i'm aware that untraceableness can't be achieved in SR4 the way it can in SR3. s'why i made it clear that all my examples were SR3-related.

as far as there being 'something' that stops botnets from being created... there really isn't, and there really can't be. as long as the things that SR (any edition) says are possible remain possible, botnets remain possible. it's like saying that logs exist, and nails exist, but houses can't possibly exist. if the devs are handing out logs and nails, they cannot reasonably deny the existence of houses. they can unreasonably deny it, which is what's happened. what ought to be done is to take measures to ensure that such things aren't overly powerful. since the devs are, at this point, pretty much committed to pretending houses don't exist, you've got two options: houserule it, or simply ignore it.

i've chosen the (in my opinion much wiser) option of not using the SR4 ruleset at all. but as this problem exists in both SR3 and SR4, it's interesting to discuss it.

...I agree, but as I mentioned to Dashifen in a PM, I'd then find myself sitting all alone at the gaming table.

Ok, looks like we all know of Agent Smith, but almost none of us as GM's have had to deal with Agent Smith. So as a GM (You are the world after all), what do you do when Agent Smith rears his head?

You say your runners have 6 months of downtime, before their next run. You're not going to check every little detail on them, as long as its "by the book" they can do it. They will get no additional income during this time, but they have all the time necessary for training, repairing, building, etc... And your hacker comes to the tables saying he has Infinite + 1 rating 6 agents, loaded with every program. It's by the book, how do you handle it?

(This is kind of the runners creating the scenario and the GM having to react, reversal to how things usually go)

As I mentioned, redundant agents don't help. They're not creative. They can't help each other with teamwork tests. If two identical agents try to do the same thing, they do the exact same thing, so only one roll is made. Post-crash, the Matrix was designed from the most fundamental level to prohibit self-replicating code. Maybe two copies of the agent can't exist in the same place, they effectively merge.
Yes, this is a lot of handwavy semi-BS ways to just say "no", but in my mind that makes all the difference between having a consistent, successful game world, and just being an asshat.

I have no problem with them cracking an Agent. They can put a copy onto every team members' commlink. They can send them off on more tasks. But none of that it is particularly problematic. As long as you don't let multiple copies of the same agent contribute to one task there really isn't a problem. (This isn't against the rules, it's one of those things that's never explicitly forbidden or allowed, so I'm choosing to interpret it in such a way that my game works) Now if the hacker wanted to buy/code multiple agents, then I'd assume they're different and could work together like any other matrix entity. Agents are still cheaper than hackers, and powerful people could get large groups of them. But now they're no different from drones, which are also cheaper than samurai and easier to get. If there's something else that makes agents different than drones, then that difference can be addressed, otherwise one isn't more broken than the other.

i dunno, Moon-Hawk. i'm pretty sure i could cause some serious problems with infinite agents, even without teamwork tests. SR3 didn't have teamwork tests, after all, and infinite agents could still screw things up pretty badly there.

Hey is - if I get his meaning - not just talking teamwork tests. Multiple, identical agents doing the same thing in the same scene don´t help. You are never better off than with one agent (except you have multiple "lives" in cybercombat, which could still be a problem).

i'm not talking about identical agents doing the same thing in the same scene. i'm talking about having a gigantic network of agents across the Matrix helping me keep on top of any number of different activities.

I'd like an infinite number of helper agents to take care of trivial busy work for me.

I bet a lot of people would. Like... Everyone.

Sure they would. But no one's 1337 trivial busy work skillz are breaking the game, so who cares if everyone has agents taking care of their trivial busy work? The entire point of having computers is taking care of trivial busy work. And in the future a lot more trivial busy work will be handled by them. I don't see how this is a problem.

If a player wants to behave like a cyber-terrorist and send millions of agents out into the world causing mischief, then they can expect to be dealt with as a terrorist by the megacorps and governments. If those agents have a way of reporting back to the hacker, then there is a way for them to be traced. Maybe not with a conventional Trace program, but if information is getting back to the hacker somehow then that information can be followed. And if their task is just to go somewhere and do something, then it can't be anything all that bad, because one agent against an important system is going to get beat down. So if a player really wants to spray virtua-graffiti on a billion low-rated hosts, then he'll probably make the evening news and things will be straightened out by morning. If he keeps it up, he'll get the GOD on his ass.

Everyone would, wouldn't they? IC running on your comm, Agents monitoring your apartment, Agents sorting news for tidbits you'd enjoy, and they wouldn't be a problem until they started to hack, because they'd all have passcodes and be low rated to keep system load down.

