« You Had to Be There | Main | The Zergling Rush (of your dreams) »

May 13, 2005



Any game mechanic specifically designed and tailored to encourage grief play is not healthy to a game community. Stealth certainly falls under that category.

No one likes feeling helpless or defenceless. For example, if in DAoC I get jumped in the frontier by a wandering Sorceror, I get frustrated, but mostly at myself for not paying enough attention. I can have at least the illusion that I had a "fair" chance to win the fight. If I get jumped in the frontier by a nightshade, it's a whole 'nother story. It doesn't matter how much attention I was paying. I can't see them until they hit me. Same thing with rogues in WoW. The feeling of powerlessness is the root of the frustration folks feel about stealth classes. Things like stunlock just add insult to injury, but you're right, that's not the main thing. Stealth is.


(I offer no solutions, just more wood for the fire...)

One of the biggest problems with stealth in MMOs is that it is fundamentally unrealistic in the way that it is (almost) always implented.

I'm standing in front of you. In the desert. At high noon. I turn invisible. Neener neener, you can't see me. Riii..iiigght.

It just doesn't make any damn sense. Sure it's fun, but it makes no sense. What exactly are we hiding under? Air molecules? It's not like rogues are magically invisible - if you walk right up to them, you can see them (assuming similiar character levels). A rogue standing on a sand dune in the sun just *shouldn't* be able to hide.

Part of the problem with stealth as a playstyle is that it's fundamentally at odds with the nature of MMO games. MMO games assume that every single person you're dealing with is some sort of adventuring hero. Stealth assumes that everyone is just a normal guy. Take a game like Thief(1,2 or 3): you're dealing with an unsuspicious public: mostly average Joes who aren't *really* expecting or looking for any sort of trouble or adventure. They're just hanging out doing their thing/job or whatever. Even security guards aren't actually *expecting* trouble. It is against these kinds of people that the use of stealth actually makes sense, and the kind of stealth available to you makes sense. You hide in shadows, you try to not make noise, etc. You screw up any one of those, and the guard (who is generally a better fighter than you - heh, except in real life, where they have pizza and beer gut) will kick your ass.

For stealth to be realistically playable and not-over/under powered it would have to implemented in the same manner as it has been in the many stealth focused games out there. The problem is that you can't *really* do that in an MMO, lighting/noise/etc are far too real time sensitive to allow for things like high ping times and lag spikes. Furthermore, you're not dealing with a computer generated and unsuspicious public, you're dealing with hundred/1000s of highly suspicious adventurer types who are on the lookout for enemy activity, so to treat the relationship between the stealth its victim the same for both adventurers and the common people is silly.

Unless of course, it's magic, in which case, it simply becomes a balance issue which is nearly impossible to fix.


Random thought: some future mmo could make stealth something like:

(1) remove the bright "name tag" over someone's head
(2) make them 50% translucent
(3) make them targetable only by actually clicking on their model (can't auto-target) in the world

This way, if you're standing in the desert, or in a really open field, you're very unlikely going to get smacked. Someone is actually forced to stay in the shadows, and move when you're not really paying attention, to lay the backstab on you. You start to even get into interesting gameplay elements like -- ok I'm gonna go stealth in this forest area, time to put on my green-textured gear.


I think it's right to say that some of the possible implementations of stealth fail for the same reason that some possible implementations of combat fail. You can't have combat that's as twitch-based as a good first-person shooter; you can't have stealth that's as twitch-based as a good first-person sneaker.

I like the idea of making stealth more immersively environmental, but at least some of that would be defeated pretty easily just by messing around with your brightness settings on your monitor, the way most darkness effects in most games are neutralized by many players. But I could imagine a more photorealistic MMOG game that might make that possible. Though in that case it would be less a game-mechanical skill, perhaps, and more a case of player cunning. This is already possible in certain ways: there are World of Warcraft rogues that I know of who carry cheap caster clothing and wear it on occasion hoping that someone won't mouseover and see what they really are, and I can think of other similar examples from past MMOGs of players using deception that was outside the game mechanics per se.


stealth Is magic, in a fantasy mmo, and in legend -- ninjas turning into blocks of wood, magicians vanashing in a puff of smoke etc.

the complaint is that stealth is overpowered. however, in WoW, there are many active stealth counters -- hunter's flare, hunter's track: hidden, hunter's mark, human perception... it's not that stealth is the problem, it's what comes after stealth -- stun-lock.

stealth just allows rogues to begin a stun-lock combo that will generally kill an opponent without reprisal. i think what people want is a way to preempt stealth -- a passive counter that say, prevents the next stun attempt from occuring, or a tweaked game mechanic allowing better "save vs. stuns".

as for bypassing content, it's only that currently in wow, invis is more reliable than see invis. in SOE games, this is the opposite, and you don't see the cry for nerf.

the issue is just purely one of balance. any rock that has no paper or scissors to counter it will obviously be unbalanced.


We had invisibility in MUD1 and MUD2. It was no problem because:

1) Anyone with enough magic could go invisible. None of this class-based "balancing" mularky...
2) Anyone with enough magic (which wasn't a great deal) could cast a visibility spell to make nearby invisible people visible.
3) It cost actual stamina points to go invisible. If this is made worthless by healers, make it that invisibility only works while you remain below some threshhold of stamina points.

If it's a gameplay-dependent decision as to whether stealth is going to help or hinder, that makes it more interesting. If it's just simple permission for one group of players to tax random other players, it's not only uninteresting, it's undesirable.



These issues are intrinsic and unavoidable if you're going to have *PvP* in the equation. It's fun to sneak up on a monster who can't see you and backstab them with little risk. It's not fun for the monster, but fortunately that monster isn't a customer paying to play your game. But when it comes to PvP, he is, and that's a problem.

