What?

OK, the break in interpretation is not whether an icon appears on the target Node. Of course it does. As defined in the book an icon is a "graphical representation". The break in discussion is whether the Agent is running on the Node it hacked access from or whether it had to transfer its entire existence onto the Node it is hacking access to.

In short, whether people can kill your Agents by turning off their computers when your Agents are hacking into them.

---

How does this not apply to Agents as well? They have Matrix Perception, they only have Matrix perception!

How does that not apply to Agents? They access nodes. Therefore they have a copy of their Icon in each and every node they access. And that means that they don't have to move the Node they are running on to access a new node, neh?

Well golly, since an Agent can be Tracked at all, that means that it has a location other than the Node it is acting on that it resides upon. And you know what that means? It means that it is accessing from somewhere when it is using Edits and Analyzes on a node it is accessing.

This is really extremely open and shut. Before, during, and after an agent has hacked itself access to a Node, it is still running on the Node that it hacked from. Actually moving your Agent onto the new Node so that it will run on that system and use that system's response and memory is an optional thing that you will very rarely do.

-Frank

Oh for craps sake.

I'm not going to debate Gibsonesque VR theory versus real computer theory. I am quoting the SR4 book in this discussion. You know, the game we are supposed to be playing.

Go read the book, the page numbers are in the quoted text. I'll wait. Then come back and tell me what the book says an "icon" is. Not what you think it should say. What the book says.

Uhh... other way around.

Most programs have icons.

There are "objects" in the Matrix. And you see icons that represent them. Not the other way around.

You can attack things in the Matrix. This is refferenced as "attacking icons" because that's what you see happening. That's much easier to say than "attack a sprite, node, agent, or independent program" and gets used a lot as terminology.

In short, it is correct to say "I'm going to do something to that icon." - but it is more specifically correct to say "I'm going to do something to the agent that is running on a distant node because it hs created a connection to the node I am in that is represented graphically with this icon."

-Frank

So you admit that an Icon is at least a graphical representation of a connection.

Admit?

That's the entire point. THe icon is just that - an icon. And the connection is also just that - a connection. The Agent is running on Node A and is connected to Node B. This allows the Agent to perform actions such as Edit or Browse on Node B.

Then, if another matrix entity spots the Agent's activities, they will see an "icon' that represents it. And they can potentially attack that Agent through the connection, which they will interpret as attacking the icon.

The core thing to understand is that all Matrix actions still have a "from" (which is where the programs generating the actions are actually running) and a "to" (which is where the results are being felt). Agents, as well as everything else in the Matrix, are capped by the Response and System of the "from". Which means that if you are hacking from a really good commlink into a shitty system then you're golden and if you try to do things the other way you're up a creek.

-Frank

No, the agent is an independant icon. When an icon appears on another node, that means the agent is loaded there and can't, by definition, be loaded somewhere else. He can't have more than one icon - he *is* only one icon.

No. When a node is shut down, all programs there shut down. They're not even crashed. The owner of the agent, once he realises the agent is gone and not coming back because it is not loaded into the Matrix anymore, can simply load the program again. Since the program was shutdown rather than crashed, he doesn't even have to wait for the reboot time. It's just a Complex Action to load the program. Once he does that he can choose to either load it into his persona, or into any node that he has access to (through his numerous persona icons).

I don't think there's anyway of destroying agents in RAW, short of maybe searching for the files in the comlink's node and editing them.

---

It applies to agents because of the part of the text you want to ignore - an agent is loaded into the node to act independantly. Alternatively, it can be loaded into your persona and would share its (matrix) perception.

It doesn't apply to agents because it says "each node you access". Access in this case is used in the present tense, so it is referring to the node you access at this moment not every node you access ever.

Since only personas can access multiple nodes at once, as specified in the Accessing Multiple Nodes section on p218, agents cannot do this. There are logical game-reasons for this. They can access other nodes, but not simultaneously, and not at the same time they are accessing their current node, because they must load into a node to act independantly. You can only act in nodes in which you have access. To access a node, your icon must move there. In this case, the agent's only icon must move there.

