Automated Bot'ing: WoWGlider vs Blizzard

[Hellinar]

Maybe I should be rooting for WoWGlider to win then. If computer bots are allowed to do what human bots do now, only more efficiently, it might spark demand for a game in which grinding 60 hours a week isn't the road to victory.

[Simond]

No, it'll just kill development of MMOGs in favour of Second Life-alikes.
Have fun with that.

[CmdrSlack]

No, it'll just kill development of MMOGs in favour of Second Life-alikes.
Have fun with that.

The intarnet sky is falling!  The intarnet sky is falling!

[Nebu]

No, it'll just kill development of MMOGs in favour of Second Life-alikes.
Have fun with that.

... or produce more games like Mount and Blade where dynamic character interaction >>> autoattack. 

[Lantyssa]

Bots do more than just get xp though.  They can gather resources and loot corpses for items and money.  They could even be used for something as simple as going from Point A to Point B, ala macro-walking back in the MUD days.

Loot is going to be as much or more of a driving factor than xp if it has any significant bearing on character performance.  When botting gives you the ability to be start your own little RMT company it is also going to have an influence.

If botting will simplify the game in some fashion, someone will use one.  The more payoff it gives, the more widespread it is going to be.  And these people consider it a challenge.  A game can never be patched so fast that someone won't be able to code around it.  Designing against bots eventually hits a point of diminishing returns.  Companies are all about getting the most effect for the least amount of resources, and once design limits are reached, discouraging people with legal action becomes attractive.

[Alkiera]

Given the history of DMCA crap, I'm guessing the answer is yes. Nowadays you don't have to break a law yourself, making it possible for other people to do that is enough.

Which is why class action lawsuits against gun manufacturers always win - oh sorry guns are real not bits in a computer, my mistake!

The point you make here is good with one exception:

WoWGlider has one use as a product: breaking the WoW EULA.  It is not possible to use a copy of WoWGlider in a way that does not break the WoW EULA, thus there are no legal uses for it.  For the developer, maybe it was orginally an exercise or test of programming skill.  But once he started selling it, it was a product that was impossible to use without breaking the law.

Guns could also be considered an engineering exercise for the designer/builder, but when sold, they have plenty of uses that are not against the law(depending on your exact location, of course).  Lawsuits against Remington for damages caused by some lunatic shooting someone are just as valid as lawsuits against Boeing because some nuts flew their planes into a building.

I'm all about personal responsibility.  If people are breaking the law or your EULA or whatever, ban/suit/prosecute THEM.  Suing someone else because the people actually breaking the law are hard to find, or because the real cause is something you can't prosecute (people are broken) is a cop-out.

It's difficult to find good real-world analogies, because so many real world objects have multiple uses.  A software tool like WoWGlider only has one use, and that use is not legal, according to the terms of the WoW EULA\Service Contract.

To me, I'd attempt to reach a settlement with WoWGlider with an NDA that added some code to the software making it detectable by Blizzard... then wait, and ban those using it.  Have the financial portion of the settlement hinge on maintaining the NDA.

Alternatively, have a few spare coders at Blizzard create something similar that worked better, sell it for less, and then later nab everyone using it.  Over and over again, so that you see people buying these things and then getting caught by Blizzard.  It'd ruin the market for them.

As a (somewhat technical) side note, most of these things function by starting the game process themselves, so they can get hooks into the DX interfaces for the keyboard and mouse, possibly even memory space.  Is there not a way to tell who your own parent process was?  If it wasn't %SYSVOL%\Windows\explorer.exe, you've got a potential problem?  Heck, just send that info to Blizzard, let them collect a list of names and investigate.  The names of common Windows UI mods like ObjectDock or Dashboard, etc would  be filtered out, and what's left would be suspect.

--
Alkiera

[Azazel]

You guys do realize that the game itself is bots right? You can't make a video-game un-bot-able because all the NPCs and mobs are just bots themselves. It's just a question of who has better bots.

That was my thought especially when people were talking about bots in FPS games. Um. HELLO? Fuck aimbots, ever played BFanything or UT200anything in SP mode?

[squirrel]

No, it'll just kill development of MMOGs in favour of Second Life-alikes.
Have fun with that.

