So here's the thing. You have this node with let's say, for sake of argument, all 4's. It's got all sorts of valuable paydata on it, so of course you want to protect it. So you buy an IC program, as well as attack, armor, analyze, trace, all that good stuff, all at rating 4. You want this IC to start defending, so you boot up the IC program, boot up armor and analyze. This is three programs running, all good. A system with all 4's should be a bit challenging, but nothing a good hacker can't handle, right?
You're paranoid, so you set the IC, an agent, to automatically run analyze every so often to scan for intruders. How often can it do this? As an agent, it gets three IP's in the matrix, so presumably it can run analyze every IP. Since hackers oppose with hacking+stealth, it seems like no hacker would be able to go undetected for longer than a combat turn or two.
Now comes the really iffy part. You have this IC running, along with analyze and attack. You want to have a lot of security in case someone breaks in though, so you hit ctrl+c and ctrl+v and poof! New IC program! So now you have two security agents! Let's further say you don't mind about reducing the response a bit, so poof, poof, poof, now you have five agents, two other programs, making the node's response 3, but who cares! You have five agents running, with analyze actions happening once every 0.2 seconds on average. And when something does get detected, heavens help it, it's gonna get slammed with attack actions out the wazoo. Obviously, if the node has even better hardware, it gets straight up ridiculous.
My biggest question is, if people have IC, why don't they just always run as many copies of it as they can without reducing response? When one gets crashed, or if a program gets crashed, why doesn't the node immediately restart the program as a complex action? Why doesn't a corporation get their resident software guy to code one rating 6 IC and then copy it a bajillion times for all the security needs?
Thanks for the help, all.
~Gelare
Full Version: IC Armies
also known as the "agent smith" problem.
look around, there are at least a dozen threads on this exact topic.
look around, there are at least a dozen threads on this exact topic.
Even better, you can run the IC on seperate hardware, and as long as you can keep buying the hardware, you can have that many IC run on the node.
As far as the analyze actions, think of it like infiltration. The IC set their analyze program to run on that node, and it alerts them when/if it finds anything. Each IC's analyze gets 1 shot to detect the hacker when he connects, if they all fail, he's fine and has snuck into the node.
As far as the analyze actions, think of it like infiltration. The IC set their analyze program to run on that node, and it alerts them when/if it finds anything. Each IC's analyze gets 1 shot to detect the hacker when he connects, if they all fail, he's fine and has snuck into the node.
I also do another Analyze whenever they do something out of the ordinary(GM Fiat). Usually treat all the "out of the ordinary" actions as an extended test (Stealthx2).
Chris
Chris
Here's some ideas I had for limiting an agent smith army:
For an agent smith army using Analyze, I would say that the army collectively gets only one roll to notice an intruder in their node. 100 carbon copy agents is not the same thing as 100 security hackers or even 100 separately programmed agents. Agents don't learn, so carbon copy agents do not have distinct experiences that allow them to notice anything that their clones wouldn't notice. However, once the alarm is raised, it's battle royale and the intruder will have to deal with all the agents in the army.
Regarding an army of attackers, I would use the D&D adjacent squares rule and say that no more than 8 opponents can gang up on each target. Of course after one attacker falls, another takes it's place the next IP, so the army will still win eventually. The only advantage this adjacent square rule gives the guy at the bottom of the dog pile is that he might have a chance to jackout (if he's the intruder) or raise the alarm (if he's the defender) before he gets the crap kicked out of him.
Also according to the RAW, the agents don't degrade the nodes they are in, they only degrade the nodes they originate in (like a hacker's commlink). You could houserule that any hacker or agent always degrades the node he is in (or is attempting to be in) rather than the commlink or node he is originating from. This way an agent smith army would just seize up any node they tried to enter. I would also say that Firewalls are smart enough to prevent enough programs/agents/hackers from entering a node that would reduce its rating to zero. This way firewalls are smart enough to prevent a ddos attack.
Other than that, I haven't seen any good rules for limiting an agent smith army (for hackers or defenders).
For an agent smith army using Analyze, I would say that the army collectively gets only one roll to notice an intruder in their node. 100 carbon copy agents is not the same thing as 100 security hackers or even 100 separately programmed agents. Agents don't learn, so carbon copy agents do not have distinct experiences that allow them to notice anything that their clones wouldn't notice. However, once the alarm is raised, it's battle royale and the intruder will have to deal with all the agents in the army.
Regarding an army of attackers, I would use the D&D adjacent squares rule and say that no more than 8 opponents can gang up on each target. Of course after one attacker falls, another takes it's place the next IP, so the army will still win eventually. The only advantage this adjacent square rule gives the guy at the bottom of the dog pile is that he might have a chance to jackout (if he's the intruder) or raise the alarm (if he's the defender) before he gets the crap kicked out of him.
Also according to the RAW, the agents don't degrade the nodes they are in, they only degrade the nodes they originate in (like a hacker's commlink). You could houserule that any hacker or agent always degrades the node he is in (or is attempting to be in) rather than the commlink or node he is originating from. This way an agent smith army would just seize up any node they tried to enter. I would also say that Firewalls are smart enough to prevent enough programs/agents/hackers from entering a node that would reduce its rating to zero. This way firewalls are smart enough to prevent a ddos attack.