Nope. Again, I think that's based on your interpretation of the words trace and track and datatrail rather than anything in the book.

You can Track and agent and its datatrail will lead back to the persona - which was the agent's original node because only personas can load programs.

You have to realize I'm using "the agent must be in the node" and "the agent's icon must be in the node" synonomously, because independant agents are just independant icons. They can't project or create other icons and they can't access multiple nodes.

No, "running on" has no reference in the book. The book uses "loaded to". Independant agents are icons loaded to nodes and must be loaded to a particular node in order to conduct actions there, independant of your persona.

Nope, it isn't.
Agents are little drones that exist as part of the matrix and hop from node to node doing your bidding as best it can with the instructions provided by its subscription link.

I think an AI would work the way you're describing, though. (That's not an interpretation of RAW, just my opinion.)

Citation please.

I have citations that programs have icons, that programs are represented by icons. But there is to my knowledge no statement that an agent is an icon. That would defeat the purpose of an icon.

-Frank

P 227. "Agents can also access other nodes independantly if instructed to and if they have the passcodes or are carrying an Exploit program and can hack their way in (as independant icons)."

The book says that independant agents access nodes as independant icons. And that's all it says. Cybercombat takes place between icons. When stuff happens to your icons, real things happen to what that icon represents, not just its image. Play down the "graphical" and play up the "representation" instead of the other way around. Your implying that it is ONLY a graphical representation rather than a representation that is also graphical.

In this case the icon represents the agent program. When one crashes, the other crashes. When one is analyzed, it pulls up information about the other.

The icon is also graphical.

p.227. "Using Agents." I've quoted it several times.

Btw, a little disclaimer so someone doesn't call me on it later: Sometimes I am referring to "independant agents" as "agents". That's careless, but accurate given that our discussion seems limited to independant agents only and I don't think we have any conflicts with how agents loaded into a persona work.


---
It would defeat the purpose of an icon? Why? An agent doesn't need something else to represent it in the Matrix. As you pointed out; it IS in the Matrix. It interacts with other icons representing agents, IC, data, nodes, and whatnot.

As I said, the icon represents the "agent program". When it is loaded, it is loaded as an icon. That's it. It's not loaded as something else and then projecting icons out to whereever it wants to access. Those are called personas, and an agent doesn't need one - or at least, doesn't have one.

I'm-a gonna have to save that one.

I don't understand your assertion; please describe how that would defeat the purpose of an icon?

Edit: And I don't ask you this question to avoid answering yours, I just want to make sure that I understand what you're asking me to do before I spout off and do something.

Whoa, you lost me with this part.
How does the owner of the agent get some mystic knowledge that his independently operating agent has been shut down? How does he know it's not loaded into the Matrix anymore if it has nothing to do with his own hardware? Maybe it wasn't on the matrix before. Maybe he loaded it onto a system that was subsequently physically disconnected from the matrix. So how does he know it's gone so that he can get the "all clear" to load another copy? Or could he load another copy before and agents' copy protection is meaningless?

But I've also failed to see anything that allows agents to have indepentently operating icons across several nodes. Not having evidence disallowing something cannot necessarily be used as evidence allowing it. It's up to you, me, Frank, and even Ancient History and his Pogo-Crutch (just seeing if he's still lurking) to make the distinction. Some of us go one way, some of us go the other.

No, those parts only say that agents can access other nodes. It also says the agent is an icon.

The rest of the interpretation you have to read "Access" which was already posted up today for your reference (not by me). It says that the agent has to hack in and get past the firewall and have an icon there for a node to be considered "accessed".

For the rest, go to p.218, where it talks about Personas accessing multiple nodes and icons magically appearing on those nodes. Each of these works under the same rules as normal accessing, but a new icon is created on each node to represent the persona.

