Arsenal Kills Agent Smith, Ding Dong the Exploit is Dead... Options

[Nightwalker450]

Agent = Pilot = IC...

Pilot Upgrade listed in Arsenal... Special Skill Required "Hardware". So the only way to mass produce agents would be a facility creating Agent Smith chips. There is a hardware modification needed for agent/pilot so you can't just copy paste agents. This is enough to kill that one off for me.

This post has been edited by Redjack: Feb 15 2008, 06:58 PM

Reason for edit: Flagged as sr4

[Seven-7]

I see text saying Agents and IC are alike, I see text saying Agents and Pilots are alike.


But there is nothing saying that they are the same.

You can't hack things with a pilot.


As the ruling you've stated concerning Hardware only says 'Pilot', then no, you are incorrect.

[knasser]

Agent = Pilot = IC...

Pilot Upgrade listed in Arsenal... Special Skill Required "Hardware". So the only way to mass produce agents would be a facility creating Agent Smith chips. There is a hardware modification needed for agent/pilot so you can't just copy paste agents. This is enough to kill that one off for me.

Where does it say that? The text under Pilot upgrade states a Logic + Software test to replace (Arse, pg.105) but that is all. An index in the book would have helped. Pilot programs are explicitly stated as software in too many places to list. If there is something in Arsenal that says they're hardware based, I don't see it as "fixing" software copying so much as I see it introducing endless inconsistencies in the game. I do agree that Pilots = Agents = IC, more or less. I've allowed Pilot programs to be crashed by an intruding hacker.

[ixombie]

Yep, I agree. Pilots are not Agents.

There's an easier way to crush the 'mass producing agents' thing though. You don't need to rules lawyer to keep players from being retarded and trying to break the game. It's called the Monty Python style giant foot of GM power stepping on a stupid idea and making it go splat.

[FrankTrollman]

You know, even if you needed some hardware to run each iteration of Agent Smith it wouldn't really matter. If each Agent cost thousands or even tens of thousands of ¥ to get up and running it would still be cheaper than training or hiring the real high end hacker that a single Agent replaces.

The Agent Smith exploit isn't copying tens of thousands of copies of your Agents for free. The exploit is using Agents at all. The fact that you can copy large numbers of them for cheap or free is just rubbing that exploit like a salt lufa on a field of paper cuts.

-Frank

[ixombie]

Agents existed in SR3 you know... It's not like SR4 goes "omg and then there were agents!" They're more available, but they're every bit as real as they used to be... And I don't think exploiting the rules just exacerbates the problem. If you use agents as they were intended, i.e. no more of them than you can subscribe to your persona, they're just like "oh, those are cool." When you exploit the rules, anyone who can afford one rating 1 agent could infinitely copy the program and crash the whole matrix. There's a world of difference between those.

[mfb]

The Agent Smith exploit isn't copying tens of thousands of copies of your Agents for free. The exploit is using Agents at all. The fact that you can copy large numbers of them for cheap or free is just rubbing that exploit like a salt lufa on a field of paper cuts.

i don't agree with how you've worded that. the problem with agents is the numbers in which they can be used. in the current rules, a single hacker can effectively run infinite agents, thus multiplying his hacking ability by infinity. if each hacker--not each commlink, but each flesh-and-blood hacker--were limited to using only one agent, agents wouldn't be a problem because they would only double a given hacker's hacking ability. boiled down, agents = extra actions, and giving any character type infinite extra actions is obviously bad juju.

the fact that the Agent Smith problem is an actual problem in real-world network security only exacerbates things. the Storm botnet is a perfect example of what an SR hacker (or decker, to be honest) could do with an intelligent application of agents. so, according to a certain mode of thought, the Agent Smith technique should be allowed and should be game-breakingly powerful.

but of course, SR isn't designed to be a world in which Agent Smith exists. it's designed to be a world where individual hackers have limited power, and that means that individual hackers can't have infinite (or even large numbers of) actions--ie, limited numbers of agents.

Agents existed in SR3 you know... It's not like SR4 goes "omg and then there were agents!" They're more available, but they're every bit as real as they used to be... And I don't think exploiting the rules just exacerbates the problem.

yes, and they were just as exploitable in SR3. the difference is, in SR4 the rules are much easier to use, making the exploits easier to use.

[Kanada Ten]

Maybe we just need a mechanic to mimic the adaptation of anti-viral software to agent spamming.

[Buster]

There's an easier way to crush the 'mass producing agents' thing though. You don't need to rules lawyer to keep players from being retarded and trying to break the game. It's called the Monty Python style giant foot of GM power stepping on a stupid idea and making it go splat.