Other than that, I haven't seen any good rules for limiting an agent smith army (for hackers or defenders).
| QUOTE (Buster) |
| Regarding an army of attackers, I would use the D&D adjacent squares rule and say that no more than 8 opponents can gang up on each target. Of course after one attacker falls, another takes it's place the next IP, so the army will still win eventually. The only advantage this adjacent square rule gives the guy at the bottom of the dog pile is that he might have a chance to jackout (if he's the intruder) or raise the alarm (if he's the defender) before he gets the crap kicked out of him. |
You're assuming that the matrix has something like gravity that keeps one from attacking from above, or that close proximity is necessary to attack the hacker, or that there is a proximity in the matrix. Everything we see is just a metaphor for the code that is happening. If 100 agent smiths are attacking you you could just as easily have them appear as a swarm of bees all hitting you with spears, or they could shrink to small sizes and be attacking you with spears or whatever. If you are in the same node there should be no limit as to the number of people who can attack you based on space surrounding you.
And, just to make things worse, I've routinely used flocks of birds or swarms of insects to represent a single icon rather than a group of many icons as odinson does above. In other words, the metaphor of any given system doesn't have to conform to any known physical laws or commonly held beliefs about the so-called Real World 
I'm not assuming anything, I'm just trying to come up with some reasonable house rules to fix a big hole in the game rules.
Personally, if I were a GM, I would start with my last suggestion I posted above and change system degradation to be localized to the node your icon is in (or is attempting to be in) and make firewalls smart enough to limit traffic to prevent excessive system degradation.
Personally, if I were a GM, I would start with my last suggestion I posted above and change system degradation to be localized to the node your icon is in (or is attempting to be in) and make firewalls smart enough to limit traffic to prevent excessive system degradation.
| QUOTE (Dashifen) |
| And, just to make things worse, I've routinely used flocks of birds or swarms of insects to represent a single icon rather than a group of many icons as odinson does above. In other words, the metaphor of any given system doesn't have to conform to any known physical laws or commonly held beliefs about the so-called Real World |
So you're saying that corporations were nice enough to make hacking their system exactly like playing a video game but they don't limit you to tactical grid movement?
Does party ice still exist?
Also, it's troubled me for the longest time, that any good decker should have had a second deck daisy-chained in to run agents/ice. You can, obviously, do this even more easily with commlinks.
For starters, you can build a nice firewall commlink that stops traces dead with IC designed to do nothing but sit there, spoof the data trail between it and you (in case anything gets past it), and analyze all signals coming through to decide if that's really you or something that's not supposed to be talking to your primary commlink, and then drops an anvil on the intruder if it doesn't like the response.
It's a hop, skip, and jump from that to a hundred commlinks loaded with agents running interference against dozens of nodes to distract every defensive hacker in the corp while you much around in the backwater system you sleazed an account for last week.
Also, it's troubled me for the longest time, that any good decker should have had a second deck daisy-chained in to run agents/ice. You can, obviously, do this even more easily with commlinks.
For starters, you can build a nice firewall commlink that stops traces dead with IC designed to do nothing but sit there, spoof the data trail between it and you (in case anything gets past it), and analyze all signals coming through to decide if that's really you or something that's not supposed to be talking to your primary commlink, and then drops an anvil on the intruder if it doesn't like the response.
It's a hop, skip, and jump from that to a hundred commlinks loaded with agents running interference against dozens of nodes to distract every defensive hacker in the corp while you much around in the backwater system you sleazed an account for last week.
| QUOTE (Big D) |
| Does party ice still exist? |
...I believe you can still get it at the corner Stuffer Shack. Right next to the 48-Paks of Spud Lite.
McDonald's sells it now, 2 bags for one fitty.
I don't think the rules have a "hole" per se because the rules leave it up to the GM to decide what can or can't be done in the Matrix. The problem when you try to "rule up" everything is that people like us keep finding holes in the rules ("Why can't you just..."). I like the SR4 Matrix rules: GM decides.
Personally, Agents and IC cannot be copied in my world. They are a wholly unique program unto themselves. Just try to make things as "reasonable" as possible. Perhaps corporate systems don't run 100's of Agents with Analyze because that affects the performance of their business applications, what with their systems pounding an Analyze program every third of a second. I like the new Matrix rules as they are flexible and fast. Running Matrix stuff in the previous editions was just a slow, painful exercise of rule-flipping.
Keep in mind the "feel" that you are trying to convey for your SR games. Are they tense, stealth-based, "caper" type runs? Are they fast-based, shoot-em-up action? Whatever your style, tailor your Matrix scenes to match, but try not to get bogged down in the rules. It's called Shadowrun, not "futuristic computer dice simulator." Try to keep Matrix stuff "in line" with the "in the flesh" part of your games and don't get bogged down in the details.
Personally, Agents and IC cannot be copied in my world. They are a wholly unique program unto themselves. Just try to make things as "reasonable" as possible. Perhaps corporate systems don't run 100's of Agents with Analyze because that affects the performance of their business applications, what with their systems pounding an Analyze program every third of a second. I like the new Matrix rules as they are flexible and fast. Running Matrix stuff in the previous editions was just a slow, painful exercise of rule-flipping.
Keep in mind the "feel" that you are trying to convey for your SR games. Are they tense, stealth-based, "caper" type runs? Are they fast-based, shoot-em-up action? Whatever your style, tailor your Matrix scenes to match, but try not to get bogged down in the rules. It's called Shadowrun, not "futuristic computer dice simulator." Try to keep Matrix stuff "in line" with the "in the flesh" part of your games and don't get bogged down in the details.