... or produce more games like Mount and Blade where dynamic character interaction >>> autoattack. 

Ya that's an awesome mmog. I think I'm still subbed to it...no wait, hang on...

Point being, there are still latency and hardware restrictions on what can happen in a MMOG unfortunately.

[Merusk]

You guys do realize that the game itself is bots right? You can't make a video-game un-bot-able because all the NPCs and mobs are just bots themselves. It's just a question of who has better bots.

That was my thought especially when people were talking about bots in FPS games. Um. HELLO? Fuck aimbots, ever played BFanything or UT200anything in SP mode?

No.   :-D

[Krakrok]

But once he started selling it, it was a product that was impossible to use without breaking the law.

There is a difference between breaching a contract (EULA) and breaking the law. I can breach EULAs all day long and not break the law.

[Margalis]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games? Is it because when you play a MMORPG the speed of light becomes slower? Or because that's a BS excuse without any technical justification?

[Stephen Zepp]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games? Is it because when you play a MMORPG the speed of light becomes slower? Or because that's a BS excuse without any technical justification?

Uhh, simple math?
100,000 X (bandwidth per user) >>> 16 x (bandwidth per user).

Not to mention fd_select(addr) or whatever the syntax is (been a LONG time) x 100,000 takes a bit more time overall than x 50?

[Trippy]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games? Is it because when you play a MMORPG the speed of light becomes slower? Or because that's a BS excuse without any technical justification?

Uhh, simple math?
100,000 X (bandwidth per user) >>> 16 x (bandwidth per user).

Not to mention fd_select(addr) or whatever the syntax is (been a LONG time) x 100,000 takes a bit more time overall than x 50?

Bandwidth != latency.

If Margalis was responding to squirrel the point is is that the twitcher the gameplay the less suitable it is for MMOGs where latency can be unpredictable depending on the number of people around you at any given time. There are, of course, ways to cheat, like doing the calculations on the client instead of the server like PlanetSide does which saves you the round trip time but that wouldn't work if you had blocking and other such mechanics in the gameplay. Even in PS there are lots and lots of times I've fallen over dead after having gone behind cover thanks to their client-side hit detection system.

Edit: extra words

[pxib]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games?

Persistance.

Latency isn't physically any better or any worse in MMORPGs, but the heavy commitments inherent to character development make latency (and other lag) particularly emotionally taxing. Every once in a while a long game of Starcraft might be decided by a hair-tearing bundle of lag during a climactic mutalistk rush, but usually it's just a quick game and if the latency goes south you back out and find a new one. In a Mount & Blade MMORPG temporary high latency means your favorite knight can't block anybody, or that your archenemy is literally impossible to hit.

Sure, you could leave the game and wait it out, or go grind some less latency-intensive activity... but while in Starcraft you can try more Starcraft because you wanted to play Starcraft, you came online specifically eager to go to the tournament with Sir Castic, and you can't just find a less laggy tournament for him.

I had no trouble with latency or lag trying Planetside, for example, because I honestly cared very little about either the war or the life of my particular soldier. There was always another base or tower to try to attack or defend. I observed latency and lag, of course, it just wasn't a problem.

[Lantyssa]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games? Is it because when you play a MMORPG the speed of light becomes slower? Or because that's a BS excuse without any technical justification?

Uhh, simple math?
100,000 X (bandwidth per user) >>> 16 x (bandwidth per user).

Bandwidth != latency.

If the bandwidth is greater than the tubes can hold, packets will be delayed and can functionally behave the same as, or worse than, a high latency connection under the right conditions.

I would think graphics lag is more likely to be a problem than bandwidth in this day and age, but theoretically it could happen, and isn't entirely far-fetched with some of the shoddy coding we've seen in games.

[squirrel]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games? Is it because when you play a MMORPG the speed of light becomes slower? Or because that's a BS excuse without any technical justification?

Uhh, simple math?
100,000 X (bandwidth per user) >>> 16 x (bandwidth per user).

Not to mention fd_select(addr) or whatever the syntax is (been a LONG time) x 100,000 takes a bit more time overall than x 50?

Bandwidth != latency.