Why don't you just fix the rules instead of being a jerk? Playing by the rules is not "being retarded", it's just playing by the rules. You wouldn't drop a nuke on a player for firing more than one bullet would you? (gasp, powergamers call that "burst fire" and munchkins use "magazines" so that can fire lots of bursts! oh noes, kill the retardz!)

I know that fixing broken rules is hard and stuff, but it's a lot better than being a juvenile backstabber.

[Kyoto Kid]

Yep, I agree. Pilots are not Agents.

There's an easier way to crush the 'mass producing agents' thing though. You don't need to rules lawyer to keep players from being retarded and trying to break the game. It's called the Monty Python style giant foot of GM power stepping on a stupid idea and making it go splat.

..or you can use the Super Handy...

[Seven-7]

D20 Cyberscape handled Agents pretty well, they basically throw skills around to help the hacker. They can't fight, they die easily, and either they complete a test or they add +2 to yours. Woopdy.

[bishop186]

Why don't you just fix the rules instead of being a jerk? Playing by the rules is not "being retarded", it's just playing by the rules. You wouldn't drop a nuke on a player for firing more than one bullet would you? (gasp, powergamers call that "burst fire" and munchkins use "magazines" so that can fire lots of bursts! oh noes, kill the retardz!)

I know that fixing broken rules is hard and stuff, but it's a lot better than being a juvenile backstabber.

There's a limit and players should know it. If the limit is exceeded, it's very much within the GM's power to say "You know what? No." I don't see how that's juvenile, that's being the GM and keeping the game fair and fun for everyone. Seeing as Shadowrun gives the GM so much leeway and interpretation power, I think that this proves even more true in SR.

[Cthulhudreams]

i don't agree with how you've worded that. the problem with agents is the numbers in which they can be used. in the current rules, a single hacker can effectively run infinite agents, thus multiplying his hacking ability by infinity. if each hacker--not each commlink, but each flesh-and-blood hacker--were limited to using only one agent, agents wouldn't be a problem because they would only double a given hacker's hacking ability. boiled down, agents = extra actions, and giving any character type infinite extra actions is obviously bad juju.

Nah, I disagree. You can seriously lay down a rating 6 agent and then your zero cyber mage now adds '12 dice of hacking' to his repertoire, and whats more, he gets three additional IP he can only make hacking actions in. Sweet. 12 dice of hacking is no slouch.

That particular problem gets worse when you consider you can do it infinity times.

It's like having the electronics and cracking skill groups is totally arbitrary as a rigger too. Electronic warfare is an autosoft, just get like, 4, maybe 5 microdrones to hold onto your commlink. Now you have 16 dice in EW at a cost of not much. But you can still get 8 - no slouch! - with just one drone.

So there are two problems

A) You don't need humans to do matrix actions

B) Once you remove humans from the loop you can do whatever it is you want to do an arbitrary number of times.

There are two problems, and while the collective suck is great, in an RPG either of them blows individually.

Edit: And bishops186 exposes exactly the short sightedness here. If a mage has 12 dice in hacking for no reason, he *is* treading on the hackers toes, but using one agent is in no way 'unreasonable' Collective storytelling game here, if the mage can do the hackers schtick in addition to being an awesome mage, then someone is broken.

[hobgoblin]

Yep, I agree. Pilots are not Agents.

There's an easier way to crush the 'mass producing agents' thing though. You don't need to rules lawyer to keep players from being retarded and trying to break the game. It's called the Monty Python style giant foot of GM power stepping on a stupid idea and making it go splat.

around these parts its known as the orbital bowine launcher...

[Spike]

Nah, I disagree. You can seriously lay down a rating 6 agent and then your zero cyber mage now adds '12 dice of hacking' to his repertoire, and whats more, he gets three additional IP he can only make hacking actions in. Sweet. 12 dice of hacking is no slouch.

Now lemmee ask a couple of questions, iffin you don't mind:

What can a normal, fairly optimal but not munchkined to death, hacker get for HIS dice pool?

Lets assume that the 12 agent hack is a fairly standard thing to do. Presumably not only is your mage tossing 12 dice at hacking attempts, but whenever no Sysop is monitoring the system, there is undoubtedly a 12 dice counterhacking agent he is tossing against. Law of averages suggests... what? Ties go to the defender more or less, don't they? All around the world, runners and script kiddies and the like are regularly tossing the best agents they can get their hands on at various money targets, for fun, profit or just free porn. And, just like those agility tests every makes to avoid tripping over their own feet, we use the 'average event'... that is everyone succeeds and no one breaks their face open, and just hand wave it away. Thusly: while people DO undoubtedly succeed, more often than not, they fail. Moreever, just throwing more agent Smiths at the problem just provokes more agent smiths (infinity+1) to stop them in a wave of mutual annihilation. This happens EVERY DAY throughout the wireless. Thus the Mage can only roll his 12 dice agent (or whatever he has...) in matters of life or death, when you absolutely need to know if he bucked the odds that time.

