Stealth / Shadow changes - Discuss

[Halaster]

If you want to attempt to move away from your spot and remain hidden, use your stealth skills.  Regular walking (or running, flying, skipping, dancing, prancing, nancing) will not be stealthy.

[LoD]

While I'm sure a few sneaky types will be frustrated that they no longer have an infallible cloak of darkness, this should be a good change.

However, I've always desired a command that would allow you to "tail" someone from a room away rather than in the same room.  Perhaps coupling the watch command with a 'tail' command would provide this result.  My own suggestion would be to make this possible as long as the subject was walking.  If the target began running, sneaking, hiding or fleeing, it would disengage the tail.

Tail would begin while in the same room as the target, much like shadow, but would involve a one room delay on coded movement, perhaps hooked to trigger by the new watch command changes.  For example:

City Street [EWS]
The skinny, long toothed man is here.

>watch skinny
You begin watching the skinny, long toothed man.

>hide
You try to blend in with the crowds.

>tail skinny
You begin tailing the skinny, long toothed man.

The skinny, long toothed man walks east.

To the east, the skinny, long toothed man walks east.
You tail the skinny, long toothed man east.

City Street [EW]

To the east, the skinny, long toothed man walks south.
You tail the skinny, long toothed man east.

Bend in the Road [SW]
To the south, the skinny, long toothed man walks south.
You tail the skinny, long toothed man south.

City Road [NS]
To the south, the skinny, long toothed man runs south.
You fall behind and lose track of the skinny, long toothed man.

That way, you could add another dimension to shadowing or tailing someone through the city/desert.  This would hold both advantages and disadvantages:

:arrow: The person tailing wouldn't be able to immediately attack, steal from, engage, speak with or catch the target.

:arrow: The target wouldn't necessarily notice they were being tailed unless they were actively watching their surroundings, which I find to be more realistic than a "hide" check that statistically should fail somewhere around 1 out of 10 or 20 rooms, which isn't many.

:arrow: The person tailing will lose track of their target more easily than if shadowing within the same room (i.e. running, sneaking, hiding, fleeing).

There have been many times that I've wanted to follow someone from a distance, but while my character would likely be able to keep them in sight diagonally (especially in open spaces), the code operates with only north, south, east, west sightlines.

Since changes are being made left and right, this might be fun to consider.

-LoD

[jhunter]

This is a neat idea.

[Morgenes]

However, I've always desired a command that would allow you to "tail" someone from a room away rather than in the same room.  Perhaps coupling the watch command with a 'tail' command would provide this result.  My own suggestion would be to make this possible as long as the subject was walking.  If the target began running, sneaking, hiding or fleeing, it would disengage the tail.

I feel this is something that is best done as a trigger in your own client.  It would be fairly trivial to have it post a command everytime you see <person> move a direction.

I don't like things like this which basically makes the game play for you.  I'd rather you guys pay attention and use the watch command change that was just implemented to enable exactly this behavior.

[LoD]

I feel this is something that is best done as a trigger in your own client.  It would be fairly trivial to have it post a command everytime you see <person> move a direction.

I don't like things like this which basically makes the game play for you.  I'd rather you guys pay attention and use the watch command change that was just implemented to enable exactly this behavior.

I don't know that this would have the "game play for me" anymore than follow, shadow or the "watch" command already does.  Why would we have a follow command at all if we didn't want the game to play for us in certain situations?  It seems to me that the same playability issues for follow would apply directly to something like this variation.

It's sort of like saying "Characters should have to pay attention when they're eating red apples, but not when they're eating green apples."

I feel that it would be realistic for someone to want to follow another character at a distance where they couldn't engage in combat, overhear whispers, steal, speak to, etc...  The trade off for that variation of follow being that the target isn't as aware of their presence unless THEY are paying attention to their surroundings.