No where here does it say that it applies to anyother kind of icon or program. It is an ability exclusive to personas. Furthermore, there are numerous parts where I've pointed out (primarily in cybercombat) where the book could have said "personas" or "persona icons" but said "icons" instead even though not every icon is capable of cybercombat. There is a deliberate distinction between the persona (and persona icons) and the independant agents that function as independant icons (only).

Normally icons can't move to another node on their own, except that p227 says they can. It doesn't say they access other nodes as personas do (in fact, it goes out of its way to point out that they don't) and so none of the material on Accessing Multiple Nodes on p.218 applies to independant agents.

I agree with this except for your word "only". As far as I can tell, it reads that you resolve agents using rules for icons instead of personas. Agents don't have personas so there's no reasont to apply anything that only applies to personas. Personas have already been quoted as being a "user interface" so there's not even a contextual reason that independant agents should have personas in the first place. The subscription link works as all the interface you need to send basic instructions.

You're also pre-supposing that iconography is only graphical. As has been pointed out already, the representation of icons goes beyond just mere appearance.

True... but I think one is far more logical than the other.
Also, the former is actually supported since there is text that says what the agent is. There's no need to specify what the agent is not. I think we can all agree, for example, that agents are not autosofts.

I'm not.

I have stated that nowhere.

Just because I have not stated it one way or another does not allow you to know which way I think it works.

Guess again.

You may get it right, you may get it wrong. No way to know unless I actually tell you, and since I haven't, you can't know.

But feel free to pick and choose the assumption that makes your argument work.

There are no new rules. Nice try.

And there are rules that govern activities between users and agents that should be resolved with personas or persona icons, but aren't. They're resolved with icons. And it specifically says what kinds of icons participate in cybercombat. There would be no need for this if all the participants used persona icons.

Just because you want to ignore text that indicates they aren't the same, doesn't mean its not there.

I don't know.
This is my justification for why a whole node doesn't instantly go on alert once you defeat the first piece of IC.

The owner would know the next time he tried to use his subscription link to send instructions to the agent. Or maybe he might notice that he no longer has a subscription with the agent; I don't know. Or maybe he was expecting it to come back at a certain time with the data its found and its been gone awhile...

This is pointless. All arguments that involve the Matrix also apply to the Matrix topology found on local networks. If the Agent was on a local system and the user wasn't connected to that local system, I don't know how he could keep the subscription link open that the book says is needed to provide instructions. If you take the view mentioned earlier that the subscription link can be disconnected when you are not sending instructions, I think it should still count toward your maximum subscriptions even if you are not using it, to be within (what I believe to be) the spirit of the rules (to limit how many agents you can control, similiar to a limit on how many drones you can control).

Copy protection? Copy protection doesn't enter anything is this discussion.
He doesn't have to load another copy - that's irrelevent too.
He can load the same copy so long as it is not already loaded and it is not crashed and forced to reboot.

He doesn't get the all-clear - he'll have to figure out. Maybe he'll send another agent to investigate. This is similiar to how the book treats IC which I still think was meant to the conceptual model behind the independant agent.



Sorry for missing your post. I usually look for your posts and find them interesting. It's odd that I missed it.

I covered this in a post earlier in the discussion.

The agent still remains subscribed to your persona, thus a connection. If the agent crashes, that connection would disappear and you would know that something happened to the agent.

hmm, on the topic of missed posts. it seems that everyone either missed my post on the first page, or is quietly ignoring it as it would more or less puncture the whole debate

No problem. I figured.

Hmmm. I see the first two cases working as long as he's got some kind of open connection to the agent, and the third in the case where he doesn't. It seems a bit iffy, but I'm willing to go with it.

I disagree that this is pointless. Offline systems are a very legitimate concern. Between wi-fi blocking paint, jammers, dead-zones, connections will appear and disappear sometimes. If the agent is operating independently there should be no need for a connection. I do agree that it should probably still be on the subscription list, even if that connection isn't currently active, though, if for no other reason than game balance.