Meanwhile: as agents are not truely AI's, and certainly lack telepathy, I would not let that player use their agent like it was a full time hacker without the Mage character sacrificing actions every so often to tell it what to do. That's as bad as buying a drone and putting in a pilot program and letting it run around like a second character controlled by the non-rigger. If the player really wants to have the drone/agent operating that autonomously then make them sit down and write out ALL the instructions, if/then statements and more they really did put into the drone/agent during downtime. Then feel free to have the agent/drone stop whenever they hit an unclear command or do stupid things like shooting the target of a hostile extraction because the player wrote 'if the party/drone/agent is attacked, shoot back' and the extraction target, with his light pistol and dice pool of three took a potshot at the troll.

Because, that is what 'not an AI' really means. The mage is gonna miss spellcasting time every time his Agent asks him 'what now boss?' in that chipper tone that is garaunteed to annoy after the third time...

[mfb]

Nah, I disagree. You can seriously lay down a rating 6 agent and then your zero cyber mage now adds '12 dice of hacking' to his repertoire, and whats more, he gets three additional IP he can only make hacking actions in. Sweet. 12 dice of hacking is no slouch.

good point. i'll amend my stance to "if each hacker--not each commlink, but each flesh-and-blood hacker--were limited to using only one agent whose abilities cannot exceed the hacker's own, agents wouldn't be a problem because they would only double a given hacker's hacking ability."

to be honest, one the best solutions i've seen to the agent problem (as it applies in almost any game system] is Frank's concept of simply doing away with the entire concept of agents as discrete entities in the rules. simply allow hackers greater ability to act in multiple systems, and ascribe that ability to the use of agents. the problem with that is that it deviates from how drones and spirits work. the difference from drones is especially problematic, since the only real difference between agents and drones is that one of them is tied to hardware. i suppose it wouldn't be too hard to explain that away.

another possibility might be to run agents more like spirits--use the summoning and binding rules. you start with the assumption that the anti-malware software of 2070 is rabid, and can quickly find and destroy the persona code of any agent (personas make no goddamn sense, but they exist in SR and the blinding pain goes away if you stop thinking about them, so we may as well use them). when you whip up an agent, you're effectively installing viral code on a bunch of random hosts, stealing those hosts' processing power to run your agent's persona. each time an agent acts, more of the viral code which supports its persona is found and destroyed by the random hosts. et voila, summoned agents who can perform limited numbers of tasks before being wiped out, and whose abilities are limited by the hacker's own--just like spirits and favors. binding would involve taking the time to hack a few well-chosen hosts and install the persona-viruses more sneakily. the two big problems with this would be a) figuring out an explanation for why a given hacker can only seed a limited number of agents, and b) figuring out why agents can't seed other agents. which are, y'know, major problems, but at least the structure is there for limiting agents in both numbers and skill.

[Kanada Ten]

another possibility might be to run agents more like spirits--use the summoning and binding rules. you start with the assumption that the anti-malware software of 2070 is rabid, and can quickly find and destroy the persona code of any agent (personas make no goddamn sense, but they exist in SR and the blinding pain goes away if you stop thinking about them, so we may as well use them). when you whip up an agent, you're effectively installing viral code on a bunch of random hosts, stealing those hosts' processing power to run your agent's persona. each time an agent acts, more of the viral code which supports its persona is found and destroyed by the random hosts. et voila, summoned agents who can perform limited numbers of tasks before being wiped out, and whose abilities are limited by the hacker's own--just like spirits and favors. binding would involve taking the time to hack a few well-chosen hosts and install the persona-viruses more sneakily. the two big problems with this would be a) figuring out an explanation for why a given hacker can only seed a limited number of agents, and b) figuring out why agents can't seed other agents. which are, y'know, major problems, but at least the structure is there for limiting agents in both numbers and skill.

What if you let hacker or agent seed as many agents as desired, but the number of "services" is for the total set, due to anti-malware data sharing?

[Blade]

As long as you can have agents, you can have this problem.
Well, as long as you can have hacking and don't use some kind of deus ex machina to prevent it, you can have a similar problem.

But there are a lot of ways to avoid it or deal with it, some better than others, none of them - as far as I can tell - safe from rebuttal and none of them canon. It's up to the GM to choose the one that's best for his game.