I thought of another way to answer this question. When designing Matrix systems and security, think of them more as real-world security systems.
For example, I haven't seen anyone post the "hole" that a corporation could fill their buildings with a security guard every 10 feet who asks, "What are you doing?" every 3 seconds. There's nothing in the rules saying a corp can't do this, so why wouldn't they? Those are the same reasons they don't fill their Matrix systems with legions of IC.
For example, I haven't seen anyone post the "hole" that a corporation could fill their buildings with a security guard every 10 feet who asks, "What are you doing?" every 3 seconds. There's nothing in the rules saying a corp can't do this, so why wouldn't they? Those are the same reasons they don't fill their Matrix systems with legions of IC.
Agents are free if you have the source code, security guards want salaries and dental plans.
there are two ways at looking at the problem, and both have issues.
1. agents have to run on the node they are interacting with. issue, what happens when a agnet is told to do a net search? issue, can a load of agents bum rush a node into shutting down?
2. agents can run on a different node then its interacting with. issue, agent smith army.
1. agents have to run on the node they are interacting with. issue, what happens when a agnet is told to do a net search? issue, can a load of agents bum rush a node into shutting down?
2. agents can run on a different node then its interacting with. issue, agent smith army.
| QUOTE (odinson) | ||
You're assuming that the matrix has something like gravity that keeps one from attacking from above, or that close proximity is necessary to attack the hacker, or that there is a proximity in the matrix. Everything we see is just a metaphor for the code that is happening. If 100 agent smiths are attacking you you could just as easily have them appear as a swarm of bees all hitting you with spears, or they could shrink to small sizes and be attacking you with spears or whatever. If you are in the same node there should be no limit as to the number of people who can attack you based on space surrounding you. |
Or the agents could just turn clipping off. Go way back to the DOOM era of games. You could be surrounded by imps (turn on god mode) and they'll all be partially intersecting each other, but it doesn't matter one bit, and they're all hitting you equally well. No clipping ftw.
| QUOTE (hobgoblin) |
| there are two ways at looking at the problem, and both have issues. 1. agents have to run on the node they are interacting with. issue, what happens when a agnet is told to do a net search? issue, can a load of agents bum rush a node into shutting down? |
Actually the first one is not really an issue.
If the agent is told to do a net search, he'll go to the nodes to find information. You ask him to get a flight ticket for you and he'll hop to the nodes of each airlines company to gather the required data.
Communication between you and the agent should still be possible because of the subscription, which should allow you to send and receive data to and from your agent.
I've seen people work on that topic. Today the main advantage is that you can send your mobile agent from your mobile phone (a few Kb to send) and then you can disconnect your phone, turn it off and go wherever you want. When you turn your phone back on and reconnect it to the internet, you get back your agent with all the data you need.
Another advantage is that the agent runs directly on the airline's server, which makes it easier and faster to query the database.
Actually, in Shadowrun both advantages aren't that important anymore: your commlink is always online, and remote queries of database should not be a problem either. But there are still some other advantages to using agents this way.
its a issue for one simple reason, that agents have their response based on the node they are running on. hit to weak a node with to big a payload and the agent basically have no response. then there is some interesting effects of the faq ruling that say that any programs running inside a agent count towards a nodes total.
there is also the question, does a visiting agent take up "resources" like a resident agent does?
there is also the question, does a visiting agent take up "resources" like a resident agent does?
I get your point.
I have answers, but they are related to the way I deal with the Matrix which, while consistent with the few rules and fluff we have (as much as it's possible), is a specific way to consider the Matrix, so they'd probably be incompatible with your or other's take on the Matrix.
I might develop my whole point of view further in a handbook if anyone's interested, but I doubt that anyone is and I don't want to spend too long working on something that nobody cares about.
I have answers, but they are related to the way I deal with the Matrix which, while consistent with the few rules and fluff we have (as much as it's possible), is a specific way to consider the Matrix, so they'd probably be incompatible with your or other's take on the Matrix.
I might develop my whole point of view further in a handbook if anyone's interested, but I doubt that anyone is and I don't want to spend too long working on something that nobody cares about.
| QUOTE |
| Actually the first one is not really an issue. If the agent is told to do a net search, he'll go to the nodes to find information. You ask him to get a flight ticket for you and he'll hop to the nodes of each airlines company to gather the required data. Communication between you and the agent should still be possible because of the subscription, which should allow you to send and receive data to and from your agent. |
If you can still interact with the agent through subscription, how can it not subscribe to the airlines nodes to do his thing ? Either the agent has to be inside a node to interact with it or any icons therein, making you lose contact as soon as you ask him to go check the stuffer shack menu, or he doesn't. In which case he can jolly well stay inside your super beefed up commlink and do his thing from there.
Can't have your cake and eat it too.
Yes, you can run the Browse program on your commlink and have it search the Matrix this way.
Searching with the agent is another way of searching, which has its advantages (for example a node can restrict access to its database to local programs. In that case, your Browse program won't be able to accses it, but an agent will) and its drawbacks.
