SR4A Agent..Huh?, Help me wire my brain right. Options

[Wiseman]

Alright, I need the DS experts to help me sort this out in my mind.

SR4A, page 234

Agents can also access other nodes independently, if instructed to
do so. They must have access to the node either by using your account
or by hacking their own. When operating independently, an agent has
its own icon. Agents may not have icons in multiple nodes like persona
programs can.

Emphasis mine.

I don't quite understand this last statement. Icons in multiple nodes requires a subscription, I get that. But if I load an agent on my persona, and access multiple nodes (only being "active" in one), he can't be used in each? This is the lesser of the two, as you could say by virtue of being loaded on the persona, the agent isn't its own icon really (but you can crash it and attack it in cybercombat separate from the persona).

If you load an agent on a separate device and send him to hack another node, once he gains access, wouldn't it then be in two nodes (thus multiple)? The node it's loaded on would have an icon, and wouldn't the node the agent hacked also have an icon now? Or does the agent have to "move" to the hacked node and load itself onto that node now?

I know i'm just not thinking of this right, blow the cobwebs out of my mind.

[Wacky]

My understanding is that the Agent can't have subscriptions to both nodes. For a fluff gaming reason, it's pilot (ie "Brain") can't handle processing the reality of both nodes so it just stands there with an odd expression on its face.

It's essentially the old Windows ME of Shadowrun programs as it doesn't handle talking to multiple networks at once (with the understanding that Agents are better as they have a much higher practical use than Windows ME and crash only if a malicious hacker is trying to do so).

That's my thought.

Sign--

Wacky

[Zaranthan]

Short answer: yes. If an agent hacks into a node, it has to move itself to that node to do anything further. If it gets crashed out there, it's done, and you have to load a new instance to do anything.

[toolbox]

Also note that the agent only has its own icon while operating independently. If you have it loaded on your system as a standard program, it has no icon and can be used in any node your persona is currently accessing.

[BishopMcQ]

You've got a few items here so I'll try to break it down.

--Agent is loaded in your persona--The agent can access multiple windows as the persona does to assist wherever your focus is.
--Agent is loaded on the same node that your persona--The Agent is able to access and assist you access, though you must have access to run programs on the node. Basically, you hack in, upload the agent and it helps you.
--Agent is operating autonomously--The agent moves itself from one node to the next as it needs to.

Read the entire section, rather than just the one paragraph to build a broader context for the information. There is also supplemental information on Unwired p. 110-111.

[Rotbart van Dain...]

..which is directly contradicting:

Like any Matrix user, the agent can access multiple nodes at once. Other nodes must be accessed with passcodes or hacked, per normal rules. The agent remains loaded on only one node, however - though it interacts with other nodes, it does not need to be copied and loaded on them.

And that's imporant - otherwise, the dicepool of an agent performing a search would change constantly - which is completely FUBAR.

[crizh]

I think you are confusing the Agent itself and it's Icon.

An Agent can be loaded and running on Node A whilst being logged in to Node B and using an Icon to perform operations on Node B.

To perform operations on Node A it must first log off Node B and log on to Node A.

[hobgoblin]

this is a serious "nerf" of agent abilities, and may well contradict info found in unwired...

it may also impact the abilities of AI, as they are basically awakened agents...

[deek]

Maybe I am thinking about it too simply. But the way I am reading all this is that an Agent can do everything a hacker can do, except act simultaneously in multiple nodes.

Seeing that a commlink, per RAW, can only house one persona at a time, the autonomous Agent would need its own commlink or be loaded somewhere that a persona could be ran. But after that, the Agent goes about its business per a normal hacker. It can access multiple nodes and whatever else a hacker with xyz programs can do.

The only thing it cannot do, if its has accessed nodes A, B, C and D, is be in each at the same time. So, if it has hacked all four successfully, the Agent would have to choose only one to be active in. While a hacker could be active in all four simultaneously.

[Wiseman]

Read the entire section, rather than just the one paragraph to build a broader context for the information. There is also supplemental information on Unwired p. 110-111.

What would imply I didn't read it all exactly? This seems overly presumptious and dismissive on your part and definitely isn't helpful. If you have some RAW to quote that proves a point or shows me where i'm going wrong, by all means, post it and I'll be most appreciative.

Now on to the meat. I'm going to quote from SR4A because it is the most recent (and more correct) and also its the only one I have in electronic format so its easier to copy and paste.

My problem specifically is that single sentence, as it goes against everything I had understood about agents, icons, and multiple nodes.

An Icon is defined as:
SR4A, 224

An icon is the representation of something that exists purely in the
Matrix. Your persona is represented by an icon in the Matrix, as are
your running programs, files, agents, IC, links to nodes, etc. Nodes
seen from the outside, data packages, and controls for physical devices
(such as switches for door locks or environmental controls for an apartment)
also appear as icons in the Matrix.

