Slashdot Mirror
New ASUS Drivers Help Cheaters?
magicmat writes: "The guys over at Riva Station are reporting ASUS's nVdidia based video cards might have a new "special weapon" in gaming. Namely, drivers that allow you to see through walls, get brighter lighting or go to wire frame mode. Is it just me or does this sound like the smartest decision a card maker has ever done for their profits?" I always wanted to be able to reposition the camera... I mean, the driver should be able to do that, right? I guess it depends how much of the world the game is actually storing on the video card at any given point, but it seems like it should be possible. (Note that these drivers are not standard or officially released, while they are for the nVidia chipset, they are technically for ASUS cards. Sorry about the confusion.)
Replying to my own message here, something I forgot:
Email ASUS and complain here (marketing), and here (tech support).
(I had those email addresses in my story submission to Slashdot as well... oh well.)
Uh huh, until someone throws that console online via a null modem cable to a PC and uses the PC to run a proxy server...
Anything you can devise, they can break.
It's just that the good coders don't bother with consoles, closed cabinets == crap games.
I remember that in one of the versions of Q3Test, you could toggle a wireframe mode, but it turned out that the players were pretty much hidden through the walls until they came a few feet from where they would be visible to you. I would assume that was because the server was being smart about not sending other players' positions over the network unless they were within your approximate line of sight.. this saves bandwidth (especially in highly-populated servers) and makes cheating more difficult.
I'm pretty sure that most modern 3D shooters do something similar, so I don't really see it to be much of a problem for those fast-paced games. In slower-paced games where stealth is more important (Rainbow 6, etc), I imagine this might be more of a problem.
problem being once again this is not a cheater dll. this is a video driver hack. you cant crc the video driver because there are hundreds of drivers constantly being released, all with different version numbers and different crc values.
-Steve Gibson
-Steve Gibson
Shacknews.com
As a reply to Temporal and CaptainSuperBoy...
There really is *no* way to trust a client. None. Anything that the client does can be watched, instruction by instruction, with a debugger. Any encryption it uses is performed in front of the determined hacker.
Any CRC of the binary, or the data files, is done by the EXE that the hacker has access to. They simply hardcode the values the server wants to get. Or if it's a changing algorithm, they keep the old files around to run the CRCs on, then use the hacked files to play.
It is provably impossible to write a program that can't be modified in this way. There are some anti-debugger tricks, but they'll probably increase general incompatibility and, they never stopped the pros anyways.
You either have to have a release schedule that beats the hackers, like updating the EXEs daily, so that they have to keep doing the work... or, you accept the fact that you can't tell if you're communicating with your certified client or a proxy.
It doesn't matter what platform it's on. As long as hardware is available, not encased in melted plastic, and development tools exist, hackers will be able to examine the game. It might be harder to disassemble Super Mario 64, or a Playstation 2 game, but it can be done, and if the stakes are high, *will* be done.
There is *nothing* that can be done to make this impossible. All that you can do is make the cheaters job harder by blocking the obvious things (like the Quake1 cheats - new player models, etc).
Anything the server does, short of sending the game out as high-res screen shots over the network at 60fps can be hacked. And even then, someone could write a proxy that would parse that pictures and auto-aim or something.
Cheaters and cheat protection will (as I think Carmack said) evolve until subtle cheaters are indistinguishable from the better players. Anything you can devise to stop cheaters, the cheaters can learn from and avoid.
The performance hit with translucent walls will likely be rather severe -- z-buffering can't be performed on alpha surfaces. That means that the triangles would have to be depth-sorted and blit in order. (every frame!)
... even if the walls are textured, it's very hard to tell what's in front of what.
Wireframe mode on the other hand would be much faster. It's really hard to play with, though (do r_showtris 1 in quake3 while cheats are enabled)
Much less annoying in Quake 3 is "r_shownormals 1", which only makes a little tick mark on each vertex, showing its normal vector. Something like this would probably be more valuable to cheaters.
The camera cant be changed in a game, at least not by a driver, because it hasnt got anything to do with any info stored on the video card.
Usually, a 3D video board, half of the memory is used for textures and half for graphics. The actual game geometry is stored in RAM. So, unless a hack is written for a specific game, knowing exactly how it handles the world geometry and where (and how) it stores the camera data, you cant change the camera angle. It should be done before the frame is computed. The driver just renders the data given to it (so it can mess with transparency or wireframe it) but you already got the scene there. No changing camera angles.