Searching with the agent is another way of searching, which has its advantages (for example a node can restrict access to its database to local programs. In that case, your Browse program won't be able to accses it, but an agent will) and its drawbacks.
| QUOTE (Sma) | ||
If you can still interact with the agent through subscription, how can it not subscribe to the airlines nodes to do his thing ? Either the agent has to be inside a node to interact with it or any icons therein, making you lose contact as soon as you ask him to go check the stuffer shack menu, or he doesn't. In which case he can jolly well stay inside your super beefed up commlink and do his thing from there. Can't have your cake and eat it too. |
because agents dont work that way?
as in, they can only subscribe to whatever it is thats giving them orders?
hmm, i wonder if it becomes simpler if one think of agents as software drones.
and i see blade have brought up the browse program. the one thing that may make a agent able to sit anywhere and search the matrix. as in, one must not forget that a agent use the very same tools as a hacker use to get things done.
bah, we can go around in circles about this for ages.
| QUOTE |
| Yes, you can run the Browse program on your commlink and have it search the Matrix this way. Searching with the agent is another way of searching, which has its advantages (for example a node can restrict access to its database to local programs. In that case, your Browse program won't be able to accses it, but an agent will) and its drawbacks. |
I was talking about agents and your solution to ASA by making them use the target node. So one last time in its full glory:
As has been mentioned in numerous threads having agents as distinct actors makes the Matrix break. Since you can copy them you can cram any amount of actions into one turn, making it statistically impossible to fail at any task that requires you to roll dice.
Agents as described in the BBB can run all the programs any user can, making able to perform any action a decker could.
Users interact with matrix nodes, agents and drones by subscribing to them, while having their programs run on their own commlink.
Now if I understand you correctly, you want to make agents need to switch nodes to interact with the objects in the target node, which, while being counterintuitive, could actually be the case, given there's only one vague paragraph in the BBB to get our conclusions form.
If that is the case you lose contact with the agent as soon as you send it off, since it's in a different node now and can't interact with you anymore.
But even if we find a way to twist our way out of that particular issue and forgoing any discussion on the way agents gain accounts, there's a real game balance with handling things this way:
Agents not integrated into a persona use up to 6 program slots out of the at best 5 program slots a node has available before losing response. So doing things this way we have just found a way to drop any nodes response to one, making any actual hacking attempt by a for real haxxor a cakewalk. Which is just as stupid as having 10 million agents roll redirect trace.
Now I don't have a solution for the ASA that doesn't just write independent agents out of existence and gives some programs limited amounts of actions they can take for themselves (similar to the Analyze example given in Matrix Perception Chapter).
| QUOTE |
| because agents dont work that way? as in, they can only subscribe to whatever it is thats giving them orders? |
I am having a hard time figuring out if you are being sarcastic, by asking questions that you think are rhetoric, since its absolutely clear how agents work. Or whether you are genuinely trying to figure things out. So I'll quote the FAQ at you, in the hopes of giving you a better understanding of what subscription is supposed to be.
| QUOTE |
| The act of subscribing is merely the act of creating and maintaining a connection between two nodes. Subscribing does not automatically grant access to a node (unless it happens to be a public all-access node) -- that is the purview of accounts. Subscribing is essentially the "handshake" that occurs between two nodes, a protocol check and very basic form of authentication so that each node knows it's connecting with the right other node. |
But I agree. Agents don't work.
ok, let me restate that, agents do not create subscriptions, their controller does.
also, the FAQs take on subscription is somewhat contradictory to the book iirc. they seems to approach subscriptions in two different ways.
hell, notice that the quoted text talks about node to node traffic, but the book talks about persona to node/agent/drone traffic. all in all, i suspect they need to get their ducks (labels) in a row before they write any more faqs on the topic.
as in, sit down and define in clear terms subscription, persona, node, and how they relate.
as it is, it seems that persona and node is sometimes used interchangeably. but that cant be as a node cant "walk around" inside another node, now can it?
as for me being sarcastic, maybe so. or maybe im just tired of seeing new threads about the same old topics. hell, i wonder why i bother to post on them. maybe its because its either that or ED/SR crossover threads on this forum right now?
also, the FAQs take on subscription is somewhat contradictory to the book iirc. they seems to approach subscriptions in two different ways.
hell, notice that the quoted text talks about node to node traffic, but the book talks about persona to node/agent/drone traffic. all in all, i suspect they need to get their ducks (labels) in a row before they write any more faqs on the topic.
as in, sit down and define in clear terms subscription, persona, node, and how they relate.
as it is, it seems that persona and node is sometimes used interchangeably. but that cant be as a node cant "walk around" inside another node, now can it?
as for me being sarcastic, maybe so. or maybe im just tired of seeing new threads about the same old topics. hell, i wonder why i bother to post on them. maybe its because its either that or ED/SR crossover threads on this forum right now?
Yes it looks like the rules were written by people having something different in mind and the FAQ by a third one, who didn't pay attention to what was written 
(no offense to the devs, I know it's a hard task).
Sma: I don't want to turn that back into the same argument... Suffice to say that both options for agent are possible according to the rules, none of them is more logical or more intuitive (not to everyone at least). I've got my own vision and it's working fine (and I've been paying attention to all discussions about the Matrix/agents to see if there was anything that could break it). If you'd like to discuss it, please use PM. I don't think Dumpshock needs another of these topics.
