatic if you're using a less overwhelming base skill. On the other hand, I have to say I'd think it a pretty impressive feat to approach inaudibly to within 2 yards of someone (after which you can step and attack safely). So I don't know that I have a problem with the odds you're getting.

It seems like distance ought to provide benefits to inaudible movement. Listen modifiers have a clear enough scale, but I don't know what the baseline distance is.

Come to think of it, the base Hearing roll to notice an incoming attack of a non-stealthy invisible opponent is penalized, before factoring in environmental noise such as a room full of machinery . . .

[chris1982]

We usually roll once per new action that could possible loud.

So e.g. once for opening the door, once for movement to the victim, and once for the attack (where failing the last roll might only mean that the victim gets a dodge at -5 or something).

[The Benj]

I find the statistics work differently because of the way I deal with the rolls:
It's only a Quick Contest if observer is alert. If they're just standing around (as most sentries are actually likely to be, really, especially regarding behind them) they don't get a roll at all unless you fail your Stealth roll.

Which means, taking Crakkerjack's example of 16 skill, that they only get to to roll if that 1.9% chance turns up... and then they may well fail anyway. Which means, against a Per 10 guy (who shouldn't be on watch anyway, but hey, your hiring practices may be more stringent) it ends up being a 0.95% chance, which raises your odds over 20 rolls to more like 83%, I think.

[entitivity]

So it sounds like everyone has different ways of doing this. I'm trying to figure out a consistent way to approach the problem as well, because it seems rather difficult to arbitrate.

[rosignol]

Originally Posted by Harald387

I'm in this boat. My general rule of thumb is to roll as if the character were trying to sneak right up beside the target - so ignore penalties for distance, etc. Apply the distance penalty *after* the roll; if it causes the roll to fail, you have a rough guideline as to how far away the character is when he's detected. If the roll succeeds, the character is free to sneak around near the target until he does something that gives him a large enough penalty that the target's detection roll succeeds.

This requires keeping track of the MoS, but I find this a lot less painful than rolling every second.

I really like this approach. It's one roll, so PCs don't feel like they're being snowballed, you ignore penalties, which simplifies things for the GM, and if they blow it, all you need to do is check the range table and you have a general idea of how far away they were when the twig cracked.

Is there a downside?

[Kallatari]

Myself, I use the following rules:

When sneaking up on someone who isn't really paying attention or expecting anything (i.e., the typical average person walking down the street as part of his everyday life), it's an uncontested roll against Stealth. If you fail the roll, the margin of failure will indicate at what range the other people will eventually notice you. On a success, you can get right up to them.

When sneaking up on someone who is paying reasonable attention for people moving about - such as a typical guard or sentry, or a worried pedestrian paying attention because he's in a bad part of town - then it's a single contest of Stealth. Your margin of victory will determine at what range such a person will notice you (i.e., compare when the penalties due to range to your opponent's sense rolls will be at such a point where you would no longer win, and that's where you're spotted). With a high enough victory, you can get right up to the guards unnoticed.

When sneaking up on someone who is paying a lot of attention - i.e., because he specifically believes someone is there, or because an alarm has been sounded, etc., then it becomes a contest each second. To me, the trick for this is that "paying a lot of attention" is defined as the person specifically taking a Concentrate maneuver as his maneuver for the turn, and he's concentrating on paying attention to his surroundings.

Note that the latter case is a rare situation, but happens. No matter how someone claims that he's paying attention while on guard, the sheer boredome and routiness of always watching the same place will cause distractions or your mind to wander a bit, so you can't always give 100% of your focus on your surroundings all the time. This is why most situations will be the middle one, where only a single contest is needed. On the other hand, the last one can come into play with very single-minded opponents (a robot or security system who can focus his complete attention on watch), or when a former screw-up has sounded the alarm, etc. Also, it gives a benefit in the middle of combat to any PC who decides as a one-off to take a Concentrate maneuver to "look around to see what's happening" as they can roll again and perhaps notice that group trying to sneak around to flank them.

Anyway, that's how I run it in my games.

[Plasmabunny]

Hmmmmm, seems like a good reason to give a robot sentry compartmentalized mind.