It may be theoretically impossible to write a program that can't be indetectably modified, but from a practical perspective, you can make it damn near impossible, especially when it is dealing with interactive servers.
I don't want to belabor the details of some of the most dirty tricks that will give even NSA crackers a pause (deliberate race conditions in self modifying code, etc), but even non-programmers can understand it isn't that hard to bury half a dozen redundant checks in the game which at rare moments calculate and send disguised CRC verification stamps to the server. All it takes is for your uber-hacker to miss just one of these for the server to mark his unique game id as one held by a cheater. The server can even delay the penalty by many weeks, so as to obfusticate when the cheater was caught.
The real reason why cheating is so prevalent is that most companies don't bother to try. Hell, they don't even bother to use any real encryption on the game authentication id they give out.
San Francisco, CA - In response to nVidia's announcement of a 3D card that will allow a game player to see-thru walls, 3Dfx has revealed that they are only weeks away from releasing a 3D graphics card that can see-thru clothing.
The technology coined "strip-o-way" allows the owner of a 3Dfx "ClearSpeed3D" card to analyze any graphic layer by layer. Similar to the upcoming movie "Hollow Man," a splashy special-effects remake of the popular movie "The Invisible Man," 3Dfx's technology will allow the user to see through-images.
In an late-afternoon press conference, 3Dfx Chairman, President and CEO Thisis Ajoke announced that, "Our technology is meant not just for the games people anymore. We have expanded our product line into a much larger market." Mr. Ajoke refused to comment what he meant by that comment, but, a source close the development effort said: "Think about it! What would you do if you could strip away, layer-by-layer, the clothes off any image of Gillian Anderson you find on the.. err.. I mean, you could possibly find out who shot Kennedy."
3Dfx is expected to release their new product before the Christmas buying season, but industry analysts already expect 3Dfx's stock to "go through the frik'in roof."
Some computer owners might have to upgrade their systems as the new video graphics technology is expected to take up 4 PCI and 1 AGP port, and require a minimum of 5 external power supplies to drive enough juice into the cards.
3Dfx is seperately negotiating with Intel to increase the mega-wattage it pumps through the system bus to make "all those dangly wires and power-strips" unnecessary.
Pricing has not been announced for the new product, but in light of Microsoft's recent announcement of becoming an Application Service Provider, 3Dfx is considering folling the software innovator's lead with their own ".accell" model. 3Dfx's engineers were unavailable for comment, as they were beating the crap out of their Marketing department at the time we called them up.
- Some hungry gremlins contributed to this story.
Hah, people criticize Ryan at MacOS Rumors for not checking his stories... nice work, Taco ;^)
Just venting (grumble). If only this was a realistic picture of the situation...
I've always had an advantage in Quake 3 because of a slight z-buffering problem with my video card (Real 3D Starfighter, 8MB AGP on a socket7 mobo, intel's initial graphics chipset). It allows me to see where people are through walls and floors without disrupting gameplay.
I've since switched to Unreal, where I don't have the problem (just a problem with OpenGL?). But people would always ask me why I still had that old thing in there :)
--
Never trust anyone over 90000.
Diablo II deals with cheating by not trusting the client all - according to Blizzard, in realm games all that's sent to the server is your mouse and keyboard actions. They can get away with this only because the game environment is extraordinarily more constrained than that of, say, an FPS.
--
--
Do I look like I speak for my employer?
.. Lihe I don't have a hard enough time trying to find a 'clean' game. What is the performance trade off? what good is seeing throught walls if your frame rate is 2 ?
The Kruger Dunning explains most post on
Back when Sc2000 was first released I did release sc2kcash.exe onto a few bbs's
Now that felt like an achievement.
-----------------------
Nicotine free Amish .sig.
Sure, the ability to make walls transparent is great for the company's bottom line, but think of how much more successful it would be if it also allowed users to make certain characters' clothes transparent.
"If one is really a superior person, the fact is likely to leak out without too much assistance" -- John Andrew Holmes
I occasionally cheat (on single-player modes) because, and I'm not ashamed anymore to admit it, I'm not good enough at most of the games out on the market.