I see the word "realism" splattered all over the GDB when other changes are implemented -- why would this be any different?  And the suggestion to accomplish this feature as "a trigger in your own client' seems contradictory to the reasoning you give for not implementing this though Aramgeddon's code.

We don't want the game to "play for you" -- so set your client up to do it?

If there are some factors involved that make you feel this feature wouldn't be feasible or realistic, then I'd be interested in hearing them, but I don't know that this feature would make the "game play for me" anymore than the follow command already does.  In fact, it places some limitations on the relationship between the shadower and the target that aren't part of the game already.

-LoD

[The7DeadlyVenomz]

He does have a point, Morgenes.

[naatok]

From a naughty monkey perspective, I'd rather not see an automated process for 'tailing' someone.  Whether that means a particularly raunchy form of sodomy or slinking after someone on the streets or in the desert.
:twisted:

I agree with Morgenes that the changes made to the watch command allow for this already.  In fact...I've already used the code to do just that.

My only beef with it is that slinking about is slower movement than walking about...so you will eventually lose your target.....unless they're slinking too!

[X-D]

I've made a tail trigger myself...quite a few weeks ago actually.

[mansa]

I've made a tail trigger myself...quite a few weeks ago actually.

Can I see the Script?

[Morgenes]

I don't know that this would have the "game play for me" anymore than follow, shadow or the "watch" command already does.  Why would we have a follow command at all if we didn't want the game to play for us in certain situations?  It seems to me that the same playability issues for follow would apply directly to something like this variation.

I'll give you that this is similar to follow, not watch.  Watch doesn't make you do anything when you notice things.  You just notice them.  

My main fear is that this would make it too easy to follow someone, making it hard for them to 'ditch' you.

If we did something like this it would only be as I described the client trigger working.  You see in the distance x moves north, so you move north.  If you can't see the person for whatever reason (light, line of sight, whatever), you won't move.  Also if I did that I would give the person being tailed a chance to notice someone walking behind them, probably based on a generic watch check.  That way if you were doing it manually it'd be safer as there would be no watch check.

I feel that it would be realistic for someone to want to follow another character at a distance where they couldn't engage in combat, overhear whispers, steal, speak to, etc...  The trade off for that variation of follow being that the target isn't as aware of their presence unless THEY are paying attention to their surroundings.

I agree that this would be a realistic thing to want to do.  I'm saying that it's possible to do it currently in the game, by watching and moving yourself, or with a client.  

And the suggestion to accomplish this feature as "a trigger in your own client' seems contradictory to the reasoning you give for not implementing this though Aramgeddon's code.

Those were two separate responses, I was providing an alternative that would accomplish it without us having to implement any new code.  Not providing a reason why we shouldn't implement it.

I'm not saying this will never happen, just that I'd rather people try doing it manually for awhile.  I've got plenty of other coding projects that in my mind would take precedence over something like this.  So if another coder likes the idea and wants to run with it, I really don't have that huge of an issue with the idea.

[Morgenes]

My only beef with it is that slinking about is slower movement than walking about...so you will eventually lose your target.....unless they're slinking too!

You don't HAVE to slink about, you could just stroll down the street, dum de dum.  Who really looks behind them when they're walking through town?

[Intrepid]

My only beef with it is that slinking about is slower movement than walking about...so you will eventually lose your target.....unless they're slinking too!

You don't HAVE to slink about, you could just stroll down the street, dum de dum.  Who really looks behind them when they're walking through town?

Err...I am accidentally doing this a lot because I leave watch <direction> on without
realizing it.  That said, it would be nice if there were a "watch ahead" option that
would allow you to watch in the last direction you moved to simulate forward field
of vision.

I already idea'd this a week or two ago.

[spawnloser]

Tail?  Meh.  Use shadow.  I picture 'shadow' as tailing someone.  You can't let them get a block away or you'll lose the person being tailed...so catch up a bit and shadow them.  If you want to hang back further, that's your problem, in my opinion.

