Meh, I would tend to think the in-your-face threat of having your data trail tracked and your hide-out blown up sort of makes up for it

Traces can be delayed or interrupted with little real retaliatory threat. Gray IC could produce expenses in a single action, and cripple the decker for the rest of the run.

Gah. Nice catch.

Still, agents use Pilot for Hacking so they still fill the skill requirement of the sidepool to change the access ID in 4A.


One day, i swear I will go through both books and list out the differences between the two damn matrixes. This shit is damn annoying already without half the debaters using different systems.

That was what the errata was supposed to do, but there have been none since the financial upheavals and no explanation why.

Hell, some recent reprints have managed to revert to the pre-errated version...

No. SR4A very clearly identifies personas as only belonging to individuals, as they are firmware associated with a commlink that allows a user to interact with the Matrix. Agents are only represented by icons. They only exist in the Matrix, and so cannot use a persona to interact with it. Nor do they need to.
An Agent's abilities change depending upon which node they are in. Personas do not. Personas are static representations of the commlink you are using.

The Hacking + Spoof Test is to ensure that one's connection is not jammed open. This is irrelevant to independent Agents as they are not projecting from one node to another, but are running upon the node they are on. Agents may not have icons in multiple nodes like persona programs can. Therefore Black IC cannot jam a connection that does not exist. No roll is necessary, required or possible.
This text was writen using 4E rules, and have been transplanted by SR4A. But, even if one were to use this, it would only change the Access ID of acommlink and not the unique Access ID of the Agent.

This is only possible with multiple hackers. Someone must be using these commlinks in order to load said persona. These Agent will only follow the commands of the persona upon whom they are loaded. This is a common technique for Hackers to protect the party, but a persona does not run independently, it is a firmware program that requires a user.
Additionally, an Agent loaded onto a persona goes where that persona goes, so unless the persona is also hacking onto a node, the Agent can't log onto that node independently.


No. Spoofing an Access ID is a Software roll. Agents have no Software skill, so they cannot make Software rolls. Even if they could, this would only change the Access ID of the commlink they are in, not the unique Access ID of the Agent.

These rules are explained clearly on pages 216-247 of the book Shadowrun Fourth Edition, 20th Anniversary Core Rulebook in the chapter entitled The Wireless World. It explains everything fully and in detail.

I don't know about 4e, but SR4a says they can't whilst Unwired says they can.

I don't know where that is, but I'll take your word for it. However, the most recent rules are the rules.
One is welcome to play an old edition, but to ignore the solution to a problem and then complain that it wasn't fixed seems silly.

EDIT: It's the first sentence of the second paragraph under "Node Movement and Action" on p. 110 of Unwired. It has been replaced/overruled by the sentence you quoted, which is verbatim from the last sentence on p. 234 of SR4A.

Interesting. Though fully aware that SR4a came after the core supplements I'd instinctively considered the supplements to overrule it. What you say makes perfect sense, though I wouldn't necessarily bet that most of the changes from 4e to the supplements were not undone in 4a.

Alright, blame that one on my insufficient knowledge of 4A. I have still been too stingy to dish out money for an update which should have been available as an errata.

So an agent can't open a connection to another node, it can only travel to another node? That won't work whenever the agent doesn't actually have access to a node, and so an agent could never independantly hack another node, because it can't run on the node it's trying to hack. So to me it seems agents must be able to open connections, too. (And Unwired very precisely says that it can do this, even multiple connections to multiple nodes, on Page 110. Now I'm assuming 4A contradicts that...)

Finally, Unwired says that an agent's access ID can be spoofed prior to loading it into a node (by the user, I suppose using the updated Software+Spoof) roll. And as long as I can do that then I don't even need to actually physically change the hard-coded access ID of the agent.



Again, 4A might contradict this, I don't know.


I don't need an independant hacker to control another commlink, I just need one to control it simultaneously. Most certainly the agent can't act independantly, controlling the persona, but once a connection is opened, the agent can slow-hack by itself, for instance, without needing further input, and I can have several personas with linked agents do a teamwork slow-hack, too. It's not nearly agent smith territory, but a definite benefit, and all I need to do for that is pick up every commlink I see on a dead body, hoping to get an R4 and up. I could also buy a nexus and run multiple personas from there.

It does, as discussed above. Unwired had problems. SR4A fixed these problems. Do not simultaneously complain about a problem while purposefully ignoring it's solution.