WWII Online is one of the few MMOGs that has "real" stealth; that is, there's no artificial mechanism that translucents a model or makes something untargetable; just the nature interaction of graphics, models, terrain, and the human eye. And getting killed -- often repeatedly -- by someone you "can't" see is a common complaint even among these hardcore realistic simmers. The more casual MMOG RPG player is going to have even less tolerance for this sort of thing, especially when the mechanisms for creating the stealth seem artificial and unfair. (Like seeing where the guy is but being unable to attack them because of "stealth".)



The problem I see with stealth lies in its implementation- particularly the"toggle on / toggle off" attitude. While the "scout" subclass is an intriguing solution, it still places all non-scouts at a disadvantage. I'd prefer a solution that is more open to all players

Let's face it, stealth in real life is not a toggle-power. It's about using terrain, lighting, atmospheric effects, and the activity of a quarry (masking sounds) to your advantage. There are some environments where stealth may seem impossible to achieve, but if that foe in the desert at high noon is busy cracking a pickaxe on a rock, you could probably gallop a horse right up next to him.

Stealth is a skill- so is detection. Anyone can look around, but identifying a cracked branch, determining range, and pinpointing direction takes practice. Someone actively scanning can often see things that a cursory look would miss. Even kids that play tag learn that they can position themselves in places that make them better able to respond to pursuers.

It's more than skill though. A kid running with a backpack flopping about is easier to hear than another. Similarly, chainmail will make noise whereas buckle-free padded will not. A lantern can increase detection, and the properly colored clothes can reduce it. Heck, a pet can alert you to things before they reach your own attention.

There are also different degrees of detection. You might hear a foe and determine the area he's in, but still not be able to see him. You may be able to see movement, but not make out the form enough to target it.

I recall that when I played "ghost recon," there was a "compass" UI that would give color coded signals to tell the user if *something* (maybe footsteps...) was detected in one of 4 directions. It was color coded, based on the certainty and proximity, IIRC, so yellow was a faint bump to a red "major threat."

I could see a similar color-coded system working with a dynamic balance of stealth and detection attributes, adjusted by gear, distance, terrain, lighting, and atmospheric effects. All classes have base values, but some are more adept at one or the other.

I haven't had the time to sample more than the more "mainsteam" games, but has anyone taken such a route?


I actually tried to suggest the outline of a stealth system for Star Wars Galaxies some months back. (The relevant Stealth System message is on the official SWG forum if you have access to it.)

The idea had two parts:

1) The game world is rich in sensory information. Examples of the kinds of sensory modes over which signals are emitted and can be detected are: EM spectrum (sound, visual, heat, radio), olofactory, mass/density/weight, and motion.

2) Every game object (but especially PCs, NPCs, and creature mobs) has baseline "visibility" and "awareness" scores for every sensory mode, and these base scores can be modified by various factors (innate abilities, environment, gear, and actions).

In such a system, stealth becomes a matter of comparing your modified visibility score against the modified awareness score of your target over each of the available sensory modes, and transmitting the results of such comparisons in an appropriate way. "A guard dog stirs, as though it smells something on the wind."

The likely problem with such a gung-ho approach is that it leads to a kind of stealth "arms race." Because there'd be some people who would spend all their time and money minimizing their visibility scores, anyone who wanted to avoid being owned by these people would have to spend time and money keeping their awareness scores constantly maxed... leaving little time for anything else.

Probably not much fun.

One potential fix would be a third rule: awareness-raising modifiers are always cheaper and easier to use than visibility-lowering modifiers. If you set the balance properly, players who just want to mind their own business can detect all but the most single-minded Bad Guys, while the more paranoid players can always detect Bad Guys by taking a more active approach to increasing their awareness scores. (This is for PvP; obviously PvE stealth results would be determined by the average awareness scores you set for your mobs.)

So here's my question: If you set up such a stealth system in which the potential target always has a reasonable but not complete edge over the stealthy player, does that make stealth both: a) sufficiently safe for those who don't want to participate, and b) sufficiently fun for those who do?



Gameplay and game-balancing are definitely relevant matters, and before just taking stealth out I would think about many times. We all agree that a game where there is no way (indeed) to counter stealth is poorly designed. And how to do that is of course an issue somewhat related to the game we want to focus on.

But as MMOG are inherently social games I would also take into account that stealth risks for a given character could be avoided by sticking around with other peers (as it happens in Timothy's telling about lvl 60 characters). Personally I wouldn't find it undesirable from design viewpoint nor against fun.
The tension between roleplaying and levellers is well-known: probably design solutions that help in keeping it balanced are better inspired, I'd say.
(Btw, thanks for raising up this interesting issue with your post)


In World of Warcraft and several other MMOGs, arguably the trade-off a stealth class makes is to be incredibly vulnerable if they lose the initiative and are attacked first. That's reasonably true in WoW: if I don't take a target down fast, I'm dead. If I miss a single stun, I'm probably dead against an equal-level player. But I still get to choose when to fight and when not to fight unless I'm being unwary or in a hurry. Meaning I can always favor asymmetrical odds--picking on vulnerable players, lower-level players and so on.

Now I would observe that this is somewhat related to real-life military conflict: victory usually goes to the side with better intelligence that picks a favorable battle ground and favorable conditions of engagement. That analogy alone doesn't rescue stealth from its problems, because it hardly means that stealth is *fun*.

I will say that the use of stealth is fun not just because you can win many battles on your own terms. I also find that it is a great choice for an explorer-type player, and for the player who enjoys solo play within a MMO (another long-standing topic). It has a "fun" to it that isn't entirely about victimizing others.


Stealth is handled in Anarchy Online similar to the way Flatfingers describes:
* 'Perception' is a skill that players can develop directly defeats 'Stealth'.
* perception is cheap to raise for most (or all?) classes.
* perception doesn't have much other purpose (it's also a prerequisite for equipping targeting scopes for critical-hit-dependent classes, but only in small amounts).
* once a hidden character engages with an opponent, they're revealed for all to see.