[Synthesis]

The major problem I can foresee with this is that the hide and sneak skills, at maxed levels, will function essentially like an invisibility spell.  A maxed assassin or burglar could spend some 95% of his time completely invisible to anything and everything, as long as he doesn't do anything to break hide.

As it was, if you were paranoid enough, you could spot people approaching you.  Now, good luck.  It significantly increases the danger and power of assassins and rangers (and other classes with both the sneak and hide skills), by significantly increasing their ability to move stealthily, from an OOC standpoint.

All of you folks cheering this change on are going to be miiiighty disappointed when you realize just how much sneakier this makes a good assassin.  Also, it's kind of a nerf on classes without the scan or listen skill, since the only way to detect a good stealthy type previously was to paranoiacally spam look in every direction.  Now, ha.  No dice.  You might as well wear a blindfold, sucka.

In fact, I'd say this makes assassins so much stealthier that you're going to see a dramatic increase in the number of them apped in the near future.  This essentially makes them unable to be detected -at all-, save for detection skills or skill failure.  This is a huge boon.  HUGE.  It may sound like a simple, realistic change...but it turns the balance of power in the city on its head, provided other things remain equal.

Personally, I suggest a stamina drain for moving while in stealthmode, since we're so hot on stamina drains for everything else.  If you're moving around snooping and pooping, you ought to be expending energy.  And from a realism standpoint, nobody should have the ability to be virtually invisible without having to resort to magick.  Or make the movement rate while in stealthmode painfully slow...if you're constantly looking for places to hide, you ought not to be clipping along at a fair pace.

Keep in mind here that I'm posting under the assumption that hide isn't broken during the process...if you successfully hide, then successfully sneak, you will be hidden throughout the entire process.  If this isn't the case, then disregard all of the above.

[TheOldGuard]

OldSchool dating back to '93 here. Came back to learn new code and return to playing again. Withheld from posting for a few months now until this thread. If there were a hall of fame for changes in the game, this would be in the top three. I've played over 80 shadowy characters easily....and this change enables a clear distinction that we "thieves" can be just that. It actually reinforces the fact "warriors" are "fighters" and darker classes, can be dark, without needing to be a Gladiator.  It has added a rich magnitude of roleplay, with the fact we can avoid aggro (Too lazy to address how this effects rangers in a positive way).

With my return, I have seen some unreal -fighter- power vs the darker classes. This stealth change was needed 10 years ago.

This change was genius-like. Kudos.

And regarding some silly thread above about 'rinth and pc interaction. Don't expect 10 nobles to be sitting around a bar talking about the latest wine....we are hidden and we like it that way. It's dark, it's smelly, and it's a struggle in all aspects of Armag IC and OOC.

Now, give me back my stink'n tailor on tradesman.

[Synthesis]

Oh. Ha.  Seems I jumped the gun a little with my previous post.

If you run a probability model with the new system, it turns out to be a serious nerf to shadowing, and sneak-hiding becomes an unsafe bet for sneaky types much sooner than you'd expect.

For example, if we assume that a person has a 90% chance to hide successfully, after sneak-hiding or shadowing someone for 10 rooms, there's only about a 35% chance he'll actually still be hidden at the end.

With a 95% chance, after 10 rooms you'll have only a 60% chance of being hidden at the end.

99% and you'll still have a 90% chance of being hidden.

Whew, look at the difference a few points makes.  Increased demand for footpads and sneaky armor, anyone?

Of course, a smart assassin playing the numbers would realize that the best strategy would be to begin sneak-hiding at the very last possible moment, giving them the fewest skill checks and hence the greatest probability of ultimate success.  Since you can't see around corners, 3 successful hide checks are all you need to accomplish to approach most locations completely undetected.

Another wacky side effect:  You can now (theoretically) spam walk around to lose or detect potential shadowers, since eventually their skill level is going to run up against cold, hard probability and they're going to bonk on a skill check.  So before you enter your apartment, take a spin around the bazaar and see if any fellows in hoods pop up. Heh.  You only need to move about 68 rooms to drop even a ninja grandmaster to about a 50% chance of failing a skill check.  (By my probably oversimplified analysis, anyway.)