You can't control two or more persona simultaneously. Slamm-O! discusses this on p. 91 of Unwired. No matter how many commlink you have, you only have one brain.
A persona isn't just a commlink, it's a user's interface provided by the commlink. You need multiple people to have multiple persona. Moreover, each persona would need to hack the same node in order to get it's attached Agent into said node, so you'd have to roll to hack the device for each persona and each time you do, you risk putting the system on alert.

Again, this is all spelled out very clearly in the basebook. If you read it, it will answer all of your questions.

You can however have up to six hands, and comlinks work fine through manual control as well, meaning you could run up to 6 'links just manually, plus one through DNI. Potentially other methods of control may add more.

How's about an anthroform drone (or even just one with some hands) controlling a comlink?

I'm fairly sure that agents can operate just like other users on the matrix. The issue that is so confusing involves their ability to travel around. AI's have this problem as well. Basically, there are three modes an agent can run in. It usually starts at 1, just after you loaded it into your commlink's memory.

1) Running Normally.
2) Connecting to an account elsewhere with the Log On(or hacking) action. This puts an Icon there, but doesn't mean its now Running on it. When you think about it, this is actually a necessary state for an agent - else it cannot just 'data search the matrix', and nothing explicitly prevents an agent from using the Log On or Hack On The Fly(which includes a free log on as part of it). Thus, agents operate like any other user, and can make connections to other nodes. (Note that AI's operate this way as well - nothing forces them to expose themselves to the vulnerability of a node crash just by being there)
3) Roaming. It physically transfers itself to another node. This may change its stats, especially if the system or response of the new node that it is now running on is lower than before.
TLDR: You REALLY NEED to keep track of where the agent is Running, physically consuming computing resources from the Processor Limit. (remember that the Agent counts as a running program itself, in addition to the programs in its Payload).



A better option would be to use a Nexus. As I pointed out earlier, each Persona on a device has its own access ID, in addition to the hardwired ID of the node. An agent loaded onto a persona uses that persona's ID, and nexi are specifically made to have more than 1 Persona Limit. Paired with the ability to set programs on a nexus available for public use by its users, well....
This is likely the exact setup that corporations use to set up IC Farms and their Botnets. Fairly inexpensive, and effective.
Clusters are able to do this to some extent as well, as the persona limit of 1(default for most devices) adds up quickly.

No. This was the case in Unwired. Fans pointed out this was broken, as you are doing now, and that Catalyst should fix it. Catalyst then fixed it. They wrote it down and published it on the 234th page of a book called Shadowrun Fourth Edition, Anniversary Edition.


The details on how persona work, and how this suggestion doesn't, is fully explained on page. 223 of the aforementioned SR4A. It explains in full and explicit detail how persona work.

No, that's wrong. Ok so I went out and got 4A. Now let me check:

The Persona is just a bit of firmware. Each commlink has one, a nexus can have several.


So that means that every commlink has a new persona program. I would interpret that to mean that there should be one dominant one, the one which you access with your trodes or DNI, and several others, but I haven't found that actually spelled out anywhere - especially seeing as how you can control a commlink with a good old-fashioned keyboard, too.

The paragraph you are adressing in Unwired refers to daisy-chaining vs parallel commlinks. Now all that shadow-talk is just fluff to me, because there are no rules to represent what you have there. While it's true that having multiple icons with DNI in multiple nodes will leave you vulnerable in all of them, you can just disconnect your auxilliary commlinks from your DNI, and then you are completely safe. You could even leave them at a different place and let the agent connected to that persona do its thing.
Which it can, because:


So what I do is this: connect auxilliary commlink. Load agent into persona. Connect to a node and tell agent to slow-hack it. The agent stays loaded into that commlink's persona program. Then dump the commlink somewhere. Rinse / repeat. I can come again after a few hours/days and check on the agent, or do the entry hack on my own. I don't need to dump the link if i'm not afraid of being traced, or, for instance, I have another agent scanning for traces (on a neighboring node).


Again, this isn't nearly an agent smith scenario, but it still means I can very generously slow-hack a lot of nodes, even without mass-probing.

Oh for fucks sake. You are wrong. Thank you for quoting pages for me, so I don't have to look it up myself.


Notes that it says Nodes.

Continuing on from the same quote block...