If Margalis was responding to squirrel the point is is that the twitcher the gameplay the less suitable it is for MMOGs where latency can be unpredictable depending on the number of people around you at any given time. There are, of course, ways to cheat, like doing the calculations on the client instead of the server like PlanetSide does which saves you the round trip time but that wouldn't work if you had blocking and other such mechanics in the gameplay. Even in PS there are lots and lots of times I've fallen over dead after having gone behind cover thanks to their client-side hit detection system.

Edit: extra words

Thanks Trippy - i was a bit snarky but this is really what I was trying to say. Bandwidth as Stephen defines it compounds the issue(s). If you have a Mount & Blade type game you have to ensure 2 things:

1. - It can scale. This is what Stephen is referencing I believe. 64 + client connections is vastly different from 1500+ and persistancy means you can't trust the client software, particularly in DIKU/Achievement based MMOG's. We're just not there yet.

2. - Latency. Anyone remember the days of HPB's (high ping bastards?). Compounding issue 1 is the fact that in twitch/response based games people with faster gear and connections have inherent advantages over others - in orders of magnitude greater than the current advantages enjoyed by current hardware/network disparity.

End of the day - we're not at the point where un-trusted client/server MMOG's can have persistant 2000+ player worlds with Mount & Blade combat.

EDIT: Speeling

[Calantus]

In CS before we got broadband I found it easier because you could round a corner and kill someone before you showed up on their screen. Client side detection is so dodgy when it comes to low ping it's rediculous. I also remember a game, and I'm not sure which one it was, but people would deliberately slow down their connection because it would give them an advantage due to client-side wankery. It might have been CS.

[SurfD]

To me, you don't want bots in your game, make a game that can't be botted.  Otherwise STFU QQ more, nub game publishers.

Please, enlighten me as to exactly how one would go about makeing a game un bottable? From what I understand, many of the bots out there are reading info directly out of the memory spaces WoW uses, and feeding back responses in the same way.  Given that the WoW guys already have their Warden thingy in an effort to try to prevent this, what other suggestions would you offer?

heck, im also in the camp of: Play the game yourself or dont play at all.  letting an automated bot run your character to farm or grind or whatever will just lead to bad places.

[Venkman]

1. - It can scale. This is what Stephen is referencing I believe. 64 + client connections is vastly different from 1500+ and persistancy means you can't trust the client software, particularly in DIKU/Achievement based MMOG's. We're just not there yet.

Dumb question here:

Does the latency/bandwidth needs truly need to scale to 1,500+ concurrent users if the game is heavily zone based where each zone is a separate "connection"? I'm not sure how to ask this, so I'll go with examples:

• Shadowbane- This was by-and-large all persistent, so you theoretically could have every player on a server in one space.
• Guild Wars- The only persistence are the public-spaces for chat and trade. Everything else is instantiated.

So now I look at Planetside vs Huxley. PS used zones but had a pretty high cap on how many could be on an island. Huxley seems to be focusing on smallish 64-person battles with a persistent world for chat and trade (and lobby/matchmaking, etc).

To me it seems like if you enforce a rigid cap on the number of people who are allowed into a battle, you could get around the latency/bandwidth issue by designing a system where each zone was a separate, err, "computer" or "IP" address or something where you don't have all of the bandwidth for the entire game or even just the entire server trying to fit through one small hole.

[pxib]

To me it seems like if you enforce a rigid cap on the number of people who are allowed into a battle, you could get around the latency/bandwidth issue by designing a system where each zone was a separate, err, "computer" or "IP" address or something where you don't have all of the bandwidth for the entire game or even just the entire server trying to fit through one small hole.

There will still be moments of lag and high-latency play just because of the way the internet works: a temporarily slow connection between the player and her ISP, slow connection between two random routers somewhere along the way to the game server, a denial of service attack in Hong Kong, a butterfly flaps its wings in Paris, France. If it is absolutely critical to the player's happiness and mental state that their character dodges incoming fire by getting behind a rock, one of these days they're going to be furious and unhappy. In a game where one death can make you lose more than an hour's worth of loot or experience, lag burps really ruin your day.