And yeah, sneaky types, your best bet is to just type 'hide' again once you reach a destination, after you've been sneak-hiding.  Since the probability of you succeeding on every attempt in every room all the way from point A to point B is very low (unless you're a ninja), you probably won't be hiding when you show up.  Typing hide again gives you a much better chance of actually being hidden where you want to hang around.  In fact, it would be better just to type hide again every 5 rooms or so.  This will reset your chances, so to speak (unless you actually -succeeded- in all those 5 rooms, in which case attempting an extra hide actually -decreases- your chances, heh).

Okay. So, I suppose I'm about done with this for now.  It ought to be...interesting...at the very least.

[Halaster]

Of course, a smart assassin playing the numbers

And would of course miss the whole point of this roleplaying game.  I really hope you don't view the game and play this way.

[Cenghiz]

Halaster beat me to it... Please stop worrying about codewise advantages, Synthesis.

[Intrepid]

For example, if we assume that a person has a 90% chance to hide successfully, after sneak-hiding or shadowing someone for 10 rooms, there's only about a 35% chance he'll actually still be hidden at the end.

With a 95% chance, after 10 rooms you'll have only a 60% chance of being hidden at the end.

99% and you'll still have a 90% chance of being hidden.

Why am I expecting Leeroy Jenkins to leave you at the mouth of a cave and run into
battle because he was bored, wanted to have fun and die with fried chicken in
his hands?  This is a perfect example of why that character was such a great counterpoint
to his number-crunching friends.

[Ava]

Of course, a smart assassin playing the numbers

And would of course miss the whole point of this roleplaying game.  I really hope you don't view the game and play this way.

Hal, I know the place where I start "playing the numbers" is when I'm worried that the game will not match my expectation of reality.

It's very, very neat that the game now models the behavior "once I start trying to hide, I keep trying to stay hidden while I'm moving."  But if I'm trying to hide, I expect my success at each point to depend solely on ability to stay hidden in that location, not on a running multiplication of all previous attempts or something. Note that I don't know how the new code works...maybe it does match my RL expectations perfectly.

Suppose I "sneak;hide;shadow templar". I'm walking down the street, slipping from one knot of people to the other, turning around to look in a shop window when ~templar glances back (do we *have* shop windows?), ducking my head--or perhaps looking innocent and nondescript.

RL-wise, I would expect three possibilities as we move along the street.
(a) He neither notices that I'm there or that I'm following him.
(b) I'm poorly hidden at that location--if he happens to look around, he notices me there. However, I'm still attempting to hide; if we move on to a darker/more crowded area, I disappear into the crowd again.  (The "hide" command maintains my *intention* to hide until I perform an action that normally breaks it.)
(c) I really screw up and he sees that I'm following him.

If I can trust the code to accurately represent my abilities and intentions, I'm fine with that. Or maybe my expectations of what the commands represent are out of whack and need adjustment...that's okay, too, but something I'd rather get straight out-of-character first.

[Morgenes]

If the templar noticed you sneaking behind him, you would at that point be exposed.  If the templar reacts to you, alerting you that you are unhidden, then it's up to you to choose to high-tail it or whatever.  If they would rather not alert you that they know you're following them, they could just continue on as if they didn't notice you, and lure you into their evil nefarious trap!

You see?  If he knows that you know that he knows that you know...well you know.

[Synthesis]

Everyone plays the numbers, to some extent.  People play the numbers when they're backstabbing, sneaking, hiding, psi-ing, sparring, whatever.  Numbers aren't a problem.