Actually, I'm starting to think the problem is only that it'd be time consuming to deal with an Agent Smith, not actually difficult. Shutting down a node in a distributed system isn't very destructive, and shutting down multiple nodes would cause a back-up network to activate, like a cascade system which would either have a much higher firewall and so on, or just have hardwired limited functionality.

theoretically true, but in reality, it's a lot easier to hide a datatrail than it is to uncover it. and when you've got a million datatrails to follow, any one of which could be false...

and from the fluff i get the impression that in SR, google will hire out the use of agents...

so why run them all yourself when you an rent, or spoof them?

When you've got a million datatrails to follow, you use your million copies of your own agent to track each and every one of them. If we're allowing multiple copies to do different tasks, then each of those different datatrails is a unique task. And if a single one succeeds in finding the offending hacker and reporting back then he get's the anti-terrorist-freedom-force-of-liberty-and-doom.
High-profile shenanigans will get traced back to you, sooner or later.

Like I said, I don't see a problem with this. From time to time, hackers will do exactly what's described. Like I said, they'll probably cause minor to moderate mischief, and they'll be tracked down and arrested after the first or second time they do it. There is a pretty solid upper-bound on the amount of trouble a single lone agent can cause in a highly secure host. True, by the law of large numbers, occasionally one out of the million agents WILL succeed in cracking a single important host and cause some real trouble for someone somewhere, but those other 999,999 who fail are still going to be causing a ruckus and getting traced back to the hacker.

This really isn't any more achievable than a team of runners scoring lots of nerve gas and letting it loose on Seattle. They could do it, and it would be bad. But it would also be stupid and game-ending. The threat of cyber-terrorism exist in the world of SR: good. It happens occasionally: good. I still don't see the motivation for runners to do it in a normal game of Shadowrun, though.

I'd let them do it. They drop major systems throughout the world causing massive panic and destruction. Then, when people go back to analyze the situation, they'll find the agents that weren't able shut themselves down and erase their logs, etc. and one bright guy or gal will say "Hey, these are all the same Agent. Let's see if we can locate its owner." So they go around trying to determine who purchased it. They eventually track down a (fake) SIN from which the purchase was made or they find the online fixer that provided the code to someone else, but either way they will eventually find the hacker whose botnet brought down the world. I may not be today, it may not be tomorrow, but they will find the source.

Plus, it's a losing scenario in my games anyway. If an agent is crashed then the program code could be destroyed. If an agent is on a host which is dropped due to a DDOS or even just a full-on viral assault, if that's how you perceive agents working, then it's likely that as the security dudes try to get things back online, they're going to be restoring from previous backups and or reinstalling operating systems on wiped (or new) drives, etc. Thus, your massive botnet probably suffers catastrophic casualties in your first go around. Granted, only a moron would give up the source code for the original agent, so you can copy yourself another botnet given the time to do so, but if you try it again, I'm probably going to shorten the time it takes for the appropriate authorities to track you down based on the new information they have.

If nothing else, as your botnet storms (pun intended) out through the matrix, forensic analysis of the response of the various nodes should lead authorities toward finding a general point of nodal origin for the botnet. Now they've potentially got a physical location for the hacker that started it all or (more likely) they've got a location or a matrix address that they can analyze more heavily.

Regardless, I think, as others have said above, it's not that this isn't possible in the game world, it's that doing so is not subtle and is liable to get you caught and tried. The biggest problem is going to be rotting in a jail cell while the various jurisdictions determine who gets to put you on trial, assuming you last that long.

well, here's the thing--you keep talking about high-profile shenanigans as if that's the worst thing that could happen. it's not, and none of the successful high-end deckers (or otaku) i've played/seen played have gone in for that sort of thing. it's not about being able to invade high-end hosts with your agent army and wreaking havoc--it's about using your agent army to infiltrate your target's lower-end hosts and then waiting and watching for the security flaw that will let you slip into the high-end hosts unnoticed. it's about knowing the exact location of your every enemy because you've got hordes of snaky bots watching through every traffic camera they can get into. yes, if a hacker uses his millions of agents to do big, loud things, then he'll have the wrath of G.O.D. called down on him. but if he plays it smart, he'll be able to pull off big things very quietly with no risk and no exposure. no one will ever come after him because no one will ever think to look for him.

now, personally, i think that kind of game can be fun. but i don't think that sort of power should be accessible to anyone with a few points in cracking.