PvP is optional in AO; players who intend to PvP put points into Perception. Players who intend to only PvE, or only casually PvP, don't bother.

They also use something like Richard Bartle's caveat: all the easily available devices for increasing your stealth drain your health continually while you wear them.

There's only one stealth-centric class; the general opinion seems to be that agents need to max their stealth skill *and* get the best stealth-boosting gear they can *and* keep all the stealth-boosting spells running to be able to sneak around opponents of about their level.

(From memory; I haven't played in a month and a half, and never played the stealth-centric class.)

Of course, agents complained that this made their play impossible, so the need to be hidden was removed from the 'Aimed Shot' skill they use to get critical hits. So high-level PvP combat now has a significant element of *all* classes, not just agents, mastering Aimed Shot and sniping at each other.


I'm curious what Terra Novians think of Mage Invisibility in WoW -- it was in Beta, then removed, and now recently there have been overtures from the developers that it might be returning.

There's the hard counter that warlocks have the 10 minute ally-castable buff 'detect invisible' as well as alchemist's detect invisibility potions... but would that be enough?

I'm pretty involved in WoW's Mage community and several mages fear that re-adding Invisibility to mages would make us into a 'Ganking' class akin to rogues. They also fear that Invisibility would preclude other changes to mages.


On another topic, I'd like to see a Terra Nova piece/analysis on Mudflation, especially as it pertains in the World of Warcraft (witness 80 dps 2h maces off of Ragnaros, compared to 52 dps swords from the level 60 non-raid instances). Casters don't scale anywhere near the way melees do in WoW.


stealth vs invisiblity

Idealy (IMO), stealth and invisibility should be completely independant of each other, and each should offer the appropriate rock/scissors/paper checks and balances.

In the EQ world, invisiblity (except possibly in the cases of rogues) was always a variable factor. Meaning the duration was somewhat random and various creatures through out the world could occasionaly see through invis.

As EQ is primary PVE as opposed to PVP, the need to be invis/see invis vs other players is mostly a minor priority while typically a very nice convenience with regards to NPC, travelling, etc.

In EQ-PVP, see invis items are relatively trivial to obtain by the levels you would generally be pvping.

In EQ there is no stealth per se.

A better system perhaps would be implimented where invisiblity/see invis worked similar to EQ, but there is separate and different 'stealth' system.

A stealth system could perhaps be implimented that the players have available both a stealth skill level, and a corrosponding 'perception' level.

Stealth could last indefinately, or be stamina (or health etc) based so that while it could be used for longer, more consistent time frames than invisibility, there is some type of systematic drain of which would eventually run out. (ie you could not just go invis in a corner and come back hours later still invis)

Other restrictions that should be implemented would be that stealth only works in conjunction with world objects/geometry. Like in a previous post, a stealth character should not just be able to stand invisible in the middle of a desert in the sun. They must be within certain perimeters of rocks or buildings or other 'concealable' objects.

If the game in question provides the means to visually fade in/out a stealth character, that would be even better.. ala predator style..

Perception checks vs stealth checks (in conjunction with other factors) would scale the 'predator effect' in a sliding scale manner.

I think perhaps the best 'stealth' implementation I've played was planetside, splinter cell and similar games.

I've never played any of the 'thief' games, nor WOW, so I can't speak to their implementations.


As usually the primary issue people have with stealth is the interation with PvP systems. You might as well be claiming that ranged attack, or magic are significantly different systems that raise issues of their own, as those would be the next two repositories of complaint if stealth weren't in the game, and probably are the primary repositories of complaint for rogue.

They are good ganking machines for WoW because of the tendancy for people to solo, and because the power gap for an "honorable" kill is far too large. If you look at stats of rogues in large scale PvP raids, they suffer considerably. In a group oriented enviornment a rogue can be expected to die either right before or right after they manage to kill someone else.

If you want to be helpless try being at low hitpoitns when a 60 warlock curses you.. Or beings too far away from a hunter, or try to take down a class that has an instant heal effectively doubling their hit points.

There are plenty of papers for the rogues rock. You are slow when stealthed, vulnerable when not. Around 1/3 of your abilities can only be applied when stealthed. Several of those can only be applied before combat starts. before You have no built in healing, low armor, moderate health.

Games without much variation between classes aren't that interesting. One of the off-shoots of EQ2's tree system is classes that are supposed to be different feel substanially the same through a significant portion of the game.

One of the biggest reasons I've stuck through WoW in spite of the lag issues is the support for stealth play. Its very easy to design it so sealth is meaningless. In most respects its not that different from other classes as far as analyzing the DPS benefit. You have a large upfront delivery of damage, then a signfificant reduction later very similar to what you might see in a tradional nuke class. You get a better sustained DPS over a long battle though due to the fast regeneration of action points.

The most significant benefit of stealth with respect to ganking lies in being able to sometimes wait for prey instead of having to go look for it. A 60 warlock can drop a 40 or 50 just as easily if they happen across someone.


Slight addendum..

Stealth implementations which display the player in a visually obscured mode instead of completely removing the character are fundamentally flawed for PvP due to the prevalance of client side hacks.

(1) remove the bright "name tag" over someone's head

Done in WoW after for years we begged to have this in DAoC.

And it really makes the difference.


Curious... Could Magic: The Gathering's landwalking (e.g. forestwalking) ability be considered a form of stealth?

It's an interesting implementation of it since it was environment based (not normally toggled or tapped activated), and generally required your opponent to be as vulnerable as the enactor. Granted the ability was purely a NPC/hirling thing, but I thought I should still inquire.


DAoC players implemented a nice anti-stealth tool: Radar.