Everything represented in the Matrix that matters has an Icon.

Also note that the agent only has its own icon while operating independently. If you have it loaded on your system as a standard program, it has no icon and can be used in any node your persona is currently accessing.

Says where exactly? I can't find anything that says that and the RAW above directly refutes this. If I load an agent on my persona, it still has a separate icon, as do all programs. If i'm in multiple nodes simultaneously, so are all my program icons (as they all originate from the commlink that has the multiple subscriptions).

You cannot crash an agent program, you have to defeat it in cybercombat. You don't attack my persona to damage the Agent, and I don't run Medic on myself to patch him back up. He functions independently (using my processing limit) as his own icon and on his own initiative (using my commlinks response).

It gets even more confusing when you try to interpert that one sentence when you consider loading an agent on its own node.

..which is directly contradicting:

QUOTE (Unwired @ p. 110)
Like any Matrix user, the agent can access multiple nodes at once. Other nodes must be accessed with passcodes or hacked, per normal rules. The agent remains loaded on only one node, however - though it interacts with other nodes, it does not need to be copied and loaded on them.

And that's imporant - otherwise, the dicepool of an agent performing a search would change constantly - which is completely FUBAR.

Exactly. And that section goes on to make the distinction between accessing a node and moving there.

Regardless though, If I load an agent into an independent node, he has an icon there. If I send him to access another system via legitimate account or hacking, whats the representation of the agent as he hacks in...the icon. While he is in the node (not loading onto the new node) he has an icon. If my commlink gets hacked while my agent is busy in another node, couldn't the intruder attack the agent icon in my home node?

I get the "can't function" in multiple nodes at once which seems implied, but the can't have multiple icons is messing with me. Why wouldn't the same rules apply to the agent that apply to the persona functioning in multiple nodes. They choose which node they're "active" in for each IP. Defenseless to cybercombat in all other nodes they're present but "inactive".

I think you are confusing the Agent itself and it's Icon.

An Agent can be loaded and running on Node A whilst being logged in to Node B and using an Icon to perform operations on Node B.

To perform operations on Node A it must first log off Node B and log on to Node A.

Maybe I am, but as far as the matrix is concerned the icon IS the agent and vice versa. Theres really nothing to be confused. Logging on a node is a single complex action and creates and icon. So an agent loaded independently on a commlink that is told to access another system now has two icons. One for the node he is loaded on, and one for the node he is accessing.

Log On (System )
You open a subscription to a node, and your icon appears there. This
requires no test, but does require either the proper authentication to an
account (such as a passcode) or a hacked account. You also need a connection
to the node’s device, either with a wired connection or a wireless
connection (by being within mutual Signal range or establishing a route
across multiple devices).

Maybe I am thinking about it too simply. But the way I am reading all this is that an Agent can do everything a hacker can do, except act simultaneously in multiple nodes.

Seeing that a commlink, per RAW, can only house one persona at a time, the autonomous Agent would need its own commlink or be loaded somewhere that a persona could be ran. But after that, the Agent goes about its business per a normal hacker. It can access multiple nodes and whatever else a hacker with xyz programs can do.

The only thing it cannot do, if its has accessed nodes A, B, C and D, is be in each at the same time. So, if it has hacked all four successfully, the Agent would have to choose only one to be active in. While a hacker could be active in all four simultaneously.

I like this interpertation, and simple is good. I get hung up though when you say acting in multiple nodes simulataneously. You can only be "active" in one node at one time, despite how many nodes your logged onto or have access. Why should the agent be any different?

SR4A, pg 225

It is possible to be subscribed to multiple nodes at once. This means
you can have icons in multiple places in the Matrix at any given time.
You are aware of what is happening in all nodes in which your icon
exists. You may only control one icon at a time, but may switch which
is the “acting� icon at will. This has ramifications for cybercombat
(Simultaneous Combat in Multiple Nodes, p. 237).

If that one sentence didn't exist, everything would work perfectly IMHO. The use of icon in the context of the restriction seems to far reaching in scope to make sense in regards to how agents are implemented and how they act independently. The agent should have icons in multiple nodes and only act in one, specifically so if he is loaded onto a persona that has said access to multiple nodes.

Maybe I need to rethink and take it as having both the loaded program icon and the "active" agent icon as two separate icons and apply the rule only to the latter. Effectively disregarding the icon from the node or persona in which it is running and only focus on the icon in the node/persona where its actually doing something.

(quick edit, I realize thats what some of you are saying, my problem was/is that everything has an icon, even programs that don't necessarily need one that are loaded on your persona. Icon is a very broad definition for fluff/comprehension purposes and not an actual mechanical definition on its own.)