I used to be able to keep up with the arcade stuff & early Doom-clones, but as each type of game has matured & there are people with lots of time to do nothing but practice, the companies have cranked up the base difficulty level to the point where it would take me months of constant playing to improve my skill level to the point where I could complete the game (I'm usually into RPGs & first-person-working-toward-a-goal-type games).
I'm busy working, I visit my family a lot, I like to get more than 2 hours sleep per night - I don't have the time or health to build up the skills necessary to complete the popular games (which are not necessarily useful to me in the rest of my life). So, in single-person modes, I cheat so that I can enjoy the game in about a week of a few hours/night.
On the other hand, I wouldn't cheat in any kind of multi-player mode, due to the ethical concerns. (Of course, that generally means that I just get my butt kicked, because I still have problems with the difficulty level...) I don't agree with your statement about cheating being a practical joke - I think people who cheat in multi-player mode just want to win in any manner possible.
It'll take video game graphics processing off of the graphics card and put it back on the main CPU, where it should be. Back in my day all of the graphics was rendered in the CPU before sending the data to the graphics card. This usually resulted in a respectable frame rate of 1-3 frames per second, not that fancy schmancy 100 FPS that all you kids think you need for a good time.
Since these disrespectable silicon valley hardware companies want to allow cheating with there cards, more and more graphics rendering will have to take place on the main processor. That's a step backwards.
To respond to UnknownSoldier's comment and a few other comments too, I have been using CAD systems since before CAD systems were made and you are wrong about rendering speeds. There are 2 different wireframe styles of rendering. The first is called 'wireframe' and it displays all edges in the entire FOV with no shading. It is always faster then fully shaded rendering. The second type of wireframe rendering is called 'line-clipping' and it is always slower then fully shaded rendering. This is because it is mathematically complicated to determine where to start and end lines. Line clipping looks like wireframe, except that you can't see wires behind other objects.
All that hogwash about graphics cards only using triangles is utter nonsense. Triangles and polygons are used for benchmarking purposes, but there are NO graphics cards that only use shaded triangles for 3D rendering.
-vax computer, vi, lynx. 'nuf said
Most (99%+) multiplayer gamers (first person shooters especially) really and truly respect the integrity of the game. If the integrity is destroyed for me, the game has lost all of its fun. I don't understand cheating, but the fact is that it can utterly ruin a game (Again, both Diablo and QW. Diablo never recovered, but the QW community seems to be making progress). The solutions to the Diablo and QW problems were both technical.
It's clear that video card drivers are bound to come out (just look at the long list of nVidia beta drivers that have been leaked). We can't count on ASUS reconsidering releasing these. I think the solution this time rests on the game designers, in the form of some kind of driver signing / certification, a la Microsoft. Don't have a certified driver? Can't play the game.
I don't know a lot about 3D, at all (I hope Carmack replies to this story). You'll see that the worst cheat in this driver is the "see through walls" and "wireframe" modes. If it's not too much of a performance hit, games may have to stop sending information to the video card about polygons that are not visible. I think this would probably kill performance though, as determining which objects block the view of which other objects is a primary function of hardware 3D.
I hope we can learn from our previous battles with cheating. Though I don't think this will do much in the long run, I hope that game companies show that they are not pleased with the decision of ASUS.
--
Networked games shouldn't trust the client. The client shouldn't have any information that the user shouldn't be aware of. If you send some data to the client, you have to assume that that user is going to be aware of that data. This includes things like the location of enemies or objects that are out of sight, as well as any other attributes that the user shouldn't know about (like say, how much ammo your enemy has, or his health).
This would prevent cheating with these drivers. That, in combination with an RSA "blessed client" scheme, like that used with Netrek, you can drastically reduce the types of cheats that are possible.
This adds some complexity to software, of course. The server has to determine what information to give to the clients. That means it needs to do basic line-of-sight determination, for example. That can be tricky though. Just because the guy's toe is sticking around a doorway, it doesn't mean I can "see" him -- I might not notice him.
A simpler solution would be for the game developers to do more clipping on their own, rather than relying on the hardware. Of course, that could potentially slow things down significantly. It also doesn't deal with other kinds of cheats that involve modifying the client itself. (particularly a problem if the client's source is available, but even binary clients can be "tweaked") Not trusting the client is the most general solution.
> Are games about fun or popularity of having the coolest shit (to the kids)?
There seem to be two (at least) basic classes of mentality among gamers of all types.