Seems to me in the long run that any system-generated invisibility is going to be susceptible to a radar hack. The info that someone is there has to be sent to the client, doesn't it? Or does it?

Environmental stealth is cooler. Well, unless you get things like gnomes hiding inside their pets, kobolds hiding inside trolls, etc.


Also as a perpetual rogue victim, I'd like to see some locational restrictions. I'd like to know that if I walk in daylight in open spaces, I am safe, but if I wander into murky forests or down dark alleys I am not. And this puts the additional constraint on the rogue that he has to take up a position and wait for a victim. Or use the terrain to get closer. Thief has those aspects and is a superior stealth system imho.

Another thing: why is it always about killing? What happened to theft? Imagine: From stealth, all you can do are incapacitation attacks, which then allow you to steal. It would make rogues into agents of redistribution.


Actual theft opens up a very different can of worms, though. As it is, one of the reasons people object to stealth is that it allows an end run around labor-time to reward ratios that other classes are shackled to. If a thief class can also take from other character's labor time, the incentive to be a thief shoots through the roof.

Unless there is some counterbalancing disincentive. Even the worst death penalty ever implemented to date in a MMOG wouldn't really suffice--somehow you'd have to have the thief suffer some major hit to their own labor time if caught. Say, for example, if a thief gets caught in the act, they lose up to 50% of their own accumulated wealth, and all of that goes to the character who caught and killed them? I can see a zillion ways to beat that system, though--using mule characters to "launder" wealth, or having a collaborator "catch" you so as to launder your gains that way.

Environmental stealth also requires the implementation of extremely good collision detection and very good environmental design as well--it would be too easy otherwise to use graphical quirks to hide, or to hide inside of other characters and objects, as Edward notes. You'd actually have to design spaces and places which helped people to hide.


I think world of warcraft's specific problem is that rogues are too effective in combat. This is a by-product of the level bound system, and can't really be fixed.

If you want to have really cool classes, you have to get away from all characters having to be equal. If you want your rogues to have superior stealth, they need to be put in the dirt when they go toe-to-toe with a warrior. Rogues should have their stealth and their backstab (and their stealing), but they should not be hunting on the same level as a fighting class.


Early UO had "environmental stealth" - no allnames, no circle of transparency. One could use the standard dye tub to dye clothes certain colors, wear certain clothes, and stand under trees/bushes and be essentially unseeable without using hide/invis. One had to be careful when moving, but it was possible to move when in the proximity of others and remain unnoticed. It used 'person at the keyboard' skill, not character skill, and it was fun.

Even if allnames/etc. was removed, as others mention, hacks would make this impossible today, as one's presence would be sent to the clients of nearby players.


In Shadowbane, one of the key distinctions of our Stealth implementation is that we've designed key classes to use stealth (thieves and assassins) to be best suited for taking out single targets. Once you stealth and backstab, you can usually take down a mage or a rogue before you need to hide and rest up again. This makes stealthers ideal for taking out the most problematic target on the battlefield (usually a healer, since in SB they can summon as well as heal).

In WoW, though, their admittedly innovative energy system for rogues allows a rogue to fight much longer before resting. A 60 rogue can easily take out a full party of 50s or lower. This subtle balance decision goes a long way towards changing the power potential of a stealth class.


As another point of reference it's interesting that a competitive, nearly PvP-based, game like Guild Wars avoids the issue of stealth altogether.


Ted>Seems to me in the long run that any system-generated invisibility is going to be susceptible to a radar hack. The info that someone is there has to be sent to the client, doesn't it? Or does it?

It doesn't have to be sent to the client of anyone not entitled to see it.



A few points:

In WoW there is a perception attribute. Humans get a bonus to it. It may not be listed on the character sheet but there must be some in-game formula for it.

Hunters have a track hidden ability. Not sure to what degree that works against player rogues however. Who knows, maybe "scout" will be a hero class available to hunters when Blizz gets around to what are effectively "sub-classes" to coin a D&D term.

Rogues have long been reviled. The cries for "nerf rogues" were second only to those of "nerf pallies." Rogue's are the only class in the game can solo *a mob* of orange threat level creatures. That right there screams game balance issues to me.

Sooner or later Blizzard will get around to some approach to fixing the Rogue -- one extreme would be to drop their DPS. Most likely they'll go the "stun lock" direction as a quick fix. Neither will be the right fix in and of itself.

You should see the effect that HK madness has had on PVE servers... they've all but turned into PVP servers in certain locales: Southshore and The Crossroads in particular. "Contribution points" are soon to be redubbed as "Honor Points", perhaps one approach, not one I would recommend since it seems to target only the rogues as a class which may be unfair, would be to also have "Dishonor Points" and maybe attacking from stealth mode could incur some of them.

I'm not a fan of forcing a layer of tactical ops onto the game -- making rogues have to wear different shaded camo gear or genuinely hide in shadows etc. all immediately detract from the fantasy element.

The fundamental problem is what's motivating players to engage in Honor Killing in the first place -- more better betterest bestest gear! It's literally an arms race. Keeping up with the Joneses. When I first started playing WoW I thought blizzard had gotten away from that mentality (which really damaged my enjoyment of Diablo 2) but I was woefully mistaken. Clearly there's a business reason to encourage this -- it keeps high-level players interested and still subscribing -- but constantly upgrading inventory has just never felt accurate to me. The source material, things like LOTR, Conan, Norse Myth, etc., just never features this sort of thing. It's one thing that appeals to me, at least conceptually, about Jade Empire.


>It doesn't have to be sent to the client of
>anyone not entitled to see it.

Great, now when I unstealth and backstab you, I also lag you for a second or two while you get transmitted ALL of my data (the stuff that would usually get burst to you when I first "pop" into existance some ways away) and load level 1 textures!

I can see fields of rapidly stealthing and unstealthing rogues lagging an entire town.