[deek]

Reading through this thread, and I may be confusing myself, but here goes...

A hacker has a home node. He hacks node A and does stuff. He then hacks node B and does stuff. He can be in node A and B, with two icons and he chooses which node to act in.

An agent has a home node. It hacks node A and does stuff. It wants to hack node B, but first needs to logoff node A. Its back in its home node and now it can hack Node B.

The agent has accessed multiple nodes, but it only has one icon "active" at one time. The hacker has also accessed multiple nodes AND it has multiple icons "active" at one time.

I think that is the difference. The hacker is able to switch between its multiple icons, whereas the agent is needing to leave one node before hacking/logging into another.

Not sure if that helps or hurts this discussion...

[crizh]

What deek said.

[Shard]

How about a "real world" example?

You could load your agent into your laser printer. Just because it's running there doesn't mean that it has an icon there. When it hacks your system's wireless hub and gains access, the icon moves from the printer to the wireless hub, but the agent is still running and residing on the printer. It's only got an icon in one place, where it's being active. From the wireless hub it hacks your cube-mate's computer and gains acess. The agent's icon then moves to the system. The agent itself is still running on your printer, but it is now active in your cube-mate's computer. From there, it hacks their monitor and prints "Hello World" with an Edit program. Poof, the icon is in their monitor and it's still running on your printer. The nodes inbetween serve as the agent's datatrail, neh?

You could attack the icon itself in the monitor and try to eject it from the node, but to crash the agent completely, you'd have to attack it in the printer, no?

[Wiseman]

Reading through this thread, and I may be confusing myself, but here goes...

A hacker has a home node. He hacks node A and does stuff. He then hacks node B and does stuff. He can be in node A and B, with two icons and he chooses which node to act in.

An agent has a home node. It hacks node A and does stuff. It wants to hack node B, but first needs to logoff node A. Its back in its home node and now it can hack Node B.

The agent has accessed multiple nodes, but it only has one icon "active" at one time. The hacker has also accessed multiple nodes AND it has multiple icons "active" at one time.

I think that is the difference. The hacker is able to switch between its multiple icons, whereas the agent is needing to leave one node before hacking/logging into another.

Not sure if that helps or hurts this discussion...

Makes sense and is most likely the direction i'll go with (see edit). I've heard enough counter perspectives to at least not have my head hurt trying to think about. There are some serious questions that still arise under this interpertation however.

First, bearing in mind Log Off is a simple action, and Log On is a complex action, and so cannot be performed in the same IP. There is a good bit of delay to command an agent to transfer to a node you're persona is active in. Doesn't that seem a bit extraordinary in time wasted for a program loaded on your persona via the same connection you're using to access a different node to act?

Second, if everything has an icon, even connections, and the agent's icon is only apparent in the node it's logged into, can you attack an agent in the node it's loaded onto even though it's logged on performing functions elsewhere?

In the case of the second, for it to be attackable, it has to have an icon. You cannot just crash the agent program, as it has a condition monitor and must be defeated in cybercombat. You cannot have cybercombat with something that doesn't have an icon. But it's running on the node you've infiltrated, it's just performing it's functions elsewhere.

So either it must have two icons based on the definition of icons, or it doesn't and you cannot attack it in its home node while it's elsewhere. There is something conceptual about this latter answer that just makes me itch.

*Edit - to avoid double post but in the same vein.

Unwired page 110, First sentence of second paragraph is distinct.

Like any Matrix user, the agent can access multiple nodes at once.

This doesn't imply they can, it says it plainly. This and everything else published before SR4A, and many elements of SR4A itself are in direct conflict with statement in the OP.

If an agent can have access to multiple nodes at once without moving, it must Log On, and therefore have an icon in each. So either I throw away/disregard a bunch of stuff in Unwired, or I perform a personal edit to scratch out the last line in SR4A.

[crizh]

SR4A, page 234

Agents can also access other nodes independently, if instructed to
do so. They must have access to the node either by using your account
or by hacking their own. When operating independently, an agent has
its own icon. Agents may not have icons in multiple nodes like persona
programs can.

This quote is immediately proceeded by text indicating what happens when an Agent is running on the same node as your Persona.

Therefore this quoted text refers to Agents that are running on another node.

An Agent that you are subscribed to, that is not running on the same node as your Persona, may access other nodes independently.

Accessing a node is wildly different to running on another node. You do not have to be running on a node to access it.

The last sentence of the above quote is clumsy and seems to have been written by someone without a firm grasp of SR4's Matrix mechanics.

It is flat out contradicted by precedent and the text of Unwired.

It should be noted that the section immediately preceding the section on accessing multiple nodes simultaneously states.

You do not have do subscribe to the node in which your persona is running; you are always in that node without the need to subscribe to it.

This clearly sets a precedent that the node you 'run' in doesn't count.