Okay, I think I see where you're going. Taking your first example: You can send your many agents out into weak hosts to gather intel on you to help you crack a high-end host. This takes some time, and there is non-zero risk of discovery, but we'll say the risk is small enough not to worry about. In return, I'd say you get a bonus to your Probing the Target roll, or maybe a reduced threshold to get yourself admin access. Hell, maybe your GM is nice and just gives you the admin passcode. But the rules for probing the target are so generous anyway that if you're halfway decent you could've gotten in anyway, and if you're really that incompetent and are just getting by with your agents, then the inevitable security hackers are going to, ahem, pwn joo.

The hordes of bots watching through every traffic camera is a pretty interesting idea. Of course, if it's this easy, then lots of people will do it, and if it gets done a lot, then occasionally the GOD will swoop down and track down a few of these Hackers just to make an example of them, so I've got to think that doing this ALL the time is a bad idea. But as an occasional strategy to track someone down, it's actually a really good idea. Of course it won't work if they're not on the street. Or disguised. But occasionally this could be used to track someone in a relatively low-security area down, sure. I think it's a clever idea, but I don't think it's a good example of breaking the game. I'd call this a use of agents, not an abuse.

I see your examples as clever uses of agents, not as big abuses.

Your argument is that the game designers don't have authority to interpret their own rules and design their own universe based on them. To an extent, that's valid. It's your game, you play it how you want to. But you don't have a basis for saying that your interpretation of what the rules must allow is the only valid one. There are a number of ways that a GM, interpreting relatively vague rules, can put the lid on botnets. First, you can go with the RaW that limits you to system x2 subscribers, and says that agents acting independently of you each count towards that limit. Second, you can rule that agents are not independent enough to form an effective botnet. Third, you could decide that matrix sercurity is tuned precisely towards ferreting out agents acting independently, and their lack of smarts makes them relatively easy to catch on the grand scale, so botnets get erradicated quickly. Fourth, you could simply decide "not in my sandbox," and tell your players that, for whatever reason, you'd rather play ShadowRun than AgentRun.

Those are all options, and each one is as valid as your decision that botnets should exist. The last one might not be too popular with players, but w/e. How the matrix ends up working in a campagin is up to the GM, not up to someone who declares unilaterally that there is Only One Way to read RAW. And the worst part of your argument about the One Way is that it's a really bad way. According to almost everyone who thinks that Agent Smith is mandated by the RAW also thinks that Agent Smith is stupid. They are interpreting the RAW in the least favorable way to "prove" that it's broken and "prove" that their total rewrites of the Matrix rules are "necessary." They haven't proved a thing, and it's very easy to fix percieved flaws in the rules with minor fixes. I don't begrudge you your way of doing things, just don't try to "prove" that it's the only correct way, because you can't.

P.S.
I'm aware that I may have seemed to be trying to prove that you were wrong druing this thread. To that, my answer is "a foolish consistency is the hobgoblin of little minds."

did anyone call my name?

clever use times infinity can quickly and easily become big, unstoppable abuses. that's the entire premise behind botnets, after all.


these are not interpretations, these are additions and houserules. you can't 'interpret' the rules to state that any agents subscribed to a commlink that is subscribed to your commlink magically count against the subscription limit on your commlink--that's an entirely new rule that doesn't exist in RAW. the rules as written work pretty much the way i've described. if a given gaming group doesn't want their game to work like that, additions and houserules are perfectly valid way of ensuring that their game remains enjoyable for them. but it doesn't change the way the RAW works.

You might want to double check the BBB before you make statements about what it does or does not say. I quote p.228:



I think that's some pretty clear RAW right there. You can have 12 agents running independently of you in the matrix, assuming you have a maxed out system of 6. It's not a house rule. It IS pretty arbitrary, and it doesn't make a lot of sense, except as a way of balancing agents. But I sure as hell didn't make it up

That doesn't stop anything. You can just buy *another commlink* Sure that adds an incremental unit cost, but its not exactly very high.

It only takes up one subscription slot on each additional commlink to link a chain together capable of running an Agent army.

you didn't make it up, but you have misinterpreted it. a subscription is an association with a specific device, according to SR4 page 212. so, like i said--you subscribe a bunch of agents to commlink A. max out the subscription list. then, subscribe commlink A to commlink B, which is your personal commlink. plant an agent on commlink A, give it an admin account on that commlink. you are now able to disseminate orders to your burgeoning agent army.