I was simply publicly correcting my previous mistaken assertion that the new code functions like an invisibility spell.  It does not, due to the exponential growth of the probability of failing when performing x number of independent events.  (Unless you have a base 95% chance of hiding successfully, and you  have say, 2 pieces of gear that each add 2.5%, giving you a 100% probability of hiding successfully each time.  And of course, nobody -knows- any of the numbers, so it's all purely speculation.)

And I only posted the statistical analysis because it's so interestingly counter-intuitive.  You'd think that an 80% success rate is good odds, but 80% over successive attempts doesn't really do much to guarantee your ultimate success.

Personally, I like to think about the numbers because a) it's an interesting intellectual exercise  b) my time is somewhat valuable, so I like to "get the most" out of the playing time that I have and c) without skills, you can't get chicks, gosh.

If my short-term goal is to get better at something, then I'll think about how to do that most efficiently, while staying within the bounds of roleplay.  If my goal is to murder someone, I'm going to make damn sure (by considering the probabilities beforehand) that I can actually do it.  Sometimes it's fun to be fabulously inept, but if you're not relying on the militia to keep you safe from harm, and you're relying on your own abilities to keep you fed, being inept is a sure way to get dead very quickly.

Obviously, my goals are not always numbers-based.  Sometimes I just want to sit around, scratch my ass, shoot the breeze and soak up the gritty atmosphere.  Usually I combine the two goals, since they are not mutually exclusive.  If it were only the intellectual exercise of number-crunching and min-maxing I were interested in, I certainly wouldn't be playing Armageddon.

Now, the way to fix this exponential growth of the failure rate, if it is deemed to be a problem, would be to set a "sneak-hiding" status, whereby one continues to hide check in every room while the status is active, even if a previous hide attempt had failed.  In this case, the probability of you arriving at your destination hidden would be only your base probability of 1 successful hide attempt.  You could possibly be spotted intermittently throughout, but at the final location, if your skill and other factors add up to 80%, you have a true 80% chance of being hidden.

[Morgenes]

I still say you have no way of knowing if you've been exposed or not.  If you think you've been exposed, nothing is stopping you from typing 'hide' again.

[Synthesis]

It's a binomial random variable problem, like flipping a coin.

And yes, you're right.  You don't know whether you're hidden or not.  The -best- bet, across the board, is to type 'hide' again once your probability of not being hidden is greater than your probability of being hidden.  That is, when p^n < 0.5.

On the slimmer chance that you are actually successfully hidden, doing a forced hide will actually reduce your probability of successfully being hidden after all is said and done.

However, since the chance of you -not- being hidden is greater, and the forced hide increases your probability of being hidden in the end in this case, it becomes a 'safe' bet.

(This is like the door-switching strategy in the "Let's Make a Deal" gameshow problem, where the switching strategy counterintuitively -doubles- your probability of winning the grand prize.)

Obviously, this all holds true -only- if you have no knowledge about your hide status.  If you -know- that you are or aren't hidden, then you know that your next hide check will actually have p probability of being successful.

The major catch to all of this is that nobody knows that their probability is,  until they reach their class's skill max.  You can approximate probabilities by keeping data on every hide you perform over time (you'd need an observer to spot you, though) and get reasonably close, but it becomes certain only when you are no longer improving your skill.

If anyone is still skeptical, google "binomial random variables," and figure it out for yourself.  I'm fairly certain I've got it right, though.

[SpyGuy]

Can't it just be made the when you shadow someone it retests hide/sneak every room like it would it you were manually trailing them?  That way you still have a chance to be noticed, but if the person you're stalking is just spam walking or ignoring their surroundings you have a chance to become hidden again.  If you are noticed said person can react appropriately, using either scan, the watch command or other nefarious means to reveal your PC when they want to.

That said, for all the number crunching aspect I think Synthesis brought up a very good point.  Thief types using shadow will need to be much more careful now and rehide if code works as he thinks it does.  Knowledge of the code shouldn't be an obsession, no one should try to min-max everything.  But basic principles of how things work, especially with recently changed stuff, are important to understand.