There is a limit to how much this can be soothed on the technical side. The rest is purely psychological, and requires a game designed to make failure less critical.

[robusticus]

To me, you don't want bots in your game, make a game that can't be botted.  Otherwise STFU QQ more, nub game publishers.

Please, enlighten me as to exactly how one would go about makeing a game un bottable? From what I understand, many of the bots out there are reading info directly out of the memory spaces WoW uses, and feeding back responses in the same way.  Given that the WoW guys already have their Warden thingy in an effort to try to prevent this, what other suggestions would you offer?

heck, im also in the camp of: Play the game yourself or dont play at all.  letting an automated bot run your character to farm or grind or whatever will just lead to bad places.

Well, there have been lots of good ideas posted in this thread.

It does seem very odd, doesn't it, for someone to pay $60 for a gme, $35 for an expansion, $15 * x for subscription... then on top of that add $25 to buy a bot to play the game for you.  What's the point, why bother with spending anything at all?  That's the reaction most people have at first when you tell them about this.  And that's the way it should be, technically: there shouldn't be a point to botting.

So if MDY wins Blizzard will start selling the ability to have your alts max level, for $25.  The vast majority of WoWGlider's business will evaporate, and the avatar business will decline heavily as well.  They may go further and start selling coin, at which time the grey market RMT business will take a nosedive, as well as the vast majority of WoWGlider's remaining business.  After that, the rest of the people who use WoWGlider (all 5 of them), can be banned.

I have a couple of questions.

Has anyone here ever quit WoW because they couldn't finish a quest or farm enchants for the nightly raid or whatever due to bot obstruction?  I ask this because I only know one person in the world who actually plays this most-lucrative-entertainment-product-ever, and she said she's never seen a bot in WoW, much less had a problem.

What percentage of the population needs to have broken the EULA before it becomes apparent the problem isn't human nature, it is the EULA itself?  The point about community being a defense is a good one.  That's what happened on Steam servers - it wasn't tolerated by 999 out of 1000 players and people ruled their servers with an iron fist.  This is different though I think the percentage against is a lot lower.

I don't see a difference between a bot and someone who plays 40+ hours per week, or a paid farmer.  To me this case is about who gets favored.  None of the above, if you ask me.  But I also believe none of those types should be persecuted, either.

[Simond]

... or produce more games like Mount and Blade where dynamic character interaction >>> autoattack. 

How would M&B combat be unbottable in any way?

Has anyone here ever quit WoW because they couldn't finish a quest or farm enchants for the nightly raid or whatever due to bot obstruction?  I ask this because I only know one person in the world who actually plays this most-lucrative-entertainment-product-ever, and she said she's never seen a bot in WoW, much less had a problem.

In the WoW cancellation questionaire, one of the reasons for cancellation, as a subheading under 'Harassment', is "Botting". I don't think that would be in there unless it was an issue.

[Nebu]

How would M&B combat be unbottable in any way?

Every game is bottable to some degree, even games like poker.  The question is: would you go to vegas and have a computer play your cards for you at the poker table?  The key to limiting/eliminating bots is making it inefficient, undesirable, and dangerous.  Hell, make the game so fun and engaging that players won't want their computer to play the game for them.

I'm sure you can think of 1000 ways to accomplish this without me listing them.

[Stephen Zepp]

How come latency is a problem in MMORPGs but not (as much anyway) in other server hosted online games? Is it because when you play a MMORPG the speed of light becomes slower? Or because that's a BS excuse without any technical justification?

Uhh, simple math?
100,000 X (bandwidth per user) >>> 16 x (bandwidth per user).

Not to mention fd_select(addr) or whatever the syntax is (been a LONG time) x 100,000 takes a bit more time overall than x 50?

Bandwidth != latency.

If Margalis was responding to squirrel the point is is that the twitcher the gameplay the less suitable it is for MMOGs where latency can be unpredictable depending on the number of people around you at any given time. There are, of course, ways to cheat, like doing the calculations on the client instead of the server like PlanetSide does which saves you the round trip time but that wouldn't work if you had blocking and other such mechanics in the gameplay. Even in PS there are lots and lots of times I've fallen over dead after having gone behind cover thanks to their client-side hit detection system.