As Bruce notes, the Hunter is the natural counter to the Rogue in WoW (similar to the Scout/Thief analogy drawn to SB above). The Human Perception ability isn't all that great, as it only adds to the base Stealth Detection rating rather than guarantees seeing stealth.

Also, keep in mind WoW uses both Invisibility and Stealth independently.

I don't have a problem with Stealth in general. It adds depth to a game. However, to make it a fun part of the game, every class/template should have a built-in way to counter it, either as a Skill, a Talent, or as a repeatable quest. These abilities could need reagents, have long reuse timers, or require specific locations (ie, the Scout analog must be lying with his ear to the ground for X seconds or some such). They should be part of those class/templates from the moment they can enter areas where stealth can be effectively used against them.

It's a careful balancing act, but an effective part of multiplayer gaming in many genres,


Timothy Burke >One of the questions I'm eager to pose here is whether the issues that stealth implementations seem to give rise to are intrinsic and unavoidable or the consequence of poorly thought-out designs. <

I think it comes down to poorly thought out design. In particular, a generic problem with most current MMOGs. Designers create a world with an “expected” maximum rate of loot and experience gain, they don’t actually code it into the world. They then react when these limits are breached by nerfing the offending skill, rather than simply coding a limit.

Rogue skills are a good case in point. The Designer expects the average rogue at level x to pickpocket about y gold per hour. Then along comes Exploiter A, and finds a spot where he can get 20y per hour. The Designer then nerfs pickpocketing to a 20th of what is was, making it pretty much useless to the average player. Cybernetics suggests you can’t set something to a fixed rate, and use to control the approach to a target. If you have a target rate for pickpocketing, then the loot take needs to drop as a player approaches the target. This makes sense in a world fiction, a skilled pickpocket could take out a dozen targets in a row. But keep at it, and tiredness and inattention creep in. Unlike in the MMOG world, where your skill improves regardless. Its this lack of negative feedback for overuse that causes so many problems with powerful skills in MMOGs. Stealth is a prime example. A bit of exhaustion would go a long way to solving stealth problems. And make playing a lot more realistic, and to my mind a lot more fun too.


You said, "The real issue is stealth itself." right after you (IMHO) made perfectly clear that the real issue was the very poor honour system. But even when it *is* stealth itself, I think if people have problems with it, it's poorly done -- as opposed to it shouldn't be there at all in the first place because you absolutely cannot have stealth in MMORPGs.


I can't really address the balance issues that stealth brings to a particular game, as those will tend be system specific, nor can I address the lag issue that Bruce brings up, since that is a technical issue that needs a technical response (and I recognise that the response may be NO).

However, I can speak from extensive personal experience as a player in Shadowbane that stealth (and theft) can bring an important dimension to a game, particularly an MMO. Specifically, stealth in Shadowbane created a need for security and recon elements. In fact, major battles were marked by large pre-battle actions that I generalize under the title The Scout Battle. Here you attempted to locate the enemy's main body and divine their intentions, while preventing the same for them. It was a large area, fast moving, extended event which inspired a level and duration of excitement and tension I have not otherwise encountered in games. This is an element worth recreating.

It is worth noting that Shadowbane stealthers, particularly scouts, were hard to catch, but easy to drive off, and were generally deficient, to the point of imbalance, in offensive ability. They had stealth, and little else. There were a few offensively effective stealthers (eg channelers), but they had an inferior stealth which could be very easily defeated, and they moved relatively slowly. Beyond that, generalized detection (tracking) of stealthed individuals was very common, even though localized detection (revealing) was not. Hence, it was possible to know a stealther was nearby, allowing you to take some precautions even though you might not be able to force the stealther to unmask. This created a tension on both sides which, in my experience, was an engaging element of play.

So, in a PvP system with multi-level checks and balances, (beyond mere RPS), where stealth and offensive power are inversely related, it is possible for stealth to produce strategic and tactical complexity potentially leading to exciting gameplay.

In PvP and PvE environments, I would also point out that stealth will be a desired ability for Explorer types, although this discussion seems mainly focused on Killers and griefers.


The core thought was:

Timothy Burke>Nevertheless, I think this is one design challenge where even a superior implementation cannot reconcile fundamentally antagonistic forms of play.

But I think stealth and asymmetrical modes of play have a place in both antagonistic (PvP) and basic (PvE) gameplay and need not be reconciled.

Just as PvE and PvP is different, playing a rogue vs a paladin is different. It's doesn't need radical fixing, just tweaking.

One elegant design is to provide sufficient tools or options for players to develop counters and evolve their attack strategies and counters. Example of this is Magic: the Gathering (cited before) and the design for Guild Wars.

Right now, MMORPG are slow to implement tools or options for any designed or emergent unbalancing play patterns. The only tools and actions so far result in "nerfing" as any new changes affect what as been spent and accumulated: amount of money, time, energy spent on specializing on one aspect of the content.


Bruce Woodcock>Great, now when I unstealth and backstab you, I also lag you for a second or two while you get transmitted ALL of my data

That'll teach you not to unstealth and backstab people, then...

>(the stuff that would usually get burst to you when I first "pop" into existance some ways away) and load level 1 textures!

Why can't stuff be sent in bursts? It's fine to send everything in bursts as usual, except the location. The client knows all there is to know about the avatar, ready to render it when the time comes, but without the location detail the radar is useless.

If you think even the fact that there's someone invisible "within range" is too much information, encrypt the data while it's sent in bursts, then send the decryption key when you want the client to know what it's been sent.



For PvP aspects invis/stealth is definatly poorly implemented. You don't have environmental factors(being in the outside in sun, walking on snow, in water, etc) being factored in, or in a very limited manner. For most MMORPG you are invis/stealthed and that is it. If people can detect you they can if not there is no random chance that they will detect you; and since alot of this is governed by the UI you see the radar addons of WoW and DAoC. Even in the old PnP games there was always that chance that you would step on a twig or that loose floorboard giving yourself away, invis/stealth was good but you had that random chance that you would be found.