(no offense to the devs, I know it's a hard task).Sma: I don't want to turn that back into the same argument... Suffice to say that both options for agent are possible according to the rules, none of them is more logical or more intuitive (not to everyone at least). I've got my own vision and it's working fine (and I've been paying attention to all discussions about the Matrix/agents to see if there was anything that could break it). If you'd like to discuss it, please use PM. I don't think Dumpshock needs another of these topics.
| QUOTE (hobgoblin) |
| as it is, it seems that persona and node is sometimes used interchangeably. but that cant be as a node cant "walk around" inside another node, now can it? |
Actually, they are. Your persona is the icon for your commlink made from your OS/running programs. In essence, your persona is the iconic representation of your node while it is visiting other nodes. It still is your node. When someone crashes your persona, its your commlink that reboots, the same as if they connected to your commlink and crashed its OS.
I'm a bit confused here. It sounds to me like some of you are referring to agents as an AI of sorts, I was pretty sure those weren't around yet. I always thought that agents were programs that could utilize other programs with a limited intelligence, meaning that they could use them with a certain set of parameters but not with creative intelligence.
Some of you are saying that you could use a browse program to find something, or just tell your agent to do it. While this is true I understood that the Agent would just be using your browse program to perform the task in your place and the results would depend on how good the ratings of both programs were.
With my understanding it doesn't matter if you copy an Agent 100 times and set all 100 copies to analyze a system. I think we're confusing the reality and metaphors a bit here. If you try to hack into a node with an Agent running analyze you may sneak past in VR by hiding behind a bush until the Agent passes, but in reality what you did is find a loophole or a workaround that fooled the particular agent/analyze program. So duping the same Agent program 100 times would be pointless once you passed the first one, they are all identical and whatever exploit you found to pass one will work on that same one and on all his copies until someone with creative intelligence takes a hand in the process.
Sure a corp could hide some sensitive data behind 10 different agents each using 10 different analyze programs but they rarely do for the same reason banks only have 1 alarm system. They will have various different forms of security but after a point it's not cost effective because you have to assume that something is secure.
This applies to the old "copy an agent a thousand times and an IC a thousand times them stick them in a thousand comlinks then attack a megacorp" They are still the same program all the copies do is increase network traffic.
Some of you are saying that you could use a browse program to find something, or just tell your agent to do it. While this is true I understood that the Agent would just be using your browse program to perform the task in your place and the results would depend on how good the ratings of both programs were.
With my understanding it doesn't matter if you copy an Agent 100 times and set all 100 copies to analyze a system. I think we're confusing the reality and metaphors a bit here. If you try to hack into a node with an Agent running analyze you may sneak past in VR by hiding behind a bush until the Agent passes, but in reality what you did is find a loophole or a workaround that fooled the particular agent/analyze program. So duping the same Agent program 100 times would be pointless once you passed the first one, they are all identical and whatever exploit you found to pass one will work on that same one and on all his copies until someone with creative intelligence takes a hand in the process.
Sure a corp could hide some sensitive data behind 10 different agents each using 10 different analyze programs but they rarely do for the same reason banks only have 1 alarm system. They will have various different forms of security but after a point it's not cost effective because you have to assume that something is secure.
This applies to the old "copy an agent a thousand times and an IC a thousand times them stick them in a thousand comlinks then attack a megacorp" They are still the same program all the copies do is increase network traffic.
I have a few questions and perhaps I'm just misunderstanding...
First. What about copy protection? The book is very clear on copy protection. Granted, once that is broken then you can copy the agents.
Second. An agent and each piece of software an agent uses pulls down the node it runs on... I fail to see how you can run a dozen of anything, let alone a hundred on one node.
First. What about copy protection? The book is very clear on copy protection. Granted, once that is broken then you can copy the agents.
Second. An agent and each piece of software an agent uses pulls down the node it runs on... I fail to see how you can run a dozen of anything, let alone a hundred on one node.
Redjack. The agent is running on the commlink you put it on. Its connected to and attacking whatever you told it to, but it is still running on the original commlink.
BlackRabite: Yes agents have limited intelligence, but according to the rules, they have use their rating for hacking test. So according to the rules, a rating 6 agent is as good as a hacker with rating 6 hacking skills.
Personally that's something I'm not really fond of (and I don't think that agents with hacking skills should be legally available).
As for sending the agent with the browse programs, there are different ways to see this, depending on how you consider agents.
If you consider them as "agents in a box" (running on their own nodes, but being able to connect to any other node without moving themselves), it'd just replace your data search skill with their own.
If you consider that an agent can't act outside the node he's loaded on, that means he'd be able to upload itself to nodes and browse them directly, which can be the only way to access some databases.
About the ability to fool 100 agents once you fooled one, some might argue that heuristic in 2070 are better than now and that agents can learn and/or act differently depending on the occasion. They might support their point by stating that the agent rolls his analyze, which might show that it can analyze well or not.
@Redjack:
1. Yes, once it's broken... but that's easy to do for corps. Especially since they probably have the source code of their agents.
2. That's a way to consider it. If you read carefully the rules, you'll see that nothing states that an agent can't run on node A but act on node B (just like a hacker will run his attack program on his commlink but will attack the IC on the node he's hacking).
Personally that's something I'm not really fond of (and I don't think that agents with hacking skills should be legally available).
As for sending the agent with the browse programs, there are different ways to see this, depending on how you consider agents.
If you consider them as "agents in a box" (running on their own nodes, but being able to connect to any other node without moving themselves), it'd just replace your data search skill with their own.
If you consider that an agent can't act outside the node he's loaded on, that means he'd be able to upload itself to nodes and browse them directly, which can be the only way to access some databases.