I used to play D&D with a group. For some of us, the fun was in the adventure, the puzzles, the humorous occurences, the challenge to overcome setbacks, etc. For others, the fun was in "winning". These people would go to the local game stores, dump out the whole box of dice, and roll each one over and over until they found, or thought they had found, a high-roller. Etc.
Similarly with wargames. For some of us the fun lay in understanding history, testing strategies, recovering from mistakes and/or bad luck, etc. For others, the fun lay in winning. If the rules had a bug that allowed patently unrealistic operations - well, no problem if it led to a win! Etc.
I, for one, can comprehend the desire to win, but if the "win" is merely nominal and results from cheating or exploiting a loophole, then I absolutely cannot fathom how any satisfaction is obtained from acquiring it.
OTOH, I recognize that my notion of fun is surely equally unfathomable to parties in the other camp.
--
Sheesh, evil *and* a jerk. -- Jade
* Is there any confirmation of this outside of this guy's site? Does anyone here actually have said drivers?
* Go to ASUS's site, and read press releases by Mr. Tsang - he is much more articulate in them than he is in the press release on Riva Station (which, coincidentally, is NOWHERE on ASUS' site).
* The Wireframe screenshots were, also coincidentally, taken *both* in the Q3 engine, which happens to have a wireframe cvar you can use if cheats are enabled.
* This just screams 'page hit scam'.
Did I miss anything?
> I'm still waiting for an answer.
Again, you illustrate my claim. I gave my answer in my original post, and you are as incapable of understanding it as I am of understanding your system of enjoying a game.
> "The greatest happiness is to vanquish your enemies, to chase them before you, to rob them of their wealth, to see those dear to them bathed in tears, to clasp to your bosom their wives and daughters." -Genghis Khan
OK, so GKh was of your camp. BFD. Did you hope to convert those of us in the other camp by citing him as an authority?
--
Sheesh, evil *and* a jerk. -- Jade
> A kid at work comes up behind me (it was after hours :-))...
...and says, "Why don't you just turn on the flying cheat." ... WHAT? GET THE F**K AWAY FROM ME! I can't fathom this personally. And in a SINGLE-PLAYER game.
Yeah, sure.
>
Hahaha! I won! Nyeh, nyeh, nyeh!
Truly, incomprehensible.
--
Sheesh, evil *and* a jerk. -- Jade
It all depends on how willing you are to sacrifice CPU time. In a short crypto class we took, we examined how to play a secure game of poker (completely randomized deck, no one knows what order the deck was in until after the hand). Its among the coolest stuff in modern crypto. Bruce Schneirs book Applied Cryptography goes into this, I believe. Also, go, chess, tictactoe, battleship, and more can be done. I think you could do a FPS through crypto by "hiding" where one player is throughout the game unless the other can see them. However, it takes VERY high bandwidth, VERY low latency and LOTS of CPU time (I think the rendering would have to ALL be software...). At the end of the game, you could check that it was all played fair by examining paths if you chose to. Not doable on current systems. We need about 2-3 more orders of magnitude to play TODAY's games, let alone whats out by then...
True, the hardware could do all of this for you, but it wouldn't help much if the software was smart enough to speed things up by not putting things in the environment that wouldn't be visible anyway. Think about it: some games base what's visible on a "per-room" basis, rather than attempting to display the entire environment at once (unless the doors are open, then it's the next room over too, but usually not 2 rooms over). The hardware would only display the software calls that were made (eg if the software was based on OpenGL then the program would probably not draw the ENTIRE environment, but just the parts it felt necessary, which may or may not be the person sneaking up on you around the corner). Usually, for speed, software does not command the hardware to attempt to draw the whole world.
-Leo
This is indeed nothing new, and stuff is available right now. Just not very wide spread. Here's Zoid bit on it also which you may find interesting:
-Steve Gibson
-Steve Gibson
shugashack.com
This sort of thing happens without the tools in at least one game that I've seen.
:)
I've got a Geforce and noticed weird effects while playing Tribes with detonator 5.22; objects would become partially transparent at large distances and seemed to "segment", each segment updating on its own as I move, sides aliasing and all.
While being able to see through walls at a distance sometimes is a small advantage, the effect is overall distracting. I'd gladly give up seeing through walls if I could have proper rendering.