[mfb]

What if you let hacker or agent seed as many agents as desired, but the number of "services" is for the total set, due to anti-malware data sharing?

nice.

[masterofm]

If you have r6 agents and are putting it onto a r6 database it will also have the same exact things for fighting you, except for the fact they they already have more agents loaded onto the system then you do. Also if you put a r6 agent into a r4 chip the agent is now treated as an r4 agent (as it is having to use the runtime on their system to fight them.) If you start loading too many agents onto a system it starts dropping the response time of the computer and eventually with an attack like that they will probably just shut down the individual node you are sending agents into. Only sending agents into a system is basically a stalemate, but a hacker will be able to use his dice regardless. If a node is under heavy agent attack all anyone needs to do is just remove/delete the data on their computer, and shut down the computer. They could also just pull the theoretical plug on their wireless and totally cut the hacker off. Agent Smiths are not the win they are just really nice when someone tries to bust into your shit.

[hobgoblin]

the thing is that some see a node shutting down as a win for the hacker...

this because such a act will probably disrupt whatever remote service that the node provides.

say you agent-storm a node controlling a surgery drone.
or hell, just your average combat drone or its control node.

[Cthulhudreams]

Capping the number of services agents gets just locks you out of the infinite loopiness thing- and it's good, because option B is explicitly caused by the fact that agents and the rest of the matrix run at two different levels of abstraction. But it still doesn't fix that anyone can become a hacker over the counter.

Limiting agents to the hackers skill size and its number of services at the same time just looks like madness. It makes all the dice pools bigger... for what reason? I suppose with half size dicepools, he can hack two nodes at once. But that would presumeably loose to any IC that could ever pose a problem for the double dice pool methodology - assuming of course that IC, being an agent without a hacker, can actually work in any way at all.


@Spike

Or, you know, as its an RPG, and the players are supposed to be part of a team, with different abilities to combine to form narrative threads, and I can tell the mage player to go stick it, and ban agents, just as I'd tell the hacker to get stuffed if he wanted an ally spirit, without subjecting himself to all the rigmarole normally required to summon one.

Also your analogy fails as the typical security drone (I imagine that to be a prime target too for mage hackers with agents) can only support 8 dice of counter hacking due to the system & response 4 of a steel lynx. Boned.

The r6/r4 chip thing being talked about before also doesn't work because plenty of hacking attacks can be launched without being loaded on. Exploit springs to mind. So does spoofing as the registered owner and sending the shut down command

@mfb

I like franks' solution too, as you can probably guess. Its easy to draw more of distinction between drones and agents - you ban all the autosofts that bestow a hacking skill, like EW and then define it as impossible for IC to ever leave the node it is loaded on ever under any circumstances what so ever. It can only attack things connected to that node. To further emphasize the break with the past, I'd change hacking to attribute + skill limited by program.

[Ryu]

I really like to apply Franks braincomputing idea to agents. Run them like spirits, each agent in use after the first applies a -2 distraction modifier. Agents roll rating+user skill.

[hobgoblin]

i just have to ask, have anyone seen someone attempt to deploy agent smith tactics in game?

[ixombie]

Why don't you just fix the rules instead of being a jerk? Playing by the rules is not "being retarded", it's just playing by the rules. You wouldn't drop a nuke on a player for firing more than one bullet would you? (gasp, powergamers call that "burst fire" and munchkins use "magazines" so that can fire lots of bursts! oh noes, kill the retardz!)

I know that fixing broken rules is hard and stuff, but it's a lot better than being a juvenile backstabber.

I have to take exception to that. What, exactly, is juvenile here? Is it saying "I don't want to play the game, so I exploit the rules and make it unplayable," or is it saying, "I'm the GM, and I will not allow you to willfully make my game unplayable." Yeah, I didn't use polite language in my original post, but I wasn't acting as a GM. I was not describing how I would call my players retards and stomp on their ideas like an asshole. I was just using colorful language to describe what I thought was the best solution to the problem.

You're comparing 'infinite agents' to 'firing more than one bullet.' One is clearly allowed by the rules, and makes perfect sense. One is based on a number of questionable interpretations of the rules, and has no place in any game. Can you guess which is which? Can you see the difference between the two?

Saying "no" is a valid way for the GM to fix broken rules. It is not backstabbing to tell a player they can't go ahead and wreck the game just because the rules in the rulebook, read very loosely, would maybe allow it. Maybe it would be backstabbing to let them spend 250k on agents at chargen and then come back with "haw haw, tricked you, your character is worthless." But exercising GM discretion to preserve the game against a blatantly obvious rules exploit? How dastardly.