It cannot do the functions it says it can without Accounts, therefore it can Log In or Hack its own. And says it can.
When operating with your persona, it can access multiple nodes, because YOU can be in multiple nodes, and it can help you in any you are in.
When operating independently, it can only be in one at a time. This is fair, but only limits an independent agent to one connection at a time - it doesn't force an agent to Roam. When operating independently, an agent has its own icon. When making a connection to another node, without physically loading itself onto it, its icon is present there - but its not taking up processor limit.

In fact, if you WANT it to roam, you need a special setup....

Agents do not roam by default, you have to tell it to do so specifically - by loading it onto another node. Only then does it run independently.

Yeah, sorry dude. Everything you've pointed me at supports my arguement.
The only-one-connection-at-a-time nerf IS a nerf, but its not a very big one.
This only reaffirms the NEED to keep track of where an agent is running at any given time. And not just where, but which settings its using.

I would also point out that nothing prevents an Agent from hacking itself onto someone ELSES persona while roaming in order to gain increased functionality, as long as it has the permissions to do so.

This node can be an enemy node by the way. The permissions do not have to be legally gained.

It seems y'all have jumped from one statement to another that is, frankly, hard to follow. Here is what I am trying to say:

Agents use a unique Access ID that cannot be spoofed by the Agent as it is a Software test or by a hacker as a Spoof Access ID roll only changes the Access ID of a commlink and not an icon.
The only way to change the Access ID of an independent Agent is a week long extended test as detailed in Unwired. This prevents the infinite number of Agents per hacker scenario and makes the game playable.

What you seemed to be saying was that Agent's Access ID's can be Spoofed easily and that this creates an Agent Smith Scenario. This argument has been abandoned.

Now what you seem to be saying is that a commlink is the same thing as a persona and that once an Agent loads itself onto a different node, it can then load itself onto that node's persona and change it's Access ID, thereby creating an unlimited number of Agents.
This is not true. Each commlink has it's particular commcode, but a persona is limited by the individual using it. When that person is logged off, there is no persona as there is no need for the firmware to be active to allow a user to interact with the Matrix, as there is no user with which to interact.
Only a persona may load an Agent onto itself. An Agent can only load itself onto a node. Therefore this run-around to create an infinite number of Agents won't work either.

While in a node with a persona or other icon, an Agent may assist that persona with a Teamwork test. So, yes you can use your bot-net to give your Teamwork dice if you load your persona onto the nodes which each Agent in the not-net has access to.
This does not mean the Agent is loaded onto another node. This does not mean that the Agent can run cybercombat from it's node onto the node in which you are hacking.
An Agent may only be on one node at a time. This is a blanket statement of fact.

I'm not trying to prove you wrong or win an argument. When I point you towards a particular chapter, my goal is not to say I'm right and you're wrong, but to point out the truth of the mechanic. I'm trying to point out that the Matrix system isn't as broken as you seem to think it is and that the rules support a fully functioning Matrix that shuts down rules lawyers who try to work their way around it. Yes, there were problems with copying Software, but these problems have been fixed.

If what you are trying to prove is that Shadowrun is a horrible, unplayable game designed by morons whose rules make no sense and should all be burned, then I shall cede the point. I suggest you not only stop playing this terrible game, but stop posting on it's fan boards as well.

Okay, people keep on pointing that out like its a holy grail of arguement defeating. But its not.

The 'new' 4a roll to change an ID is Hacking+Software.
Agents do not have Software.
Agents use Pilot for Hacking.
You default to program-1 on the matrix, not skill, because program replaces attribute.
Software is not a program.
They do not NEED Software to make the test, because they have the other skill, Hacking, and thus actually have a dice pool.
Test is declared, dice pool is figured out.
0 dice for software. Rating dice for Hacking. Therefore, an agent rolls 1-6 dice to change an access ID under Anniversary rules.
In 4th edition rules they roll Pilot+Spoof, and cannot make the test without a spoof program
As I pointed out before, agents may be loaded onto persona in order to use their ID, and may also do so illicitly if they have the proper permissions.

Therefore, an agent is still quite capable of botnetting, but its not as easy as it was, as only high rating agents stand a chance of making the necessary test to get a new ID on a commlink - and they must be a mook to do it, because it takes the hacking skill.

Patching an agent takes Logic + Software, which under a general rule that agents roll Pilot for any tests involving normal attributes, would allow them to default. Unfortunately it says in the book that they can't do that.

Unwired at least maintains that the user can spoof their access ID when loading them onto his node, and they then keep that access id. Interestingly enough it is not clear what happens when that agents self-replicates - I would assume it replicates with its original hard-wired id.