Then on the PvE side you have the same thing, if the MOB can see you they can if not you can walk right over to them and there is nothing they will do, and in most cases all MOBs had the abaility. This was primarily done to prevent players from doing an easy walkthrough to solve the quest. Future wise DnD-Online looks like they may be fixing it, from what they have said they are expecting people to have invis/stealth and should beable to use it. They talk about how you will beable to go into a dungeon and either kill everything you find or use stealth and sneak around, remove lights so they you can get by MOBs or even cause distraction to get around them.

Guess we will have to see.


Darniaq seems to be posting above that *every* class needs a counter to stealth. That's a design decision, not a law, and it looks like the EQ2 designers are betting the farm otherwise.

Their theory is more on the order of 'every *group* needs an effective counter to stealth'. They're not trying to balance all sets of 1-on-1 PvP combinations, which was something that generated lots of heat and not too much light on the Anarchy Online forums (or, apparently, among the designers).


I will say that I think some complaints in WoW about stealth are a bit overstated on the game-mechanical specifics. Many classes do have effective counters if they're ambushed when they're at full health and mana--I think warlocks are the only class where I would simply say that they should lie down and die if an equal-level rogue jumps them.

The key, of course, is that "full health and mana". In purely tactical terms, this is what I'm getting at, that the stealther has choice of attack, and this is a huge, intrinsic, asymmetrical advantage, whatever else might follow in terms of abilities in combat.

The academic question that I think is interesting about stealth has to do with asymmetrical experiences of the virtual world: a player who has stealth of some kind, especially stealth that has no energetic cost (can be on all the time save when attacking, as in WoW) really lives in a different gameworld, experiences it differently.

I almost think there's the beginning of an argument here for what I might call "overlapping instanced virtual worlds"--gameworlds where players are in the same world all the time but also able to go "out of phase" with the social universe, into a private space, while still present in the world and among other players. To me, that's the main appeal of stealth--not that it's an "I win" button, but that it allows me to move in and out of soloing and cooperating (and conflicting) as I will it, as the mood strikes me. Maybe that's something all classes, players, etc. should have in some fashion.


>Why can't stuff be sent in bursts? It's fine to
>send everything in bursts as usual, except the
>location. The client knows all there is to know
>about the avatar, ready to render it when the
>time comes, but without the location detail the
>radar is useless.

Well, I am not a modern MMOG programmer, so I don't want to dig too deep a hole for myself here. However, a lot of that information is generally distance-dependent, I think. Still, sure, you could send some information and just not location; while that would stop a "radar" the players would have potentially still have an "early warning" system -- they'd know a stealher was within range, just not where.

>If you think even the fact that there's someone
>invisible "within range" is too much
>information, encrypt the data while it's sent in
>bursts, then send the decryption key when you
>want the client to know what it's been sent.

I think most MMOs encrypt this data anyway, don't they? Still, I'm not saying a software solution is impossible. You are, however, creating a whole new codepath to accomodate stealthing, breaking a lot of convenient "assumptions" one had before when you could just push character data around according to simple rules, and clients now have a whole new structured data list to accomodate these guys. And that still doesn't really address the potential graphical lag, but that's more of a problem with lots of stealthers at once.



Game data is not encrypted. That's just a smoke screen. Take a look at all the server emulators that spring up. Every programmer out there is not an encryption specialist. Clearly, all this data is for the most part plainly readable. They just say it's encrypted to keep the looky-loos out. Meanwhile, even the most modest of encryption would go a long way to reducing hacks.

As for the lag on backstab... I don't know where you've been playing, but I get that all the time, stealth or no stealth. I think the only thing that is going to fix those problems are 'the next big things' in bandwidth. I read something a while back that said it could end electronic bottlenecks, so keep your fingers crossed for that. It'll be a big help.


I don't get it. Why is there this assumption that a whole lot of data has to be transmitted when a stealther uncloaks? I don't play any current MMORPG, so maybe there's something I don't know, maybe players now have the option of adorning their character's head with a high resolution image of their own face or something? But the way I'm seeing it, the looks of a character on screen are described by something like, use this model with these colour parameters and these textures -- where the actual model and textures are present at the client's machine already -- and slap a couple of simple stats on it, like player name, level, maybe health points -- nothing fancy. So there should be less than a KB, not very much data to transmit? If there's any problem I can see, it's loading the most part of the 'new' data locally, from disk, and pushing textures onto the GPU -- nothing network related. If this takes so long that the client is lagged, there's a simple solution: just don't freeze the client during this operation. Just keep every possible model loaded at all times, at least a low poly version of the mesh. I don't think there's a game out there that needs more than 1 MB for low poly versions of all possible player characters combined. When someone uncloaks, render him/her instantly, in a simplistic way, then flesh the representation out as the necessary data, like textures, becomes available. Smooth this all over with a nice decloaking effect, maybe an alpha-channel fade-in with a little blur. As I see it, there's no real technical problem at all, just lazy programmers who are satisfied with a solution as soon as it works just good enough on their high-end machines. Or did I miss something?


"Overlapping instanced virtual worlds" have a certain appeal. It's similar to running around as a ghost in WoW, but with more options.

For example, WoW can create an expansion that offers new classes and abilities to interact with the spirit world. Shamans, warlocks, etc. can cast spells that allow them to interact with both the material and spirital plane. It adds a whole new dimension of gameplay.

Another interesting experiment is to create a pocket of space where players of different servers can phase across different servers to test server-to-server links. To limit ill effects of this experiment, only chat is enabled.