There are agents better then rating 6 and response chips better then 6 as well. An agent 6 is not the win and trying to take over the world with one of them will get you seriously insta-kacked. I mean it took teams of the elite of the elite shadowrunners. Teams. There were agent smiths running around on the internet destroying everything in its wake. I don't think the agent 6 program is comparable to the super virus that caused the second crash. I mean there are delta clinics that basically are almost impossible to find, and it's not as if a GM can never make up some fluff of his or her own from time to time.

Now who's house ruling? There are no rules about daisy chaining commlinks together. I have as much basis in saying you can't do that as you have in saying that I can. I get the feeling that Uniwred will answer the question in the negative. After all, I don't see the devs acting to encourage something which definitely destroys the cyberpunk one-on-one netdive feeling of the Shadowrun matrix.

the words 'daisy chain' never appear in the book, no. but as i said earlier: if you provide a person with some logs and nails, you can't complain when they use those to build a house. the rules for subscribing devices to other devices make it possible to daisy-chain commlinks; daisy-chaining commlinks (and other devices) is a logical application of the rules that have been presented.

You must have glossed over the part where I stated that letting the hacker/mage/sammy/puppy run their agents as 'second PC's in the net' is what really allows this sort of abuse, regardless.

Let me repeat it: Agents are NOT AIs. They are clever programs that can solve problems, but they are not... say it with me... AI.

IF the Mage is smart enough to tell me in advance... and this sort of thing I would write down/have them write down that their agent/drone is going to shoot/hack any security drone then I, as the GM, have the right to put in situations where such blanket orders get misunderstood.

See example of hostile extraction target getting popped because he fired his holdout at the troll sammy.

You can't complain that the mage having a hacker in the box steals the PC hacker's thunder if you aren't actually making that Mage control his HIB. Its like letting one player have two characters and everyone else has just one... of course that player gets to do twice as much.

Most people who deal with Agent Smith in SR4 never realize that is what they are looking at.

I mean sure, if you find yourself with a copy of your pet agent camping on a drone you are subscribed to handing out Medic actions to your main comlink every IP while you hackastack up in AR so that you can't take physical or stun and regenerate to full Matrix Health every round - you know you are doing wrong. But most of the time you start using Agent Smith you do so by slapping an Electronic Warfare autosoft on your drones. Or by detailing an agent to defend all of your devices. Or by having an Agent try to exploit you a space on a device. Or something else that sounds perfectly innocuous. Indeed, doing the things that Agents are supposedly actually for.

It is only after they start doing that that the player realizes that the technical skills that they spent 80 BP on are completely meaningless in the face of just loading a second agent. Indeed, meaningless in the face of the agent that they already have.

And then people cry.

-Frank

If you're the GM, I can't complain when you do that. But if I'm the GM, you better bet I can. I'm not saying daisy chaining is illogical, or that it can't be allowed. I'm saying, given the parade of horribles it allows, given how it destroys the Matrix theme as written, why would you?

There is only one reason to allow Agent Smith: so you can "prove" that the Matrix rules suck and justify rewriting them. But you need to recognize that Agent Smith is not part of the RAW. It is one interpretation of the RAW. It is an incredibly broken interpretation of the RAW. It is, therefore, not a GOOD interpretation of the RAW. You can criticize the devs for leaving it as a possibility instead of closing the loophole. But proving that the RAW could be abused does not prove that it's bad, it only proves that it's imperfect.

When you find a perfect game system, please let me know

The rules allow for the subscribing of devices together. Can you give me a quote that specifically prohibits commlinks from subscribing to other commlinks? As far as I can recall (not a Matrix expert), there are even rules for subscribing your whole team's commlinks together. If you can do that by the book (and it's kind of a major part of the game), how can you logically disallow a Hacker from setting up a commlink chain of his own, for whatever purpose?

i'm saying it's not something you have to allow, any more than you have to 'allow' people to fire guns when they've got a -2 penalty from cover, +2 from their smartlink, -4 from lighting, and +1 from a combat drug that enhances Agi (or whichever stat it is that applies to firearms). that exact action is never specifically covered by the rules, but the rules provides a framework in which that specific action is definitely possible and completely rules-legal, no interpretation required. same deal with daisy-chaining commlinks.

Yes, you as GM have the power to deliberately screw over player tactics that you don't know how to otherwise handle. But not the right. And rules that favor such tactics aren't any better than GMs who punish players for using them.