About the ability to fool 100 agents once you fooled one, some might argue that heuristic in 2070 are better than now and that agents can learn and/or act differently depending on the occasion. They might support their point by stating that the agent rolls his analyze, which might show that it can analyze well or not.
@Redjack:
1. Yes, once it's broken... but that's easy to do for corps. Especially since they probably have the source code of their agents.
2. That's a way to consider it. If you read carefully the rules, you'll see that nothing states that an agent can't run on node A but act on node B (just like a hacker will run his attack program on his commlink but will attack the IC on the node he's hacking).
| QUOTE (Tarantula) |
| Redjack. The agent is running on the commlink you put it on. Its connected to and attacking whatever you told it to, but it is still running on the original commlink. |
My point exactly... Its a never ending battle of scalability... Resources are not unlimited.. And those nodes running th agent off protecting other nodes are themselves subject to attack, being taken over and serving as a trojan to the protected node....
| QUOTE (Blade @ Aug 28 2007, 02:54 PM) |
| 1. Yes, once it's broken... but that's easy to do for corps. Especially since they probably have the source code of their agents. |
This is true for the companies that write agent software, but not for others. Also, its illegal for the ones who don't write agents. Whistle blowers would rat out the companies. This software is easily audited running on their systems... The ones who write the software still have to have the hardware to run it.
| QUOTE (Blade @ Aug 28 2007, 02:54 PM) |
| 2. That's a way to consider it. If you read carefully the rules, you'll see that nothing states that an agent can't run on node A but act on node B (just like a hacker will run his attack program on his commlink but will attack the IC on the node he's hacking). |
See scalability issues above.
| QUOTE (Redjack @ Aug 28 2007, 02:39 PM) |
| I have a few questions and perhaps I'm just misunderstanding... First. What about copy protection? The book is very clear on copy protection. Granted, once that is broken then you can copy the agents. Second. An agent and each piece of software an agent uses pulls down the node it runs on... I fail to see how you can run a dozen of anything, let alone a hundred on one node. |
1) There is no copy protection if you have the source code for an agent (i.e. you wrote it yourself). You can make all the copies you want for free.
2) That's part of the argument. The RAW seems to say that agents run from the hacker's commlink or some other source node and their icon moves around like a hacker's icon (i.e. not degrading nodes that they are in). As far as I can see, the only way to fix the agent smith army problem is saying that agents always run on the nodes their icons are in or are trying to be in.
Redjack, you're right, resources aren't unlimited. But, as I've shown in other threads, its MUCH cheaper to buy and maintain a few more agents every year 3-5 or so, instead of paying for a security hacker. The agents can do everything just as well if not better, and they don't take a salary.
| QUOTE (Tarantula @ Aug 26 2007, 04:48 PM) |
| Even better, you can run the IC on seperate hardware, and as long as you can keep buying the hardware, you can have that many IC run on the node. |
I think it's a lot easier to visualize the abstraction level if you think of Nodes as a Network, not a single device.
Adding another machine to a network often doesn't make the network that much different.
p. 204
"Th e Matrix a complex organism, a vast collection of billions
of nodes all linked together in various networks that are
themselves linked together."
p. 208
"In order to enter some nodes (devices or networks), however—
especially private ones—you must actually log in to an
account."
p. 211
"In active
mode, you can both access and be accessed by other nodes
(PANs, devices, and networks)."
P. 216
"Node—Any device or network that can be accessed."
Substituting the word Network for Node in almost all text of the Wireless chapter makes a difference in understanding.
Dire. If I buy 101 commlinks, and 100 agent programs and 100 analyze programs. I can have all my data on commlink #1 which is rating 1 across the board. Commlink #2-101 have an agent running on them and are rating 6 across the board with the agents/progs at 6. Commlinks 2-101 run one agent w/analyze each, all of them have the agent connect to commlink #1 and analyze on it. None of the agents suffer any kind of response penalty at all.
Thats the point I was making with it. In this example, commlinks 2-101 are what I was referring to by "seperate hardware" and "node" was commlink 1.
Thats the point I was making with it. In this example, commlinks 2-101 are what I was referring to by "seperate hardware" and "node" was commlink 1.
| QUOTE |
| If you'd like to discuss it, please use PM. I don't think Dumpshock needs another of these topics. |
Heh, so much for that.
But back on topic.
| QUOTE |
| 2) That's part of the argument. The RAW seems to say that agents run from the hacker's commlink or some other source node and their icon moves around like a hacker's icon (i.e. not degrading nodes that they are in). As far as I can see, the only way to fix the agent smith army problem is saying that agents always run on the nodes their icons are in or are trying to be in. |
Agreed on the the problem (now there's a surprise!), but I don't actually see how this is a solution. Unless having 29 copies ((less if you give them actual programs) of any old rating one agent try to access the all sixes ultrasecure node to drop its response down to one is actually what you were shooting for.
Here's a completely different proposal for a solution to agents:
Agents are sophisticated pieces of software that are close to Artificial Intelligence, in fact they are so far advanced that they at times seem to be able to read their users thoughts. By monitoring the input a decker is receiving and applying their advanced predictions algorithms onto it, they are able to perform many of the minor annyoing tasks a matrix user faces daily, thus freeing up his attention and time for the really important things in life.
There are still rumors of independent agents roaming around, even after the disaster this led to last time (for details see Emergence).