Where you take a hit for loading things is entirely implementation dependant. There nee not be any more lag from a sealthed character than for any other character coming into view. The appropriate textures and models can with be loaded already without giving much of anything away.

In fact its quite likely that a modern client would already be sending information on characters to preload before they are within visual range. Characters without attached locations, because location may not be relevant yet. A stealthed character would like like any other preload.

Hypothetical griefing by stealth/unstealth is less likely than griefing by groups of characters with clothes changing macro scripts running. After the first state change the resources are probably already loaded. Not to metion that stealth implementations typically have a refresh timer which prevents a character from stealthing again immediately.

As far as encryption of data goes, The issue is again one of the client. The client decrypts the information and maintains it in memory. Many hacks even take direct advantage of client code to operate. Encryption is largely irrelevant when you are one of the endpoints of a transaction.

Anyway, I delved a little more deeply into some of the problems on one of the PvE servers (PvP servers have a significantly different situation). What I saw was this:

The ganking group currently consists of a small group of people usually 2-3. The levels range of the gankers is disproportionate to the level range of the area they are in. Most of the gankers with be 10+ levels higher. One, the "bait", will be several levels lower than the typical level for the area. The higher levels will travel stelathed with the bait. (IE: Rogue / Shaman / other class with invis). The bait will flag, and attempt to get another player to attack. After the fight starts the higher players will make short work of the attack, then they will continue. They will attack targets of opportunity.

There were also occasions of just pure grief play. These seems to rely on a large number of people having been lured into flagging for the chance at a honor point. The higher levels will then just plain gank people, and sometimes camp the corpses. When higher levels from the opposing side arrive the gankers will wait out their flags and just harass flagged players while waiting for chances to kill lower players without interevention.

Some things that might help with this behavior are:

Publicize the number of points required to gain rank. Players who haven't participated are unaware don't realized there are a significant number of points required to get any benefit at all. This makes them more easily lured into attacking. And more likely to try such baiting tactics themselves.

Narrow the level range for which honor points are gained. The range for Honor Points should be smaller than the range for regular experience.

Apply the same sort of damage proportion penalty on group honor points as on regular experience. Higher levels characters shouldn't be able to help lower character farm honor points.


Thabor>Encryption is largely irrelevant when you are one of the endpoints of a transaction.

But if you don't have the decryption key until it's sent to you, and it isn't sent to you until the server determines you need to know, what use is it that you're an endpoint?



Some time reader, first time poster.

I am definitely an Explorer type player - I like to roam the geography unmolested, and that's what I use stealth for when I have it available. It occurs to me that WoW already has a mode of stealth perfectly suited to Explorers: death! Not only are ghosts invisible and intangible, but they also have greatly increased speed. If ghosts had the ability to swim below water, I wouldn't have spent any time alive.

I would love an MMO which offers 'ghost' as a playable archetype. Ghost stealth seems to avoid the problems of rogue stealth: on the one hand, it is perfect versus mobs, and nearly perfect against players (except perhaps shamans, necromancers or the like). On the other hand, ghosts, being intangible, cannot use their stealth to appear out of nowhere and damage players. A ghost's offensive capabilities versus players could be limited to illusory monsters. In group PvP, ghosts play the role of scouts and anti-scouts.

The main problem with this is that ghosts lack any obvious capability to kill mobs and otherwise participate in standard advancement. Also, a game which uses PvP as part of an anti-grief strategy might be confounded by the presence of obnoxious ghosts who cannot be killed. Still, I think it's an avenue worth exploring.

EDIT: Looks like it's being explored already. This is what I get for coming in late.


There's definitely an argument to be made for a "phantom" mode that's independent of the PvP/PvE combat implications of stealth.

The "bait" strategy, by the way, is a favorite of some of the aggressive PvP guilds on my server. At this point if you see a caster class sitting down drinking or eating, you pretty much have to assume there are two or three stealthers right nearby. Actually, I think that's fairly inventive: it's a social hack as much as it's a game-mechanic.


But if you don't have the decryption key until it's sent to you, and it isn't sent to you until the server determines you need to know, what use is it that you're an endpoint?


I should have been more clear about which points I was responding to here.

You are correct that witholding the key would prevent them from getting at the data until the appropriate moment. I just see your suggestion of withholding location as being a better implementation.

I don't see texture and model information to be critical enough to encrypt when you've already removed the location. In fact it would prevent that data from being pre-loaded, and add some overhead for decryption.


Ian Zeilstra> I would love an MMO which offers 'ghost' as a playable archetype.

This concept was discussed for the "Force Ghosts" in Star Wars Galaxies, where it seemed like a natural. I don't know if it was ever actually implemented, however, or if so, to what degree.



Thabor>I don't see texture and model information to be critical enough to encrypt when you've already removed the location. In fact it would prevent that data from being pre-loaded, and add some overhead for decryption.

Oh, OK, I misunderstood what you were arguing.

Yes, I agree: you only need to remove location information for this to work. I only mentioned the encryption thing (which I first read about on MUD-DEV, and somewhat embarrassingly missed the point of it initially) as a counter to people who might want to argue that knowledge of the presence (if not location) of invisible characters was still giving too much information for client-crackers.



> Hellinar wrote:
> I think it comes down to poorly thought out
> design. In particular, a generic problem with
> most current MMOGs. Designers create a world
> with an “expected” maximum rate of loot and
> experience gain, they don’t actually code it
> into the world. They then react when these
> limits are breached by nerfing the offending
> skill, rather than simply coding a limit.

We actually considered the concept of capped xp gain rates or capped level rates (no more than X levels can be gained each month), but over time we decided against it.

We concluded that players would ultimately dislike the system enormously, and it would end up being biased against too many types of gameplay patterns. For example:

If someone takes a week off work, are they only going to be able to enjoy the game for a few days when they quickly hit the cap?

Will college student customers hate the game in the summertime when their amount of gameplay time might increase?