Okay, here's my conflict. It's the part where you say "so long as it is not already loaded" implying that he has some way of knowing whether it is loaded anywhere in the world on any system vs. "he'll have to figure out" implying that he does not know. I see three options:
1) He can load as many copies as he wants. Copy protection is meaningless on agents, or multiple copies are 100% redundant and offer no benefits. Possible, I suppose.
2) He can only have one copy loaded anywhere in the world at any given time. This seems like what you're saying, but then how does one copy know about the other? Or how does his commlink know whether some distant out-of-touch copy is still running or not. It seems weird.
3) He can only have one copy on a system and any and all systems possibly connected through it though the matrix. I don't really like this either, is also seems to imply some strange knowledge of everything that's happening anywhere on the matrix.
Do you see where I'm getting with this? If he has to have knowledge of the agent being crashed before he can load a new one, how does he get this knowledge in cases where he is not in contact with the agent?
Maybe this is a better example:
Say the guy connects to a host that happens to be another commlink sitting on a desk. He loads his agent on to that commlink, where it runs independently. He now disconnects from that commlink, and turns off it's wi-fi. It doesn't crash the instant he disconnects, or it's not independent. (unless of course it works the way Frank is describing) So in theory its still running on that commlink. Or maybe it's crashed. We have no way of knowing, do we? It's Schroedinger's Agent. So now the guy connects to another commlink with no wi-fi sitting on a nearby desk. Can he load his agent on to that link? It can't possibly depend on whether Schroedinger's Agent is still running, can it? He either can or he can't. If he can, he can run as many copies as he wants, right? If he can't, then a lost agent is lost forever and can never be reloaded, and that's kind of odd, too.
Okay, that's enough of that.

It's no problem, I didn't take it personally. And thanks.

But what if the agent is running on a host that you have no connection to, because the host is off line? I'm fine with it still being subscribed, but you don't have a connection. (in my example) Are you saying that an agent will crash if it ever loses connection with the commlink that it is subscribed to?

I'm rereading it, and I'm not getting it. Either I don't understand what you're saying, or I don't understand how it punctures the debate.
Maybe try saying it again?

basically, by saying that when a persona successfully connect to a node, it uploads a small program that acts like the interface between the node and the hacker.

a agent does something similar, but uploads itself rather then some interface program

if you want it broken down into nice little steps then here:

1. initate connection with a node

2. node asks for credentials

3. credentials are verified or faked/hacked.

4. interface transfer initiated by node

5. if its the hacker then the comlink sends over the persona interface. if its a agent, the comlink (or whatever node the agent is on) instead puts the agent on ice, and pass it over to the target node, where its thawed and starts doing its stuff.

and yes, im just pulling this out of my ass as im typing it

in the end, do we need a single, unified way of understanding SR4?

if you dont like how a SR4 rule works, house rule it and shut up.

No, I didn't imply anything. Or, more accurately, I didn't mean any of the things you suggest I implied. A hacker cannot load an agent that is already loaded. It's not stated that way anywhere as far as I know, but it is implied in parts of the text - my previous examples where it mentions that some cyberlimbs might be merely protected by a single piece of IC. There's no reason to assume this unless you assume that the IC program can only be loaded once. It has reprecussions for everyone in this debate if we don't accept the statement as true.

Now you can copy the agent program, in which case you then have two agent programs, and you can then load both inside your persona or a node. Everything legal there. I don't like it, but it seems to be the case. We can discuss house rules to limit this abuse in this thread.

An agent can be loaded only once per program. Multiple copies of a program just mean you have multiple programs and each program can be loaded once. Each loaded program counts toward your active program limits or subscription limits, as appropriate.

It does seem weird. But as far as I can tell, that's how it works. You can tell whether or not a copy is running at any given time if you bother to check and you have a subscription link to that agent. If you try to load an agent then that has already loaded, it would probably just eliminate (unload) the other agent icon. But that's not in RAW as far as I know.

Nah, this is kind of clumsy.

I do, I suppose, but it doesn't seem unreasonable that you can instantly tell when a program is already loaded if you try to load the program. The only case when this would get weird is when you are starting the program on a completely closed system and the agent is running off the Matrix. I really don't think the rules were made to accomodate this situation and its not unreasable to assume that this one fairly rare situation wasn't taken into account.