Edit: extra words

Thanks Trippy - i was a bit snarky but this is really what I was trying to say. Bandwidth as Stephen defines it compounds the issue(s). If you have a Mount & Blade type game you have to ensure 2 things:

1. - It can scale. This is what Stephen is referencing I believe. 64 + client connections is vastly different from 1500+ and persistancy means you can't trust the client software, particularly in DIKU/Achievement based MMOG's. We're just not there yet.

2. - Latency. Anyone remember the days of HPB's (high ping bastards?). Compounding issue 1 is the fact that in twitch/response based games people with faster gear and connections have inherent advantages over others - in orders of magnitude greater than the current advantages enjoyed by current hardware/network disparity.

End of the day - we're not at the point where un-trusted client/server MMOG's can have persistant 2000+ player worlds with Mount & Blade combat.

EDIT: Speeling

I admit that I was providing a smart ass answer, but a lot more goes into things than you'd think off the top of your head.

I will concede the point that bandwidth != latency (and in fact I wasn't making that point, although looking back at my short answer I can see how it seems like I did). Here's the logic chain that (hopefully) summarizes what I was getting at:

Assumptions

A: what players think is latency, lag, and performance/responsiveness is made up of a metric ton more factors than they are even aware of.
B: Many games even now still use client side hit detection, or at a minimum proxy/imposter based hit detection.
C: Non MMOG fps games are commonly peer to peer, while MMO games are by definition Client/server based.

Definitions:

1) Latency: single trip average time of information transfer between a sender and receiver. In some cases players assume round trip, but there is a hidden cost of server side processing time and possibly event saturation that is added in that makes it a non-network deterministic delay.
2) Lag: a user subjective broad based term used primarily to indicate an unsatsifactory state of blocked waiting for information. This is mostly used to indicate "network" lag, but in reality nowadays is primarily due to either cpu overload, or blocking data transfer from the hard drive. It does of course still apply in not a few network situations.
3) Responsiveness: how accurate and fast the control object (usually an avatar) is to user input.

First and foremost for the short answer I gave, the "game design myth" (my term) that ping is the end all be all to how good a player can be at an FPS is not nearly as supported as the end user experience indicates. I certainly agree that there is always going to be a delay between event generation (user input) on a client, and event resolution on a server, followed by authoritative updates from the server to each client, but ping is not quite as big a factor as it used to be, and certainly not measurable as a human experience at the 10-50 ms level.

What is measureable is a combination of several things:

--client side hit detection. In this scenario, a game is going to allow a user to actually directly apply a move input to their copy of the simulation, create a projectile, and determine it's game changing result (authoritative) before anyone else knows about the instantiation of the projectile. As has been noted in posts above, this combination of scenarios actually quite literally makes the worst ping player one of the most successful, since no one can respond to the projectile created before it's collision is resolved in many cases.

--server side hit detection (some MMO style)--in this case, instead of allowing the user input to directly create a game changing projectile and resolve it's collision on the client's time scale, the client generates a move event that is networked to the server that includes information about the pressing of the trigger. This information is then transmitted to the server (with the latency involved in the trip), processed at the network layer, then the event layer, then the processing loop, the result of that triggered event creates a projectile, which is then processed authoritatively in the game world. Additionally, that projectile will then be networked back to all clients in an observable range (normally called the scope), factoring in the single trip latency again, followed by network layer/event layer/processing loop delays at the receiving client's simulation, finally resulting in the new projectile on each of the client's simulations. Finally, this object will now do the following:

--interpolate between the last 2 updates sent by the server
--extrapolate if/when needed based on the last receieved update, and last received trend information (velocity, etc) if an update is "late"
--respond to a later server sent network event to indicate a collision/destruction based on the authoritative server simulation processing.

In this case (Server Authoritative, no client side hit detection, MMO style) ping rate --does-- matter, but it's my contention that the other factors involved in an MMO scale game over-ride simple ping rate/latency due to a lot of factors:

--total amount of bandwidth to process--even with huge server systems and amazing network infrastructure, it simply takes time to iterate over all packets that arrive. Since games are using UDP (which is a good thing), the games have to implement their own application level ACK/NACK of packets/events, so that takes processing time per arriving packet, as well as simply the total mass of events that must be processed each game loop cycle. These scale totally differently than the total number of events that are applicable in a small number of players style game,and even more so in a P2P networking architecture vs a Client/Server architecture.