Basically, look at it this way. Can a mage run around with unlimited spirits? Or an otaku? No, because the rules address those situations. It is the fault of an incomplete ruleset, which did not address a situation it should have.

You are inferring a positive from a negative. That's a logical non sequitur. Just because it doesn't say you can't, doesn't mean you can. The book has no rules on how daisy chains are limited, if at all. Any limits I perscribe as GM are thus within my sole discretion. GMs are supposed to make calls based on the rules to make sure that a game is fun. You're saying that I can't make a ruling that would prevent the Matrix system from being broken because it wouldn't be logical. I say, it's illogical to allow players to break the system just because the book doesn't say they can't.



It's not punishing the players unless they go ahead and spend their money on dozens of commlinks and the facilities to massively copy agents, and I don't stop them. Telling players that they shouldn't expect to break the system is not a punishment. Telling the players that they should expect to use the Matrix as it's described in the theme, and not according to a loophole in the rules is not a punishment. It's called playing Shadowrun. It's not like Agent Smith is hard to handle - to fight massive botnets, the NPCs will just get their own massive botnets. It will be a Matrix system based around people sending swarms of agents at each other. But that's not part of the theme. That's just a questionable interpretation of the game mechanics in action. If I want to play the game according to the theme, and not according to the worst, most game-breaking interpretation of the rules, which is obviously an oversight, that's totally within my prerogative. And I think my players will agree that that's how the game should be played. They will not feel punished.

You're talking like a GM patching up an obvious loophole and oversight in the rules is some kind of crime. You seem to think that an RPGs rules ought to operate on their own, like a computer program. Put in player input, out comes a guaranteed output. That's not how it works. That's how MMOs work. In an MMO, if something has an unreasonable drop rate, you can go farm your heart out and get rich until they make an official code update. Since the game is not about anything but scoring lewtz, doing anything and everything necessary to get lewtz is acceptable. But Shadowrun is about more than lewtz. It's about a compelling story and an exciting play experience, regardless of whether you "win" or "lose." In Shadowrun, if there's an obvious oversight in the rules, you can patch it up with a quick ruling and play the game as it was intended. You can preserve the theme, and you can end up playing the game that you all expected to play. And have a good time doing it. And that's supposed to be a punishment for the players? Give me a break.



If you know of a perfect game system, I'd love to give it a try. You are criticizing the system not because it's bad, but because it's imperfect. You are cricitizing a system that, if interpreted in THE WORST WAY POSSIBLE, would not be fun at all. How about let's NOT use the silly loophole, and play the game as it was intended, and we'll all find that it's relatively pleasant. You're telling me that if there's a loophole, then I MUST let the players use it, or I'm a bad GM. Except if I do let the players use it, then it's a bad game. So... what? Let's all play D&D? Let's all play SR3? Because whooee, that was one perfect system... Loopholes are just that. Loopholes. Oversights. There is no rule that a GM can't fix them at will. In fact, there's a rule that a GM can fix whatever they want, whenever they want, for any reason. I cannot accept you guys criticizing a GM who would fix a dumb loophole to improve the fun of the game. It's incredibly obtuse to suggest that a GM trying to make the game fun is somehow punishing his players.

But agents are quite advanced. They are as smart as drones, and drones can deal with ambiguous orders like 'patrol this facility and use tasers on any unauthorized intruders, using team work with other drones' (That is what the book describes that 'dozens' of those beagle? Weasel? patrol drones can do.)

That requires

A) Being able to understand the concept of an intruder and segregate them from base personnel

B) Being able to formulate on their own the required tactics to get their taser onto a guy who might be a highly mobile shadowrunner

C) On the fly co-ordinate a plan with other drones to surround and take down a guy!

Other drones are smart enough to find, chase and destroy micro drones with minimal guidance. Drones can clearly take very basic orders 'attack that, protect this' and translate them into results. Hell, with the autodocs you can just dump a guy with a completely unspecified problem and shout 'fix him' at the drone, and the drone will get right on that.

That is some very smart stuff. Very smart indeed. If a pilot 4 drone is capable for formulating such complex ideas as teamwork and organisation, a mage can tell his agent with a free action via a DNI to 'shutdown that drone' and the agent will be able to formulate a plan of action and implement it with no further guidance.

Edit: In other news they are clearly artificial intelligences, and while they may not be able to pass the turing test they can understand concepts. The different between the AI that plays chess against me, and them is that they are much more complicated by several orders of magnitude. Deus can certainly pass the turing test and is several orders of magnitude more.