Would people feel like they HAD to be hitting the cap or else they were falling behind the curve?

I really like the simplicity of this idea though. Instead of having to constantly balance and rebalance things, figure out what rate of advancement you want and literally CAP IT. But the above concerns eventually outweighed the perceived benefits.

Perhaps we just never saw a good way to do it.

This is a pretty big tangent that I would love to explore further either here in another thread or on MUD DEV.


I don't really understand the coincept of capping experience gain while still forcing grind.

Since you're defining the rate anyway why not just dissassociate it from play activity entirely, and base the level/money on the delta between the character creation date and the current time.

Not that I would expect you to get much of an audience that way. Aside from the already metioned points such a system eliminates your ability to catch up to friends, and to pass competitors. That is a big component to lose for all the achievers out there.


It might be worth mentioning again the system that EVE Online uses (which I thought was pretty clever).

Instead of learning new abilities as a function of experience, in EVE it's a function of time. Each ability-granting skill has five levels, each of which takes a certain set amount of time to learn.

The exact amount is determined by applying a few modifiers to a base time for each level. Level 1 generally takes about 15 minutes; level 2 one or two hours; level 5 might take about a week. (And learning happens at the same rate whether you're online or offline.)

The obvious benefit of this approach is that it completely eliminates "grinding for skills." (There's still grinding in EVE; just not for skills.) The obvious downside is that the game's early adopters have the largest skill sets, so it's tough for a newbie to compete effectively. Two or three months' worth of play eliminates much of this gap, however.

It actually seems like a pretty slick approach to me, but then I'm not an Achiever. As Thabor noted, this kind of system might not go over well with Achievers who need to be able to convert energy (time & effort) into dominance.

Others seem to like it, though.



To the original subject, my opinion is that the problem is with the stealth mechanic itself. I'm not a WoW player so I can't comment on that game in detail, but as I understand it stealth allows a player to walk right up to a target regardless of the situation. Mister Rabbit has basically covered all the points on this one, but I'll go ahead and list my own.

1. It's illogical, i.e. the desert at noon situation.
2. It's immersion breaking. You *know* logically that if someone simply ran across a salt flat straight towards you in full daylight that you would see them.
3. It's too powerful. The ability to choose the battle and the enemy needs some kind of drawback or hoop-jumping.

A system based on perception and stealth scores is probably the best answer, however it needs to be balanced in a way that the stealth must be played logically. For example, a rogue needs to stick to dark areas, wear camoflage, etc. as Ed said.

As for the areas like the salt flats with no available cover, well, stealth has moments where it *shouldn't* work. Still, adding smoke bombs or perhaps darkness bombs and using environment-attuned camoflage could still allow limited use in those places.


I used to play Rogue on WoW.

Someone said that stealth is inherently unfair in an MMO.

But as a rogue, I loved the stealth aspect. It is so much fun.

The big thing is though, at least when I played, PvP was limited only to PvP servers. On a normal PvE server, you had to type /pvp or attack someone who had pvp enabled to fight another player.

Don't players give consent to griefing if they choose a PvP server? While I don't want to go so far as to say "If you don't like it, don't play it", people on PvP servers did choose to be on PvP servers. That should be taken into account when discussing this stealth issue
in WoW.

Also, in PvP, I know many mages discussed spamming their level 1 spell that did AOE damage in a circle around them. You could spam it regularly with little energy loss, and it would cause any rogues near you to take damage and drop out of stealth. Player solutions like this to stealth really appeal to me. (I realize that constantly spamming a spell while questing is tedious and unrealistic, but it's just an example.)

Another interesting alternative to stealth is the "disguise" ability. I know many MUDs use this ability to great effect. Wolfenstein Enemy Territory, an FPS with some RPG elements, implements this system really well.

Could an alternative to stealth be the ability to disguise ones' self? You could disguise yourself as a player or perhaps as a mob even. Then there would be a chance for a player to "see through" your disguise.

The whole disguise idea adds a lot of necessary strategy on the part of the rogue. I know as a disguiser in Wolfenstein, I had to know how to act like an enemy player. I had to play the part, which required careful skill on my part. Also, other players might be able to tell by my movements if I'm really on their team or not.

All in all, I think one of the problems with stealth is the helplessness you feel when taken by surprise. If there were a way to detect rogues in some way, getting ganked would be far less infuriating and far more embarassing. As it stands now, most lone players have no hope of combatting their would-be killer without resorting to drastic measures. That is not good.

I think that some stealth type system can be sucessfully implemented in a graphical virtual world. It's just a matter of finding the right system.


It seems that the whole balancing and rebalancing proposals of stealth in WoW will only add to the frustration of players who are having a great time, regardless of any disatvantages that may apply towards some classes.
I play a rogue on a pvp server and i'm having the time of my life. It's aboslutely true that most classes have a quick and easy way out of being instantly killed when i have gotten the jump on them, mages have blink, pallies have shield, priests have fear, etc. The point however is that in any mmorpg some classes will have advantages over some while having disatvantages over others. This shouldn't subtract from the enjoyment of the game but should cause players to think and develop strategies to win in a situation, even if they didn't get the upper hand in the beggining, such as a rogue stealthily backstabbing them whilst they were unaware. After all, thats a rogues job :).


Stealth, disguise, etc.... to me, the problem with every implementation of every type of deception skill ever to see the light of day in a multiplayer game boils down to an inherent focus on the wrong side of the equation.

To use the prototypical example: can you sneak up on someone in the middle of a flat, empty field?

The truest answer is not "it depends on how stealthy I am", but rather "it depends on how aware (or distracted) my target is".

My prediction: if/when you see a game that offers as many options, effects, and activities for -Perception- as for -Deception-, you'll see the everpresent "stealth" problem start to go away.

The comments to this entry are closed.