This topic is hugely more complex than even what I've posted above--we haven't even begun to discuss the fact that by definition (and direclty due to latency) each and every simulation (server, and each client) are in different timelines, and in fact may be in multiple time lines at once (I'm talking semi-quantum branching type stuff here, at least as an analogy)--if an update takes 50 ms to get to Client A, and 500 MS to get to Client B, by definition the server, Client A, and Client B are all in different times. Integrating this dichotomy into a playable multi-player experience is a very involved process.

Networking is more than 40% of the Boot Camps I teach every few weeks, and it takes at a minimum 8-10 hours to cover most of the challenges....and my assertion originally was that ping is just one of the very important factors, and one that games themselves have by evolution mitigated extensively, and it's no longer as important as it used to be. In addition, the nature of MMO vs Small Multiplayer has quite a few factors that mitigate latency as well.

[Merusk]

How would M&B combat be unbottable in any way?

Every game is bottable to some degree, even games like poker.  The question is: would you go to vegas and have a computer play your cards for you at the poker table? 

You bet your ass I would, and I'd wind-up winning quite a bit more than losing.  This has been done before with very large degrees of success, which is why it's illegal to do so in casinos.

[Stephen Zepp]

PS: I think I've linked to this before, but Jason Booth (was at Turbine, not sure if he still is) wrote a very interesting article about dealing with latency from a game developer's perspective--specifically about the timeline dichotomy that latency introduces:

Fast, Secure Interactions in Latent Environments

It's more about AI than event handling, but pretty interesting.

[Nebu]

You bet your ass I would, and I'd wind-up winning quite a bit more than losing.  This has been done before with very large degrees of success, which is why it's illegal to do so in casinos.

I guess that's the difference between me and many gamers.  I play games to have fun.  While winning is fun, it's not the game to me. 

On a side note: computer programs are successful in poker largely because most people are bad poker players.  I doubt a computer could win the World Series of poker, but who knows... maybe I'm way off on this one.  Vegas hates anything that closes the odds gap between the player and the casino.  I think they are against this sort of thing in poker because it would chase the fish away.  The house gain in poker is all about the rake, so they don't have to play the odds. 

[Jayce]

 I doubt a computer could win the World Series of poker, but who knows... maybe I'm way off on this one.

If IBM could program a bot in 1997 to beat Gary Kasparov in chess, then I submit that it's possible to program a bot to beat any game.

[Venkman]

Stephen Zepp provided great thoughts in terms even I can understand :) But I did want to respond to one thing:

There will still be moments of lag and high-latency play just because of the way the internet works: a temporarily slow connection between the player and her ISP, slow connection between two random routers somewhere along the way to the game server, a denial of service attack in Hong Kong, a butterfly flaps its wings in Paris, France. If it is absolutely critical to the player's happiness and mental state that their character dodges incoming fire by getting behind a rock, one of these days they're going to be furious and unhappy. In a game where one death can make you lose more than an hour's worth of loot or experience, lag burps really ruin your day.

There is a limit to how much this can be soothed on the technical side. The rest is purely psychological, and requires a game designed to make failure less critical.

But this happens already, no? And probably with even MORE such moments because afaik in normal FPS competitive play,  it's still individuals that do the hosting, with all of the variables associated with p2p connections.

I only make this point because I wouldn't want MMOs to be held to any higher standard than which players in such relevant other genres already accept. Heck, I just want MMOs to reach those standards.

[pxib]

You bet your ass I would, and I'd wind-up winning quite a bit more than losing.  This has been done before with very large degrees of success, which is why it's illegal to do so in casinos.

I guess that's the difference between me and many gamers.  I play games to have fun.  While winning is fun, it's not the game to me.

This is precisely what I was talking about.