I wonder what a storm botnet would do to Yakuza operations... If it was turned to the express purpose of fucking the Yaks over

I'm not saying I don't know how to handle their tactics. I'm saying an Agent is not a PC and should not be treated like a PC. The player can not just simply handwave away commands to their agents or pilots. I won't let hackers do that either, for that matter. The book goes through some trouble to remind us that agents and pilots are, in fact, not AI's.

What I explicitly am saying is that the following situation should never come up at my, or anyone elses table if you are bitching about Agent Smith.

'GM: Ok, so the security team moves (rolls some dies, does some stuff)... and the Steel lynx pops out of the hatch in the corner and levels an MMG at you.

Mage Player: Okay, my agent is going to hack the drone while I blast the security team with a stunball.

Hacker player: But....I was gonna hack the drone...'


Instead it goes more like:

'Mage player: Okay, My agent is going to hack the drone while I blast...

GM: Hold up. How does your agent know to hack the Drone? Are you telling it to hack the drone?

Mage: Yeah... and then I blast...

GM: Whats with this blasting? Are you directing your agent to hack the Drone or are you blasting something? Make up your mind, dude.

Mage: Okay, fine then. My agent was preprogrammed to hack...

GM: What? Is that written down on your sheet? Did you tell it to scan the matrix for signs of hostile drones? Hook it up to a camera or something? Where did it magically get this command down?'

Though I am, of course, fine with a player taking the time to hook up a HIB to a camera and writting down all the actions the HIB will respond to 'automatically'. And I'm also fine with adding a complication if they are at all sloppy or lazy about it. That's the price of trying to get free extra actions, regardless.


Just like if the players take on an NPC 'merc' to fill a hole in the party. No player decides what he does. They give him commands, and then he does things according to his own loyalties and agendas... usually within the wants of the party, but I'd be ignoring a great source of 'action' to make him a blind, bland, robot who simply did exactly what he was told and nothing else.


Heck, lets address the simple 'hack the drone' command. The agent is a program, a clever program, but just a program. Without further guidance its typically going to do things as straight forward and logically as it can, regardless of what the player thinks is important.

Simply 'hack the drone' the agent, depending on its normal use, could start a slow, methodical hack, particularly if it is regularly used for stealthy hacks. If the player routinely uses it as a matrix sledge hammer, it will prefer direct, and probably loud hacks. Useful in combat, less so when you want to be stealthy. Thus, I feel it is justified to take a full action to get your HIB/Drone to do things exactly the way you want them to, round by round if necessary. Simply 'shouting' hack the drone gets an immediate response, but quite likely one not suitable to the need at hand... even if the drone is hacked 'this round', the agent hasn't been instructed to actually DO anything with the drone, thus it (the drone) will continue to follow ITS last instructions until the next time the Mage shouts at his HIB to 'Get it to stop shooting us!'.

Thus, a living hacking (PC, that is) is preferrable to a HIB even with equal dice pools. EVEN if that PC hacker does nothing but send out HIS HIB's to do stuff, its the dedication to MONITORING (using his actions) his agents (and rolling their dice...) that makes him the Hacker, not the source of his dice...

Make sense?


Also: I have been a long proponent that multiple agents on the same task are reduntant as they will essentially be doing the exact same thing at the same time. Hacking attempts are adaptive situations, not applications of brute force, else we'd have a stat measuring the battery power available to boost an attack. If password xyxzzz doesn't work, it doesn't matter how many Agents put it in at one time...


EDIT::: @ Cthulu: And we have programs that are fully capable of beating grand masters of chess, but couldn't win a checkers game against a five year old. One can presume that some programmer (probably NOT a mage) sat down with the corporate security drones and coded in some pretty extensive rules on what is, and is not, and intruder. Logically, one could then find gaps in the program based on how a Drone Pilot program recognizes 'Intruders'. That's what a Shadowrun team should do for legwork. Maybe it's 'wear official Lonestar uniforms', maybe there is a RFID badge on all authorized personnel, maybe there is a list with facial recognition software that their faces can be added too... You could even go so far as to abstract it to a single 'spoof' roll.

If you wanted to take it further, Shadowrunners could even 'play with' drones much as if they were 'bots' in a video game. Say ALL drones running team coordination software from Ebol Empar Corp will do three passes from the left followed by one to the right every time, giving knowledgeable teams (who have done their legwork) to know to expect that every fourth pass will be from the right instead of the left (giving them +2 dice against the drones...what have you)

I probably wouldn't make it that simple, but...