Maybe it's just me, but playing in wireframe so I can get a few more points simply isn't worth giving up my delicious graphics.
---
Where can the word be found, where can the word resound? Not here, there is not enough silence.
"Where shall the word be found, where will the word resound? Not here, there is not enough silence." -T.S. Eliot
incredibly difficult. this hack is done at the driver level and not at the game code level. they can crc the client all they want and it will still return the same value as anyone else.
-Steve Gibson
-Steve Gibson
shugashack.com
I can see it now: Quake map designer sues nVidia for violation of the DMCA. "They circumvented my protection I placed on the other players by putting walls in the map." Gawd damn, go call Jack Valenti.
Yes, the framerate _may_ actually be _slower_ in wireframe with some video cards.
Back when I first got my Voodoo 1 there was some demos showing off the speed. The card was actually SLOWER in wire frame mode! It didn't make any sense.
BUT if a card doesn't have points and lines as primitives, but only triangles, then the geometry/lighting processing may be taking more time then the time saved from not texturing.
All though, I would think that with modern cpu's and graphics cards with geometry engine there shouldn't be any frame rate hit.
~
"Triangles are the 3d pixels of today's graphics"
- Anonymous
IMHO, cheaters sometimes cheat because they cannot play otherwise. Even an empty victory is better than no victory for some. For single player gaming, I suppose cheating is acceptable for those miserable souls; they ruin it only for themselves.
However, multiplayer games are different. Cheats *do* cause others not to enjoy the game in this instance, and the only explanation I can think of is the same as above: Defeating others even with underhanded methods is better than not defeating them at all.
They simply can't play. Have you *ever* seen a cheater skilled in combat not involving cheats?
Yeah, cause any stupid shit can plug a console in, it takes some ammount of brains to master typing. It's also a cheaper system.
But you have to be pretty fucking blind to think a console game can beat a PC game in depth or interactivity.
A lack of onboard storage (large flash ram or HD) cripples Consoles for anything but stateless games like Mortal Kombat or some pathetic RPG like Zelda where 'saving' the game loses everything about you expect for some basic numbers...
Unlike playing Deus Ex or System Shock 2 where you can revisit an old area and see the corpses, right where you left them and with the bullet holes in the right places, etc. If you can compare a console game to these in terms of depth, you're smoking crack...
Yeah, the idea of not stopping cheaters, just ostrasizing them. Works for me... Except, then you get into the case where a sysop needs to examine a game log and determine if something was cheating or not. And if it's something like Quake where the servers are distributed, how do you know the server admins are honest and not labelling someone as a cheater just because they won a game...
The idea works if you never go outside of your group of friends, but it has problems when you go beyond that at all... And I do like the occasional sixteen player games which I can't get enough friends together to play.
But overall, yeah. If you know KillMaster is a cheater, just don't invite him to your games.
It's when you go to a global form of this that it's a problem... Do you let users register names so that only one person can have a certain name, or what. How do you block people, CD Keys?
In other words: Is that a rocket coming at m--**BLAM**.
- A.P.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
Like we really need this bit of technology. As if people don't already come up with enough ways to cheat, we certainly don't need to be making it even easier for them. Sure sure, "it's just a game". But a) people pay good money for many of these games, and cheating can really ruin it for them, and b) how you play reflects how you live. I know most people still don't believe this simple precept, but if you're willing to cheat in a game, or even encouraged to do so with tools like these, chances are your real world behavior is going to learn a bad lesson from your gaming experience.
Unfortunatley for 3Dfx, they're going to have to rush their development cycle! It's already been done.
Wah!
The problem is that client movement is predicted rather than synchronous from the server. You don't want to come around a corner, see nothing, because the server doesn't think you are there yet and hasn't sent you the information about something that will, after a round trip delay, pop into existence ... Yes, you can do better than what the Quake engine games do (region to region visibility lists), but because players can change direction unexpectedly, the server doesn't really know where you'll be / what you'll be able to see by the time its updates reach you. You can tighten the filtering of what the server sends to clients, but they will just lie about their latency to get a broader uncertainty range. Or you can be strict, and only people with truly low latency will be able to see an entity as soon as they come around the corner. It's even more complicated when clients predict not only their own movement, but that of other entities.
(anal article correction: The driver is made by Asus, not nvidia, although it may work on any nvidia card.)