Even if you allowed it, it hardly seems like it would matter much to the writers. Certainly not enough to warrant changing their whole concept of the Matrix rules, assuming they had already written them.

Oh man, this made me laugh out loud. My co-workers think I've funally cracked.

I would have to say that the rules don't imply that this is possible and they do imply that it is not. Further, your end result here has an effect in the game that clearly wasn't meant.

So I'd have to say not he can't, but I don't have any justification for it beyond rules and game balance.

Altogether though, I don't think it matters all that much either way. So private nodes are cheaper/easier to protect than nodes connected to the Matrix? Altogether I don't know if I have a problem with that... it actually makes sense, although the reasoning behind the conclusion seems rather silly. And Shadowrunners break into private systems a lot...

It really does sound like more of a topic for my copy protection thread which tries to identify problems with the current "copy" rules and suggest house rules to fix them.

There are explicit rules detailing how Agents and Persona are different. For example, in cybercombat "Agents, IC, and sprites have an initiative equal to Pilot + Response" (p. 230, SR4) whereas a Persona's initiative is based, in some way, on the Intuition of the Hacker behind that Persona.

Beyond that example, there are many places throughout the hacking chapter which detail the ways in which Persona act as opposed to Agents. I don't truly feel like going though and finding them all.

It baffles me that you can feel that there is a lack of explicit rules proving similarity or dissimilarity. The fact that I've just posted a rule explicitly showing a dissimilarity between Agents and Persona proves they are not the same thing. They may be related and they may have 95% commonality, but they are not the same.

You don't see a red delicious apple and a mackintosh as the same, do you?

Yes, the Unified SR4 Theory, which has been suggested by hermits but never proven.

One is not a requirement of the other. We can house rule it and shout at the top of our lungs.

Actually, I think the purpose of the thread was to put everyone on the same page so that house rules can then be written together (or at least evaluated) from a common perspective. It's a breakoff from another thread.

Dashifen, my argument only refers to iconography of Persona versus Agents, and the Cetiah's argument of the agent code must be where the icon is, and there can't be multiple icons.

Agent and Persona attributes are in fact based on different sources, though you will note that ultimately they do have the same set of attributes.

"as an independent icon" means "as separate from the Hacker's persona icon." If the Agent is running as a Persona application, it does not get an icon for itself, beyond, perhaps, a special effect on the hacker's icon. The Agent's independent icon can be detected separately from the hacker being that they are two, dare I say it, independent connections.

Thanks, cetiah. I think I understand your position better now. I'm not won over, but at least I understand what you're saying.

Regarding the above quote, don't forget the possibility that two instances (I'm shying away from the word "copy" because I don't want to talk about copy protection here) of the same agent in one place are completely redundant. If it were to be the case that multiple instances in the same place are redundant, and that a vulnerability which causes one to crash is so trivial to apply to all instances that it is essentially an automatic action for the attack program, if the user wishes, then we can allow for multiple instances of the same agent or IC to be spawned in multiple places for multiple unrelated tasks, but not add anything to one place.
The fact that it has repercussions for everyone is no reason not to consider the possibility.

Oh, and I'm glad you enjoyed Schroedinger's agent.

Dashifen: Okay, see, in my view you have a big chunk of program on a commlink that spawns an agent, and could potentially do so again. Your version is interesting, though. It certainly handles my example tidily. But of course, the copy protection will be broken and so the user can create an infinite number of individual copies (each of which will be destroyed when crashed, sure) so how do you prevent that from becoming a problem? I apologize if you've already explained it, this thread is threatening to overflow my brain.

hobgoblin: Okay, I understand better. Basically the same thing Dashifen is saying. I still don't see how that eliminates the problem that the copy protection will be broken and the person will have an infinite number of identical agents at their disposal.