[kzt]

How about a "real world" example?

You can't use real world examples for the SR4 martrix. It's one of the major weaknesses of the entire system is that it makes less sense the more you understand of actual computers and networks. It's probably OK if your sum total experience with computers comes from reading books written by this guy on his mechanical typewriter and watching TRON....

[hobgoblin]

it depends on how specific you make your examples.

sure, you cant apply direct tcp/ip knowledge to the protocol layer of the matrix networking system, but i find that i can make quite nice sense out of it using general description of real life setups.

[Falconer]

Okay above... IF the agent is loaded into your persona... and you're accessing multiple nodes at once... then it can access multiple nodes at once. So Unwired isn't in contradiction w/ SR4a at that point.


I'll point out SR4a is the newer book and supercedes unwired. (and quite frankly they did a massive rewriting of the matrix section thankfully, so I think it fairly safe to say... that they intended to kick these kind of agent smith abuses out). Put another way... SR4a are the core rules published AFTER, the optional unwired expansion book. I'd say that's pretty clear which has precedence and includes any rules fixes.


Here's the problem w/ agents running remotely and only icons popping up... any node can be flooded w/ agents WITH NO RESPONSE DEGRADATION. Congratulations agent smith is now about the only workable technique.

As the other poster pointed out quite significantly, IF the agent is running autonomously in a node, then it MUST be logged into that node (the laser printer). It cannot move it's icon to a new system w/o removing it's existing icon from the laser printer... hence it must move itself and place it's new loading on it's new 'home'.

And yes AI's suffer the same problems... AI's also get a significant 'home node' benefit as well which does not move with them outside their home turf.

[Wiseman]

I'll point out SR4a is the newer book and supercedes unwired. (and quite frankly they did a massive rewriting of the matrix section thankfully, so I think it fairly safe to say... that they intended to kick these kind of agent smith abuses out).

I don't agree that SR4A made Unwired obsolete. Specifically because they advertise it for "advanced rules" in SR4A itself.

Also they can errata anything they like and I don't see anything for unwired other than the removal of (a very similar situation) one line about spoofing agent's access ID's when loading them on a node, which directly allowed what they were trying to prevent in the agent smith scenario.

Here's the problem w/ agents running remotely and only icons popping up... any node can be flooded w/ agents WITH NO RESPONSE DEGRADATION. Congratulations agent smith is now about the only workable technique.

If they're acting independently, they can't have the same access ID and be in the same node, since access ID comes from their software, you'd have to patch all the agents in an extended test to have any sort of "army" (or buy multiple programs).

So no agent smith swarms there. If you referring to corporations using auxillary nodes to run programs in their defense, and send in agents to the compromised node from elsewhere, they already do so by way of hackers, spiders, and security nodes.

As the other poster pointed out quite significantly, IF the agent is running autonomously in a node, then it MUST be logged into that node (the laser printer). It cannot move it's icon to a new system w/o removing it's existing icon from the laser printer... hence it must move itself and place it's new loading on it's new 'home'.

This statement is directly contradicted in Unwired, and this only flies if you're choosing to toss out Unwired as having their "advanced rules" trumped by a single line in the agent section of SR4a. Again, I can throw out a book and all my previous understanding of the mechanics, or I can get rid of that one line.

I can provide thousands of examples, real life or shadowrun matrix wise where you'd have to have agents with multiple icons, in more than one node at once. The icon in another node represents the agents connection, the icon in the node in which it is running is the closest to the "true" program icon, but as far as cybercombat and track and all the other mechanics, they are one in the same.

If it didn't have multiple icons, it would have to copy itself to the vendor node to update for a legitimate patch. It would transfer its running program to the site. That makes absolutely no sense. Its a program running on my commlink.

Am I to believe that the program copies and removes itself from node to node when its merely an autonomous dog brain using my comm to "fetch" information? And thats accepting you burned Unwired's entire sections on agents...

[crizh]

Unwired p110

Like any Matrix user, the agent can access multiple nodes at once. Other nodes must be accessed with passcodes or hacked, per normal rules. The agent remains loaded on only one node, however though it interacts with other nodes, it does not need to be copied and loaded on them.

Is that clear enough for you all?

Until I see errata explicitly contradicting this I ain't interested in any further argument on the subject.

That is the RAW.

The text in SR4A is cludgey and does not explicitly contradict the above. If you infer it's effects to their logical idiocy you find that the above quote cannot be true.

It's not a difficult leap of the imagination to infer therefore that the text the OP quoted is just poorly written by someone with an incomplete grasp of it's ramifications and that it will be errata'd in due course.

Certainly less of a leap of imagination than it takes to assume that it overrides the explicit rules in Unwired.

Perhaps Aaron or Tiger Eyes might like to jump in here?