Rules:
Agents come in two ratings. You can only ever have one Agent effect you at a time.
Rating one Agents give the user an extra initiative pass (max of 4 passes per turn applies) for the purposes of using electronic devices or interacting with matrix objects, that the agent can access though the users persona.
Rating two Agents are specialist systems that provide an additional two bonus dice to one of these skills: Cybercombat, Electronic Warfare, Hacking, Computer or Data Search.
5k for rating one agents, 10k for rating 2.
But since that means dropping independent agents altogether it may or may not float anyones respective boat.
Honestly, those agents are fairly worthless. They're supposed to be the drones of the matrix, and having them do what you suggest makes them the smartlinks of the matrix instead.
| QUOTE (Sma) |
| Agreed on the the problem (now there's a surprise!), but I don't actually see how this is a solution. Unless having 29 copies ((less if you give them actual programs) of any old rating one agent try to access the all sixes ultrasecure node to drop its response down to one is actually what you were shooting for. |
You have to give agents programs. They can't do anything without them!
Someone suggested limiting the number of agents connecting to a node to the standard number of subscriptions. That sounds good, are there any problems with that that I'm missing? (especially if combined with system degradation?)
According to the rules, subscription links personas to nodes/agents/etc. but there's no mention of subscription between two nodes.
And if you do apply subscription limits to nodes, it'd mean that nodes can't handle more than a handful of users. You can decide to rule that big servers are clusters of small nodes connected seamlessly to each other so that a lot of users can be connected on the same server even if they are on a different node.
That way one user won't get more than a few agents with him in his "node", but each user on the server will have the same number of agents in their own "nodes". But it means that you can't have two hackers together in the same "node".
And if you do apply subscription limits to nodes, it'd mean that nodes can't handle more than a handful of users. You can decide to rule that big servers are clusters of small nodes connected seamlessly to each other so that a lot of users can be connected on the same server even if they are on a different node.
That way one user won't get more than a few agents with him in his "node", but each user on the server will have the same number of agents in their own "nodes". But it means that you can't have two hackers together in the same "node".
| QUOTE |
| They're supposed to be the drones of the matrix, and having them do what you suggest makes them the smartlinks of the matrix instead. |
Since I haven't seen an interpretation of Matrix Drones that doesn't lead to game balance problems, removing them from the game is the whole point.
| QUOTE |
| You have to give agents programs. They can't do anything without them! |
They do not actually do anything! They fulfill their purpose just by accessing the node! Adding in copies of programs just makes it worse!
| QUOTE |
| Someone suggested limiting the number of agents connecting to a node to the standard number of subscriptions. That sounds good, are there any problems with that that I'm missing? (especially if combined with system degradation?) |
Being able to degrade any system to ones at will by dropping a shitload of agents on it ?
Limiting the amount of agents to subscription limit only makes the attacker need more higher levels of agents with a bigger payload.
| QUOTE (Tarantula) |
| Dire. If I buy 101 commlinks, and 100 agent programs and 100 analyze programs. I can have all my data on commlink #1 which is rating 1 across the board. Commlink #2-101 have an agent running on them and are rating 6 across the board with the agents/progs at 6. Commlinks 2-101 run one agent w/analyze each, all of them have the agent connect to commlink #1 and analyze on it. None of the agents suffer any kind of response penalty at all. Thats the point I was making with it. In this example, commlinks 2-101 are what I was referring to by "seperate hardware" and "node" was commlink 1. |
If you abstract this, which is what the rules and dice rolls are about, you think of 2-101 as a network, they count as a single Node.
Also, if you have Commlink # 1 + 2-101, together they count as a single Node.
So you get
Node/Network A (Commlink # 1 + 2-101) versus Network/Node B (1-100 + Agents) as a single contested dice roll
That's what abstraction does.
If you want to get into the individual machines and devices and network topography and it's effects, that not supported in the rules in any way whatsoever, hence all the discussions on the effects of topography.
At the table it's going to boil down to the GM rolling some dice, and the player rolling some dice in a contest to resolve the result, and the abstract system supports that very well.
I'm not interested as a player or GM in mapping out the topography of a network and the capability of individual devices on the network because it's a lot of work for which there is very little material benefit in the game I'm playing.
| QUOTE (Sma) | ||
They do not actually do anything! They fulfill their purpose just by accessing the node! Adding in copies of programs just makes it worse! |
Making fun of my punctuation? Please try to show some maturity, this isn't the Pokemon forum.
I wasn't making fun of you or your punctuation, and apologize for any perceived slight.
I have been asking for a couple of posts if your proposed solution to ASA is having agents needl to run on the node they are accessing or interacting with, and thereby letting copies of agents (and the programs those agents run) reduce a target nodes response.
I also have been pointing out how that completely does not work, unless you add another layer of houserules, whose ramifications I am unable to even guess at, since you refuse to even acknowledge my first question, by either replying, asking for clarification.
Given all this I am going to assume you are not interested in discussing your houserule. Which is completely A OK. It would have saved me time spent rephrasing my point in the interest of making it more clear, if you'd said that right from the start.
I have been asking for a couple of posts if your proposed solution to ASA is having agents needl to run on the node they are accessing or interacting with, and thereby letting copies of agents (and the programs those agents run) reduce a target nodes response.
I also have been pointing out how that completely does not work, unless you add another layer of houserules, whose ramifications I am unable to even guess at, since you refuse to even acknowledge my first question, by either replying, asking for clarification.