Personally, with this camp I'm leaning towards the camp that if the copy protection is broken and multiple identical instances of the same agent are created, they're either redundant and multiple copies don't really help (such as in combat) or they're ability to coordinate and work together on a task has already been taken into account (such as for datasearch)
Overall, I think I like the Frank Trollman camp's (sorry to lump you all into one group, arbitrarily and unfairly singling out a spokesman) assertation that an agent ought to be able to run on one host and still have some sort of presence in multiple other hosts, just like a persona can. I know, I've seen the posts that say that tabbed browsing is an ability unique to personas, but I still don't like the idea.
I like the simplicity of saying that agents are just like hackers, their program is running on host A, and they can have an icon and a presence in hosts A, B, and C. If the agent would rather run on host B, he can copy himself there. Similarly, if a hacker with a rating 3 commlink hacks a rating 6 host and gets administrative access, he could load copies of all of his programs onto that host and have it run for him, basically giving him a rating 6 commlink and using his own commlink only as a relay for the simsense signal. I don't really have a problem with that.
Of course this whole interpretation does lead to the problem of agents replacing the team's hacker, but I like the idea of limiting agent ratings to about 4 except in the case of extremely SOTA military stuff.

All in all this is a great discussion. We're really doing well on our "exchanging information to name calling" ratio.

I use the copy protection rules on p. 228 (next to the program cost table) so to crack the protection it's a Logic + Software (x, 1 hour) where x is between 10 and 20 (game master's discretion). I've decided that glitching reduces the program's rating by 1 for each glitch regardless of whether or not you succeed in breaking the protection and a critical destroys the program beyond repair. Due to the ability of Agents to become somewhat overwhelming, I'd put the threshold high ... maybe 18. To be honest, in four semesters of gaming SR4, I've never had a character break the copy protection on a program because they either (a) don't take the software skill and don't want to screw up their programs or (b) they don't care. Plus, I do not allow the software to be purchased at character generation already broken and I have a house rule that you can't break the copy protection during character generation. Thus, someone has to take the downtime action of breaking into their programs and runs the risk of screwing them up. No one's taken that risk in my games so far.

If it did become a problem, and by problem I mean not only breaking the copy protection but then handing out copies to team mates and generally making a nuisance of themselves, I'd start making a luck test using their edge as a dice pool when the agent software is analyzed by an opposing icon. Failing that test, the analysis reports that the agent's copy protection has been compromised. If the luck test is glitched, then not only does the analysis report the compromised protection, but it'll also alert the appropriate authorities to the problem and a critical glitch might not only alert the proper authorities, but also begin a trace back to the hacker to get more details about that hacker's location, software, etc. to be given to the aforementioned proper authorities.

Then, the company that makes the Agent software could track down the hacker and does some kneecapping style justice or have him arrested for copyright violation, whatever. All the more reason to have good false identities and trade them around.

Alternate idea: you've got Agent A (source) and Agent B (copy) an the hacker wants to make another copy. Roll a luck test. If they fail in any way, they accidentally copy the copy (thus degrading the rating). Glitches further reduce the rating and critical glitches reduce the rating and accidentally remove the original. Now they've got crap agents and have to buy a new, higher rating one with copy protection and start over.

But, like I said, no one bothers to break the copy protection on their programs in my games (so far) so it's not been a problem.

Crap ... have to do something about that. You're a doody-head!


______________________

Edit: Wanted to add this but didn't think it was worth another whole post. Hopefully people notice it. I also don't think of Programs in the terms of files which can be copied as we copy files today. Instead, the Agent is a programming entity. Sure, it's executable code at its most basic level, but the user never sees it that way. In VR, you execute your attack program by picking up its gun-icon, pointing it at the bad guy, and pulling the trigger. Agents work the same way. When you tell you agent to go find you the best Mushu Pork in Denver, the bug-icon of your Agent wanders off and does so. There's no code left on your commlink when it does that -- it's actually gone off to do your search. Thus, you don't actually have the Agent and cannot instantiate a new one until it finishes what it's doing and comes back.