I think it's safe to say that they can't self-distribute a botnet of unique agents by themselves anymore. I still believe they can be spoofed by the user, or multiplied into several independant unique agents via multiple commlinks. This is a significant gain to PC hackers, but not totally out of line, because it is to be assumed that most opposition will have cheap access to unique agents anyway. You just have to use them as a GM.

Which means you can't use it on the Matrix. You can only use programs on the Matrix, not skills. If you want to use 'skills', you need the appropriate Autosofts. Software is explicitly excempt from being available as an Autosoft. This means you can't default, because the roll in itself is an impossibility.

Udoshi:

I don't see how this says what you seem to imply it does.
You can access nodes as much as you please, because you're an actual user, and have a persona.
All it says is when you're accessing a node different than the one your persona is running from (ie. your 'link) -say, looking for a book in the public library node- and you happen to have an Agent running in the node your persona is running from, this Agent may help you with your Data Search action from the node it's running in, without having to log in to the library's node as well.


In order to run on a node, without being loaded into your persona, the Agent has its own Icon there. It may not have icons in multiple nodes. This means if you tell the Agent to go look for this book in the library by itself, it would have to Log On to the public library node. Since it can't have an icon in multiple nodes, it will have to stop running on your 'link, and start running on the library node. This will cause it to assume the Matrix Attributes of the library node. It will also make all trace of it running disappear from your 'link, and show up on the library node instead. It will also use up processor limit on the library node, instead of on your 'link. If it subsequently wants to report its finding back to you, it will need to move back to your 'link's node in a similar fashion.



...Or by the Agent moving itself there, when it needs to access another node. Because it has to move in order to access it, to satisfy the limit ofnot having icons in multiple nodes, as I explained above.


Except that loading a program into your persona can only be done voluntarily, through using the appropriate action.


Indeed. When you have hacked your way into an admin-level account on the public library node, and are currently accessing this node through this account, and have an Agent running on the 'link you're using, this Agent can do anything admin level gives it clearance for, respecting limitations generally valid for Agents. If you tell it to look for any user accounts owned by people named 'Suzie' and delete them, it can, without any further intervention from you, because it's riding off your clearance, using the Attributes of your 'link.
However, if the Agent wants to do this on its own, it needs to Hack its way into an admin-account, Log On, again moving itself from your 'link to the library node, switching to the Attributes that has, and running on its processor limit.

You're both right and you're both wrong, at least by the letter of the rules. Honestly guys, this was an old arguement and caused some discussion when SR4a first came out.

Unwired's description of how agents function independently directly contradicts with SR4a's single sentence



It's sad but true. Now people argued SR4a was the later edition and thus superseded Unwired. Other's (including me) argued Unwired is listed in SR4a as advanced rules, thus they were more extensive.

Saying that people complained about Agent Smith and they wrote that in sentence in SR4a for just that reason is speculation, unless you have some evidence that's why that sentence is there or some interview with the writer themselves discussing their intention (by all means I'd love to know)

Actually the Unwired errata that disallowed you to spoof access ID when loading an agent was the extent of stopping Agent Smith. Nothing stops you having multiple agents, or even a botnet, but copying a single agent and bypassing rules for unique access ID's logging into the same node is pretty much dead.

You can copy an agent, load it on all your teammates 'links, and it is individual for all of them provided it is run with their persona. Independent agents however don't get that benefit, and thus must have software rewritten. Agents cannot default on skills they don't have, and they don't get software skill.

We can guess what the writers meant all day, but even using icon as the definition is ambiguous to start with. I can easily argue that the Agent program Icon only exists where the program is running, and all other Icons associated with multiple connections are Icons of subscriptions. Therefore the agent may only "run" on a single device at a time (using that devices attributes). Copies of agents running on independent nodes have their own icons, but share an access ID.

Edit: In effect i'm saying that rule disallows a single agent program to Load itself on multiple nodes at once, effectively copying itself without copying itself...

Ah, an unwired errata, it's good that you mention that. I'm notoriously bad with knowing about these pesky things... .

Here we find the following goodies:

Concerning Personas:


So there ARE rules. You have to switch, but you CAN have several. I don't understand the modifier thing. This looks like shoddy editing, once again, as it should be a -4 DP mod, I suppose, since a +4 threshold mod is completely out of line. And in any case threshold mods are only normal with extended tests.



Ok, so there I am finally actually contradicted. Which is fine, I am content with using multiple commlinks.