While Vegas tries to pretend that the reason folks go there is to have fun, most folks choose Vegas over, say, an equally expensive Carribean Cruise because they're dreaming of hitting a jackpot or a great game and coming home with a big pile of money. Sure there are great restaurants and shows, but there's a reason you've got to walk past three hundred metric tons of slot machine to get to them. Using a computer to play the game for you in Vegas is a great way to make a lot of money (and/or get in big trouble).

Using a computer to play for you in the basement with Lenny, Doofus, Carl, and Sharkey on Saturday night is just a great way to be an asshole. Using a bot may take away some of the fun of the game for you, but it has the potential to take away a LOT of the fun for your friends.

Comparing MMOGs to poker is also worthwhile in as much as most folks only play either for the shiney. The only way to win is to step away from the table. Until MMOGs are as much fun as games people enjoy even when there's no money on the table, they're going to have folks eager to cheat to get the phat virtual lewtz. And this is the internet, so ruining the fun of other folks in the game only feels as bad as cheating some random tourist in Vegas, rather than potentially losing friends around the table in the basement.

On a side note: computer programs are successful in poker largely because most people are bad poker players.  I doubt a computer could win the World Series of poker, but who knows... maybe I'm way off on this one.

Professionals don't make a living off the World Series of Poker. They make a living largely because most people are bad poker players. I imagine the computer would manage pretty well in the World Series, but just like any of the masters at those tables, whether it actually won would be a matter of luck.

There is a limit to how much this can be soothed on the technical side. The rest is purely psychological, and requires a game designed to make failure less critical.

But this happens already, no? And probably with even MORE such moments because afaik in normal FPS competitive play [...]

The problem with moments like this is qualitative, not quantitative. MMOs are held to a higher standard because they have been designed in a way that inspires a higher standard. They won't reach the FPS standard until their options are as bland, identical, and disposable as the ones in the average FPS... and in my mind that would be a step in the wrong direction.

[Stephen Zepp]

 I doubt a computer could win the World Series of poker, but who knows... maybe I'm way off on this one.

If IBM could program a bot in 1997 to beat Gary Kasparov in chess, then I submit that it's possible to program a bot to beat any game.

The problem here is that chess is a fully determinstic game, with a very simple rule set, very limited game world, and one that can be fully evolved to a win state each and every game turn given enough processing power.

I can't mathematically prove it, but I'd suggest that given the non-determinstic nature of mmo games, it's not 100% possible to create rule based AI that can handle any game.  Sure, we can create rule systems that can approach win states and adjust, but there's quite a bit more research needed to be "AIftw!".

[Nebu]

The problem here is that chess is a fully determinstic game, with a very simple rule set, very limited game world, and one that can be fully evolved to a win state each and every game turn given enough processing power.

I can't mathematically prove it, but I'd suggest that given the non-determinstic nature of mmo games, it's not 100% possible to create rule based AI that can handle any game.  Sure, we can create rule systems that can approach win states and adjust, but there's quite a bit more research needed to be "AIftw!".

That's what I was trying to emphasize in my poker analogy.  Thank you for wording it more concisely.

[Morat20]

[cool stuff]

I was rather shocked at the sheer amount of effort that goes into something as simple as synching clocks (well, it's easy if you have a master clock -- I'm talking getting four or five computers to agree on what time it is without a master). It's not surprising that something like an FPS shot cycle would be considerably more complex, especially if you cannot fully trust client-side information.

[Jayce]

 I doubt a computer could win the World Series of poker, but who knows... maybe I'm way off on this one.

If IBM could program a bot in 1997 to beat Gary Kasparov in chess, then I submit that it's possible to program a bot to beat any game.

The problem here is that chess is a fully determinstic game, with a very simple rule set, very limited game world, and one that can be fully evolved to a win state each and every game turn given enough processing power.

I can't mathematically prove it, but I'd suggest that given the non-determinstic nature of mmo games, it's not 100% possible to create rule based AI that can handle any game.  Sure, we can create rule systems that can approach win states and adjust, but there's quite a bit more research needed to be "AIftw!".

I see your point.  However, I think that if you restrict the scope to a certain activity in a certain part of a certain game world, things become a lot more deterministic.   

Also the exercise may be infeasible due to the amount of work it would take (I imagine IBM has more resources at its disposal than the WoWGLider guys) but it could still be possible.