I have it on good authority that you smell like poopie! Should you choose to harass me further I shall have my dad beat up your dad, which he most assuredly could.

Yeah, it sounds like your house-rules patch the problem pretty well. Thanks for the clarification. And yes, I saw your edit.

There are definitely a couple different ways of looking at agents, and the different ways are logically consistent within themselves, but whenever the two camps meet and try to talk about anything that is in any way related to the matrix there are problems.

Let's try a little experiment. What would it take to actually win you over? By that, I mean what is your criteria for being in one camp or the other?

P.S., for everyone else, this thread was meant to convince supporters of my view that we were wrong if the RAW text was strictly interpreted. As far as I'm concerned, everyone should be in "my camp" by default until the opposing view is proven otherwise because the burden of proof is on the original poster and everyone supporting his view. All of my posts have been assuming this to be the case. Although I think I and others have done a very good job of showing how the text supports our view that agents must be loaded onto the node that they acting in and can therefore only exist in one node at a time, really, I don't have to. So long as I can show the slightest bit of doubt otherwise, OPPOSITION wins the debate by default.

I'm sorry. I don't want to be dismissive, but I have no idea what you are saying. "then we can allow" sounds like a house ruling to me, but I'm not sure. Which means even if I agreed, it doesn't really lend itself to this discussion. But like I said, I think I'm radically missing your point here.

Yes and no. In order to discuss any issue, both sides generally have to assme "all things are equal" except the thing being discussed. If we allow outside issues and problems in the analysis, it throws off the case for both camps. But agan, I might be misinterpreting what you are talking about.

Still enjoying it!

Maybe. Your persona would have to go back to your comlink's node, copy the program (which as Dashifen is pointing out may take awhile) and then load it. You could just say "I use my infinite storage space to make ten thousand copies during downtime" but that is a problem with ANOTHER section of the rules being discussed in another thread. It has nothing to do with the rules concerning each individual agent program.

That's a little like saying, "since its possible to make an Army of agents with infinite hacking capability, we should just assume each agent has an infinite hacking capability," which, come to think of it, is also a topic being discussed in another thread, which originally spawned off to this discussion.

It doesn't, but that's because this topic doesn't address that particular problem.

Me too. But 1) I think this is a topic for another thread, maybe two others that already exist to discuss these issues; and 2) I don't think my players will accept either of these answers without me having to pull the "I'm the GM and what I say goes" card.

Okay rules and debate aside form the moment (since this preference has nothing to do with either): I'm rather disturbed by the idea of IC protecting multiple nodes at once or that drone-equivilent dog-brained agents can be just as good if not better than hackers. I don't think either of these interpretations are in the spirit of the rules and I personally don't prefer them.

I think if we had rules for Artificial Intelligences, there would be a clearer distinction on the differences and capabilities of dumb bots, agents, and artificial intelligences. Basically, I see agents and dumb bots as the same thing and other people seem to equate them closer to AIs without sentience.

I think this has been the best argument countering my position so far.

We agree on the meaning of the words independantly and independant. If you could help me convince Frank of this definition, I'd really appreciate it. Independant = "seperate from the (user's) persona". Couple of important distinctions here: "seperate from persona" is the only requirement here, not "seperate node from the node that the persona is in" and I put user (or "Hacker") in perenthesis because only users have personas so its clearer if the word is taken out of the statement.

That addresses definition of "independant" but that doesn't address the context of the statement, which you are ignoring. The agent hacks in as an icon. Period. He has no persona and recieves none of the benefits of personas. If they wanted to give him a persona, he would have one and there are multiple instances where they went out of their way to say "icon" instead of persona, especially in cases where IC might be involved.

This kind of proves my point, though.
If Agents were treated as personas and were identical in all respects, cybercombat would refer to "attacking personas", not "attacking icons". Instead, it goes out of its way, taking a less convinient route, referring to "attacking icons" but specifically stating what kind of icons can be attacked - Personas, Agents, IC, and Sprites. Nodes and independant programs can't be attacked in cybercombat.