Apparently, performance is significantly slower, so that's at least one disadvantage of using this to cheat. I for one don't understand cheaters anyway - whats the point of playing a game if your just going to cheat? It stops getting fun really quick. I never play anyone I don't know anyway, partly because I'm not very good, but also because most of the fun of playing is social, i.e. get a bunch of guys together, drinks some beers, blow each other into mip-mapped chunks with a rail gun.
---- I made the Kessel Run in under 11 parsecs.
These are not Nvidia drivers, these are ASUS drivers built off of the NVIDIA ones. In other words, they aren't "Official"..
:)
:)
Then again, neither are the 5.32 series Detonators from Reactor Critical, but I use those
Happy fragging with fast drivers
Other 3D rendering thingies, those that communicate more directly to the hardware or come in more flavors, will be much trickier to deal with. As for open source drivers...
Well, on linux, God Mode will finally include omniscience.
My mom is not a Karma whore!
No shit. I can not believe that ./ would stoop to this level.
Linux O Muerte!
There's only one perfect solution that makes seeing invisible people impossible (far as I can see): don't send information about their location to the client.
You'd have to make the game server not send information to a client that a client shouldn't be able to see. Send a "disappear" message when it goes out of view and no more messages after that until it reappears.
Otherwise, you could cheat by having a hacked client anyway. Supermap or something.
Only problem is the crazy amount of processing this would require on the server. It wouldn't be quite as bad as the processing the client must do, but still, you'd have to calculate a view for every player, which would get pretty bad.
Just a silly suggestion from a non-game-writer.
--John Keiser
I submitted this story about 8 hours ago with the *correct* facts, but it didn't get posted.
Anyway, ASUS are releasing a modification of Detonator 2 drivers (based on 5.32 I think) which will add features allowing you to switch to wireframe mode, partially-transparent-everything mode, and also add extra lighting to a scene.
This has nothing to do with nVidia bar the fact their drivers are being used as the base, and Slashdot are making them look bad by not checking facts before posted a story.
Obviously neither the person who posted the article to Slashdot, or CmdrTaco actually read the URL supplied in the article.
*sigh*
This is a pretty short sighted move by ASUS. Another company, Wicked3D, tried this a while ago and met with a lot of anger in the gaming community. I really hope this happens again, and ASUS decide not to release the drivers. Otherwise the online gaming world will either be based on trusting your opponent (not likely), or everybody cheating as much as possible, and so will begin a horrible downward spiral into out-cheating each other, rather than gaming.
Obviously, the client needs to have the same validation at its own end, but would this be unworkable?
---
Book(n): Utensil used to pass time while waiting for the TV repairman
Really, if you think about it, isn't cheating in this way just defeating some security through obscurity the developer created? After all, the information about what's hidden behind the walls is there, these drivers don't create it... It's just that it's been in too obscure a format to read before now.
So those who seek to powergame and "be the best" will have another tool which helps them to do so more easily. Those of us who play for our own enjoyment (isn't that the point of games?) will ignore them, and continue to enjoy the games. Will the two groups meet? Unhappily? Of course. They do so now even without this.
Now, the question really worth discussing here is, is it possible to create a multiplayer game where there's no advantage to be gained by examining memory, or intercepting the data stream? What would such a game be like, and would it still be able to attract a wide audience?
AR Schleicher (Jerrith)
ars@iag.net
to me-- from someone who has worked at a Babbages for a while-- how many kids even play games anymore? Most kids I talk to just buy games (like Diablo 2, etc) and get the best video card on the market at the second-- and play until one of their friends gets something better. The same stance is taken by the grade-school kids with Pokemon cards.
Now I know that adults play games too-- but the whole idea of being able to "see through walls" seems pretty juvenile to me. Are games about fun or popularity of having the coolest shit (to the kids)?
I did however get it to finally work with XF4.0.1 and their last set of drivers. OpenSource aside, it does work and kicked ass against my Voodoo3.
Although the site uses Quake3 screenshots to showcase these new features, Quake3 is not where the worst problems would be seen. In Quake3, there is an advantage to seeing where another player is, but you can't do anything to them until you go to their location.
A mod for Half-Life, Counter-Strike, which by far is played online by more people than Quake3, features rifles where wall penetration is a factor -- i.e. if you can see someone through a wall, you can probably shoot that player.
--