Chasing down targets is simply following sensor data at that point. Doesn't require that the drone be able to make 'critical' decisions like 'is this guy lying to me, his RFID tag says he's a freindly but he seems... suspicious.' or 'He said hack the lynx... hrm... maybe he wants me to turn its gun on those other humans over there? Yeah... that's it...'

Both those are outside the realm of a Non-AI program, no matter how clever.

this is not a case of inferring a positive from a negative. SR4 page 212 specifically states that you can subscribe a device to a specific network.

I repeat my point regarding agents: The real control on the functionality of this is the characters ability to give commands/interface with his devices. Just because he has 18 commlinks daisychained and running agents doesn't mean he gets the ability to set 180 agents on 180 different tasks.

Sure, you could set it up so all the agents got exactly the same orders at the same time, but I don't see that being that useful. Its like daisychaining calculators together to compute Pi to the last digit. They all are running the same calculation, and will return the exact same results at the same time. About the only use is error checking....

As I said, I am no Matrix expert, but I am sure I recall rules about how a team of shadowrunners can subscribe their commlinks together into a kind of network an as mfb said above, SR4 pg 212 has rules for subscribing devices to a network). Assuming I am not mistaken about this, then I am not doing anything like what you say. I am stating that there are rules for subscribing commlinks together.



So where do you draw the line on the number of commlinks able to be linked together? What arbitrary number do you choose as the cut-off point? Even with 5 (a common amount for a team of 'runners), a Hacker could string a lot of Agents out with that.

Before you get even more defensive over GM prerogative, I am not in any way attacking that. I am saying that arbitrary and non-sensible rulings should be kept to a minimum when using it. I am also in no way defending the viability of the Agent army. I am just saying that it would be better to find another way to limit Agents than to pick a random number out of a hat as a limit to linked devices.

Except that presumption is *wrong* because I can copy the pilot program from my honda jazz, install it in a Main Battle tank with no specialist knowledge of cars, tanks, guns or anything, (it's a software test) and the Jazz pilot program now can blow things up with the main gun.

I can buy a rating 6 pilot from amazon.com and install it in my car, my hunterkiller drones, my anti tank guided missle and my steel lynx with a software test. Said pilot program is now coping with grid guide, being a rocket propelled anti tank missle, knowing how to use an electric wielder on resisting targets and firing LMGs in suppressive fire, none of which are things that I the guy who installed it knows.

clearly whatever harness you just slapped the pilot in provides it with information for form decisions, but a pilot program is quite a powerful autonomous decision maker and can understand very difficult and often abstract concepts. The mage can definitely go 'shut down that drone!' as a free action and get on with blasting, or 'get that hacker out of my gear!' and try and frag him manually too.

But even if you disagree with all that, there is a defined test in the book for understand complex and ambiguous orders (stating that they can and do understand complex and ambiguous orders) and a rating 6 agent gets 12 dice towards that test. It is going to get a success. It can BUY three successes, unless you assume that agents get stressed or put under pressure

Edit: Additionally, you can transmit low density signal to nodes you are not subscribed to - placing phone calls or whatever. You can clearly have all agents in a conference call and just order them by shouting at them too.

actually, it does mean that you cand sent 180 agents on 180 tasks. it just takes 180 actions. you can prep your agent army with commands beforehand ("A, you use medic on my icon if it takes any damage; B, you attack any icon that attacks me; C, you brew me up a coffee with two creams and one sugar, and none of that soy shit this time..."). or you could play it slow and smart, building up complex sets of commands to achieve longer-range goals.

with a loose interpretation of the rules, you could also rank your agents such that lower-order agents can take commands from higher-order agents. then you can give the higher-order agents complex tasks, which they break apart into commands that they pass on to the lower-order agents.

I guess it all comes down to do you think 'shoot that guy' is an okay order for a drone. If it is, why isn't 'hack that drone' ?

I'm not sure how this refutes my point. Clearly, at some point, some programmer sat down and wrote the software, thus we can safely assume that some programmer has, at some point, defined common commands for that software like 'stop intruders'... even going so far as to define the word 'stop' and the word 'intruders'.

And I feel safe in assuming that a skilled dedicated computer/drone expert for a corporation is going to customize those basic commands and definitions within the legal limits of his software liscence... or not depending on his bosses.

