As I suggested earlier: responsibility. An Agent is only responsible for one persona's worth of stuff: it's own. The Agent is doing things more or less just like a persona. Very small and reasonably fixed amount of system load. The IC, on the other hand, is responsible for the entire node, regardless of how large the node is. As a node grows into an immense mainframe with thousands of users, the IC's responsibility grows to continue to monitor all of those thousands of users at once. So as a node gets larger, IC's workload increases but an Agent's doesn't.

I know. It's a puzzle to me too. I just see a problem with Agents eating up so many resources.

Although as I think about it more and more, it all fits in SR4's desire to streamline the world as it were. I can see that treating them the same does make applying the rules much easier, less exceptions always help that. I guess I'd just like it spelled-out better in SR4.

I'm not a stickler for rules, but I do think the details are often overlooked or misread I've been guilty of that myself.

What's the most important for me is that my group is clear so there are no misunderstandings.

Actually my interpretation is following the rules as written.

pg 228 Talking about running Agents independently of the hacker..."In this case, the agent doesn't count toward your persona's active program limits like running programs do but it does count as a subscriber towards your subscription limit."

No where does it say that loading the agent on the node makes the node have it as an active program. In fact the rules specifically state that it still counts as subscribed to you EVEN WHEN OFFLINE and thus can be tracked. Furthermore it only says that it USES the nodes Response, it doesn't say it becomes an active program for that node. IMO the agent becomes a separate matrix entity with its own persona, just like a hacker.

The processing power comes from the host that the Agent is running on, but it takes so little of that power that there is effectively no limit. This is what Grinder (I think it was him) is/was getting at with a difference between IC and Agents. Because IC "expand" with the size of the system they are running on, because they are responsible to oversee everything, they are significantly more complicated than Agents. Therefore, they count towards Response decrease. Agents, OTOH, are basically automated users. Since nothing in the rules talks about a limit to how many users can be logged in at one time, that number must be "effectively" unlimited. Therefore, Agents do not count towards Response limits.

If you go the other way (Agents do decrease a Node's Response) then you are open to the possibilities of the "Agent bomb" attack. Even moreso if you allow Agents to be copied.

On the Agent copying note: TBRMInsanity and I were talking this one out and he and I agreed that Agents should not be allowed to be copied. Mainly, for game balance reasons. However, if you need a "fluff" type reason, here's one: Agents, being complicated independent programs, have a unique "code structure" for each one, individually. This strong "sense of self" is necessary for the Agent to "move" or copy itself from node to node while retaining all state information, its "thoughts" if you will. The individual "uniqueness" of the code runs throughout, so if you want a true "second" entity, the Agent code must be re-written in its entirety. If you copy the code of an Agent and try to run that second copy 1 of 2 things will happen: 1) the copy will fail to run ("That program is already running"), 2) the previous running copy of the Agent will be deleted and will be "reset" back to the location of the second copy.

I might be convinced that if you wish to create another Agent of the same rating as an existing one you can decompile its code (using the same rules as the "pirating software" test) to use as a "basis example." This would provide a dice pool bonus or some reduction in time for creating the second Agent, but not much.

Well what information does a particular drone or Agent need to be on your subscriber list? The Agent must have some piece of information that allows it to find and communicate with your Commlink, be it your Commcode or your Access ID. On a relation to RL level, I never imagined IP address coming into the SR rules anywhere. I saw your Commcode as a combination phone number/email address (I'm pretty sure the rules state as much). Your Access ID I see more like your MAC Address. IP Addresses can and often do change depending on many things but a device's MAC Address is always unique.

My end point is: if you have an Agent on your subscriber list, it can be used to Trace back to your Commlink.

Here's the thing: if #2 goes away, I have no problem with #1 & #3. Under SR3 a Deck (personal computer/electronic device) had the limitation of Active Memory for simultaneous running programs, Matrix Hosts had no such limitation, they could run whatever the heck they wanted. I believe that the Response Decrease rules were a conversion of the "spirit" of the previous Active Memory and were never intended to apply to Matrix "Hosts."