[bhodi]

The other issue, Jayce, is the input medium. With chess, the boundaries of the problem and the input/output mechanism is very straightforward. It becomes a number-crunching game (if you want to brute force it) or sophisticated guessing code. Entire books have been written about it, but our little MMO bot is much easier. The problem shifts to the input/output mechanisms. A (sophisticated) botting program is going to need to create a sense-think-act feedback loop; somehow receive data from the target program and then act on it. Only the most rudimentary bots would work without it (UO Macroing programs, where you walk in a circle, press these keys at this interval.) MMOs are very tricky, because you have to intuit your sense data, and the 'act' part is relatively simple -- often just a keypress.

There are basically three ways of getting information in this regard. I'll talk about each in turn.

1) Sniff the data stream

This is traditionally been done with things like EQ radar, setting up some sort of proxy to grab the data coming into and leaving your computer. Because a lot of information is server side (to prevent the next issue), you may be able to get enough working information to use this as your input and create a bot feedback loop. I don't know that it's ever been done in practice, generally this is used for spoofing and additional information that gives the player an upper hand.

2) Look at the running program in memory

SoftICE or another debugger lets you look at and edit the game during runtime. This is the most versatile, the most common and the most dangerous (to developers). If the data can be found in running memory, and it isn't encrypted, it can be read. Often it can't be changed, as the server does sanity checking, but since everything that's displayed on your screen is somewhere in memory, this makes it easy to get at. It makes the most sense for WoWGlider and other programs to use this method to keep track of where your character is in the game world. Developers counter this by obfuscation and encryption, but it's essentially a cat-and-mouse game.

A simple example of using this on output as well as input would be that WoW program that let you fly around the world. The WoW devs made two mistakes - the first was to leave the unencrypted X, Y, Z location axis of the player sitting in memory, and the second was not to do any sanity checking server side of the client's reorted location. The program simply changed those values in memory, and the server didn't do any sanity checking, so the server accepted it and you were suddenly somewhere else. Do it very quickly, in small increments, and you can outrun the deeprun tram.

This was heavily used for farming reagents in the early days before WoWglider. It's also been used endlessly for 'trainers', programs that give you infinite lives and such in single player games. Remember GameShark? That falls in this category. So did some of the old memory resident cracks for dos games.

3) Read the game output

In this, the bot attempts to pretend it's a person and only recieves input that the game would normally output to the player. The most common things that fall into this category are the pallet matchers. I personally used this one for my ffxi fishbot, since it was the most easy. It works best on static text; With some clever coding you can do pretty well. You look for a specific color, or collection of pixels with a specific color or color range, and once you detect it, you can do something. With enough sophistication, you can read text , status boxes, things like that. It falls down when trying to create a bot that walks around, because you're pretty much unable to reconstruct location from a 3d rendering of the environment. You can cheat and use overhead maps (hit M, scan for your location, adjust) but you generally end up falling down pits and get 'stuck'. Remember botters walking in circle? It's becuase the act is out of sync with the sense. The program thinks you are in a different location than you actually are.


Let me give you a hypothetical example. I've been playing planetside recently, so I'm going to use that. Let's say that you want a combat edge over another player, and have decided you want a way to have it automatically both use a healthkit and turn on your personal shield when you take damage. Output on this is easy; all you need to do is transmit F1 and then F2 keypresses.

Option 1 seems a bit of overkill, but you'd set up a proxy server and look for indicies of someone hit you. I don't know what planetside data transmissions look like, but I suspect you could find health changes in there somewhere (an 'I hit you' packet, for example).

Option 2 is easier. Hunt around in memory for your total health. Watch that value. When it changes, you know you're hit.

Option 3 is just as easy. Because your health is in a fixed location in the screen, and is even represented by a bar, it's trivial to have a program watch pixel location X, Y, and do your keypress when it detects a change. (you'd still have to deal with loading screens, but you can just put in a 'don't change if it turns black' clause.

Now, that we've got the background, you can see that 'act' is only a third of your sense-think-act bot AI model. Of course, most agree that thinking is the more sophisticated trick. But we're diku here, it doesn't take brains - just time.