I don't have my book handy to check you on the universality of software, but I feel reasonably confident that if I were to assess a dice pool penalty for taking a civilian car pilot and slapping it in a military tank, I wouldn't be cheating the players. That's the price of using pirated software outside the bounds of its design specs.

Then we get to the abiguous commands.

'Stop that drone' is pretty ambiguous to a HIB, I'll admit. With a high level agent, you are probably better off with that than the more specific, but less generally useful 'hack that drone'. Hacking is a specific action, the agent will happily bust into the firewall and set up a spoofed account, just like it was ordered.

Stop, however, requires the agent to determine, for itself, what constituites 'stoppage' and what, of the tools available, are best to perform that act. That may include hacking and spoofing an account with the authority to power down the drone, it may mean tracing the drones commands back to the security Rigger and frying his brain. Heck, I'd be tempted to put in a staged 'comprehension threshold' sort of thing. Threshold 3 means it groks the players basic intent. Threshold 4 means it understands that it's gotta do it 'RIGHT FREAKING NOW', threshold five means... well, at that point I'm forced to ask the player for more complex ideas... but that's fine. Five could be 'and give the boss control over it' or something.

But at no point does the ability to roll lots of dice comprehending ambiguous orders mean that the agent/pilot is performing tasks exactly as if the player was playing the agent... which seems to be the source of the problem, letting players run rampant with their 'free dice'...

I am confused. Are people trying to tell me I MUST allow Agent Smith botnets, even though they're broken? Even though they would ruin the game? Or are you just trying to say that it was a failure on the part of the devs not to explicitly disallow them? What, exactly, are you people trying to prove? I think I'm having trouble engaging you because I have no idea where you're going. If the rules allow Agent Smith... then what? Then we must gnash our teeth at the devs being imperfect? Then we must toss out the Matrix rules and completely rewrite them? I don't get it. What is the point of this argument?

I'm gonna risk a double post here:

I think ordering a drone to shoot that guy is fine. I also think it takes a good and proper action to do it right... particluarly if there are multiple 'that guys' available and the drone has multiple options on the shooting side. Which gun to use, etc.

Think about it: pointing a gun at someone and pulling the trigger is an action. How is 'pointing' to a guy and telling someone/thing else to pull the trigger significantly different?

Either way, you did that instead of blasting him with a stunball this IP

I don't think anyone is saying allowing multiple agents running around doing the bidding of a player unchecked is 'good'. But we have two different ways of addressing the problem, at least.

One group seems to want a new set of rules designed to prevent/account for Agent Smith abuses, thus keeping them under control.

the other group says that there are already rules in place that can be used to prevent Agent Smith from breaking your game.

Neither camp is very unified.

the only real argument i'm partaking in, in this thread, is whether or not the RAW allows Agent Smith. i believe i've shown that it does; what a given gaming group chooses to do with that is up to them. other than that, i'm exploring the possible ramifications Agent Smith can have on the game and on the game world, as well as the ramifications of houserules designed to curb Agent Smith.

come over to Oz

What if the guy uses a DNI and transmits orders radiopathically, logically that can come with significant associated information, like urgency, and a mental picture with a big red 'target here' on it, and if you disagree that can be a free action, then it can also come with the ability for the pilot program to actually read the players mind (literally!) and see who or what he's talking about when he said 'shoot the guy on the left' or 'shut down that drone' Kill all those guys resolves your dilemma incidentally

So no, in my view that didn't take any sort of IP and is probably quicker than opening your mouth. Shouting doesn't block you form stunballing, so logically a high bandwidth radiopatch shout shouldn't either. I'm just trying to remember what sort of action shouting is in SR though at the moment, I'm not with my books.

And nerfing the pilot program would be a house rule - if the guy makes the software test there are no penalties in the book. I'm not discussing house rules at this junction because that removes all basis for common discussion. Resolving ambiguous areas is fine though!

incidently, even if it does take a simple action, I'm not sure what that solves. It's going to take the drone many complex actions to accomplish that task, and the mage is free to act in the mean time.

Ok then. I don't really have a problem with that. I was responding to what people seemed to be saying: that not does RAW allow Agent Smith, it requires it, and it's wrong to put a stop to that kind of craziness, because GMs are bound to follow all interpretations of the RAW no matter how lame.

Really though, I think Unwired will close this loophole once and for all by giving us actual rules on how daisy chaining works, which will affirmatively prevent infinite chains.