*whew* You know, I'm not really sure. A summary, maybe?
Seriously, I'm starting to get lost in the details here.
What do I need to be won over? I need an interpretation of the rules that is reasonably logical/possible (although does not have to be "in-line" with modern computing), internally consistent, does not logically progress to game-world-destroying conclusions, and is fun.
You may have already provided that, but I can't see the big picture because the pieces of the puzzle, while assembled neatly in your head, are scattered across a dozen pages of posts.

When we get right down to it, it's a question of whether an agent can be running on hardware A and functioning in system B, just like a hacker does all the time, or is it mandatory that they run on the system they are affecting? (independent agents only, of course, no one has any questions about agents running within a hackers persona)
In the one camp, we have agents behaving just like hackers. They can run on one host while simultaneously accessing multiple hosts in the same "tabbed browsing" way that hackers do. Anything that affects one affects all. I don't really have a problem with this approach.
In the other camp, the agent is a single chunk that can only affect the host that it is currently running in, with no possibility of "tabbed browsing". I don't really have a problem with this approach either.
Both camps have some odd conclusions, but both seem to have some pretty good arguments backing them up.
For either camp, I'm just going to shut up about copying agents, because it's off-topic and misses the point. And I admit that. The question burns my brain, but it is a question for another thread.

The way I'm currently seeing it is, either one, with a little bit of clarification, could work. The book is obviously ambiguous, and can be interpreted either way. I point not to the book for evidence, but to the length of this damn thread. Neither one is stupid, and neither one is necessarily wrong. But they can't both be right. And in the interest of having productive discussions about anything we need to figure out which, which I don't see happening short of an official answer.

So in summary, I don't think it's possible for either side to be "right" or "wrong" until we get an official answer, but what you can do is give a clear, complete explanation of how things work in your interpretation, and if people like it they can use it.

As for all your other questions, I don't know anymore. Much of it was my blathering on about the copy protection issue. Sorry about that. As for my "we can allow" comment I wasn't talking about house rules, I was talking about possible interpretations of the rules and the possible implications thereof. Probably not worth pursuing, 'cause at this point I'm getting pretty confused.

Summary of intent:

Signed by: So far, only Cetiah

• ]Agents have to be loaded onto whatever node they access.
• That is, the agent icon has to enter that node.
• Agents only have one icon.
• Persons can access multiple nodes at once, but agents are more limited.
• Agents Response stat changes, depending on which node they enter.
• Agents and IC are the equivilent of drones that must be sent out to perform a task, not the equivilent of security hackers which can provide overwatch security on all nodes at once.
• The concept of where the agent is running off of or the location or status of the agent's souce code is irrelevent and purposefully left ambiguous within the rules. It can be on your coffee machine for all anyone cares.
• Within the Matrix, the "agent" and the "agent's icon" are the same thing. You can't have more than one agent icon unless you have more than one agent program. Source code and such things exist outside the Matrix reality and don't matter.

Does that help at all? Did I miss anything?
Anyone else want to sign this?

Yes, that helps. Thank you.

The only thing that I still have issue with is the source code thing. If you go with this interpretation that the "agent" and the "agent's icon" are the same thing, I think that necessarily leads to Dashifen's conclusion that if your Agent is crashed or lost, it is gone. There is no loading it again from the source code; you had one copy of it, you sent it away, it crashed, it's gone.
And I'm actually fine with that interpretation.

But it still doesn't prove that the other camp is wrong.

Well, no. It can't "prove" anything. These are my conlcusions.
And there's no evidence to contradict it, which is the point of this thread.


As for the souce code confusion, yeah, I original felt that way, too and handled things just like Dashifen does. In fact, I think it makes the game better. But strictly speaking, we have rules for Crashed programs. We know what happens when a program crashes and an agent is just like any other program. So my former interpreations and Dashifen's current one is against RAW, but a really cool house rule that I feel helps promotes game balance and gives hackers a view that they are "sticking it to the corps" whenever they trash a good piece of IC.