Given all this I am going to assume you are not interested in discussing your houserule. Which is completely A OK. It would have saved me time spent rephrasing my point in the interest of making it more clear, if you'd said that right from the start.
| QUOTE (DireRadiant) |
| If you abstract this, which is what the rules and dice rolls are about, you think of 2-101 as a network, they count as a single Node. Also, if you have Commlink # 1 + 2-101, together they count as a single Node. So you get Node/Network A (Commlink # 1 + 2-101) versus Network/Node B (1-100 + Agents) as a single contested dice roll That's what abstraction does. If you want to get into the individual machines and devices and network topography and it's effects, that not supported in the rules in any way whatsoever, hence all the discussions on the effects of topography. At the table it's going to boil down to the GM rolling some dice, and the player rolling some dice in a contest to resolve the result, and the abstract system supports that very well. I'm not interested as a player or GM in mapping out the topography of a network and the capability of individual devices on the network because it's a lot of work for which there is very little material benefit in the game I'm playing. |
What is the rating of your new abstractly formed Node? Each of those commlinks can run a rating 6 agent with 5 rating 6 programs at full capacity, and all 100 of those agents can act in commlink #1 with no decrease in performance.
If you abstract those 100 commlinks to a single node, then you'd have to add together all their system ratings. So then it'd be a node with a 600 system. But, it'd still be able to run all the agents as intended.
Or, you could average their ratings, in which case it'd be straight 6's. And suddenly it can only run 1 agent instead of 100.
Whats your point?
| QUOTE (Tarantula) | ||
What is the rating of your new abstractly formed Node? Each of those commlinks can run a rating 6 agent with 5 rating 6 programs at full capacity, and all 100 of those agents can act in commlink #1 with no decrease in performance. If you abstract those 100 commlinks to a single node, then you'd have to add together all their system ratings. So then it'd be a node with a 600 system. But, it'd still be able to run all the agents as intended. Or, you could average their ratings, in which case it'd be straight 6's. And suddenly it can only run 1 agent instead of 100. Whats your point? |
You can either try to figure out how to
Add 1 + 100 Commlink ratings and figure out their affects and then figure out how they interact working against the 100 commlink chains, which involves a large order of math calulations, and rules which do not exist, which makes it all rather difficult to resolve.
Or Abstract it to
Node/Network of 101 versus Node/Network of 100, which is in the rules.
Your choice. Your game.
Dire, what do you mean? What 1 + 100 are you talking about? And abstracting it to 101 vs 100? Why is it vs? The agents are protecting the commlink, not blitzing it.
This is a ruling I've been considering for a while:
What if identical copies of agents, AIs, (programs?) etc could not subscribe to nor operate from the same node at the same time? Only the first counts (or maybe most recent, but I'm thinking first). If you threw in SOTA rules (which is my easy deterrent to the logic 1 hacker) then accumulating 100 unique agents would beimpossible difficult for a lone hacker and probably deemed an unnecessary expense by (the been counters in) the Corp. Corporations would of course have as many unique agents as the GM wanted to justify, but that was always true of spiders anyway. I've been considering implementing this, and I would truly like some feedback as to how well this would or wouldn't work and why. (Incidentally, this would give huge incentive to a hacker to keep his agent programs private.)
My SOTA rules: every hacker picks a random number 1 to 6 to assign to each program and agent (and sometimes 2 numbers for R&D intense things like exploit and firewall). Every 2 months or so game time the GM rolls a d6. If your program's number matches the die roll, decrease its rating by 1. Want the rating back up? Start coding, you'll have a new program written in no time I'm sure. (Please give feedback on this too.)
I've always liked SOTA rules in regard to the Matrix even while I bitched about updating my characters programs. To me they reflected the constantly changing computing world. A several thousand dollar computer is usually just a paperweight in less than a decade's time.
SOTA for things like cyberware? Not so much. The legs that made me twice as strong as a typical human suddenly don't because a new model came out? No. --But that's just my tangential ranting.
What if identical copies of agents, AIs, (programs?) etc could not subscribe to nor operate from the same node at the same time? Only the first counts (or maybe most recent, but I'm thinking first). If you threw in SOTA rules (which is my easy deterrent to the logic 1 hacker) then accumulating 100 unique agents would be
My SOTA rules: every hacker picks a random number 1 to 6 to assign to each program and agent (and sometimes 2 numbers for R&D intense things like exploit and firewall). Every 2 months or so game time the GM rolls a d6. If your program's number matches the die roll, decrease its rating by 1. Want the rating back up? Start coding, you'll have a new program written in no time I'm sure. (Please give feedback on this too.)
I've always liked SOTA rules in regard to the Matrix even while I bitched about updating my characters programs. To me they reflected the constantly changing computing world. A several thousand dollar computer is usually just a paperweight in less than a decade's time.
SOTA for things like cyberware? Not so much. The legs that made me twice as strong as a typical human suddenly don't because a new model came out? No. --But that's just my tangential ranting.
Nice dodge, would you mind clarifying what you were saying in regards to my post?
I like the idea of SOTA rules, but in practice, I'm really split. I feel like program upgrades should be included in lifestyle costs, but the upgrades are likely very expensive (greater than or equal to the cost of the entire lifestyle, depending on the lifestyle), so inclusion doesn't logically make sense.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.