I refuse to believe that Matrix tech between 2064 and 2070 regressed to the point where only 6 enemy or friendly IC/Agents can run simultaneously before performance is affected. The move to the Wireless Matrix didn't replace all the old hardware, it merely created Wireless access to that hardware (this is how I see it), so I think it's ridiculous that the same computer systems are now somehow significantly less powerful. I refuse to believe that the whole Matrix runs on modified Meta Link Commlinks (that is the cheapeast way to get a Rating 6 Node right?).

This paragraph is specificly for matrix nodes running IC. Its not about any low level devices or comlinks. Its about full blown matrix hosts.

I would not use this rule either, but it has to be crystal clear, that this (full blown matrix hosts have infinite resopnse and only efective responsefor game mechanics) is a house rule then.

And one further note: Forget the fluff timeline consistency. We have a brandnew ruleset here. There is no reason to argue anything using timeline arguments (old cyberware, old matrix, old whatever). We have a complete reality backswitch. The fluff about Crash 2.0 is there just to make it bearable.
Argueing about dying characters because of changed essence rules are sensless. As is arguing about matrix nodes after and before the rule changes. (Reality backswitch = it has always been as it is now in the rules)

Well, according to the RAW any device that can be accessed through the matrix is a node, therefore the rules on page 228 DO apply to both commlinks, devices and full blown matrix hosts.

Since each node is a device in its own right I do believe that one has to think in the SR3 line with different nodes. A corporate system for example is not 1 full blow server/commlink – it is a network of nodes that share programs between them.

By the RAW programs take up response if they run too many programs and I see that as a good thing. My hacker for example will never run things at maximum response since he wants flexibility, but he can still run up to nine programs including two agents running three programs each.

Sure, they are all rating four but in the end that’s just one lousy D6 I loose if I want to go number crunching. Corporate hosts and matrix hosts will probably very seldom run more than 4-5 programs since they don’t need to run many active programs. Most of these nodes are storage devices and the only things they will probably run is Analyze and 3-4 IC programs.

Now, let’s get nasty and say they run rating 6 node and decide to have 11 active programs, that’s -1 response. What could we face as a hacker?

1.Analyze
2.Agent (Analyze,Stealth,Attack,Armour)
3.Agent (Analyze,Stealth,Attack,Armour)
4.Agent (Analyze,Stealth,Attack,Armour)
5.Agent (Analyze,Stealth,Attack,Armour)
6 IC: (Analyze, Track, Medic, Armour)
7 IC: (Analyze, Track, Medic, Armour)
8 IC: (Analyze, Track, Medic, Armour)
9 IC: (Analyze, Track, Medic, Armour)
10: IC: (Stealth, Attack, Black Hammer, Blackout)
11: IC: (Stealth, Attack, Black Hammer, Blackout)

The Agents and IC are all rating 4 and roll 8D6 each and they ALL run Analyze constantly. Do we really NEED a higher rating on most civilian systems? Also, remember that they could be running other agents/IC from another node and roam inside this node without taking up its system response.

Also, the node might have firm instructions only to run a specific set of programs and will delete any illicit program uploaded that does not fit its instructions.

Hopefully we will get a lot more information about how node networks would look like in unwired.

I've got it! IC=Agents and Agents=IC right? An Agent uses the Node's Response as its Response limiter, because it runs on the node right? So if IC=Agents then IC can be swapped for Agent in the sentence above. The Agent bomb attack will not work. The system will simply say "I can't run that." Done. A node will not run more IC than its Response can handle, and it will not run more "intruding" Agents than it can handle. Done. Rule exploit plugged. Everyone is happy.

If only it were that easy. An agent running admin privileges could break that rule fairly easily. A node could have a rule "No more than X agents running, period," but then the agent/hacker with admin could rewrite that rule.

I'll bite. Which operating system does not allow an admin (root) user to modify anything, including the system files?

We're making it RAW because the RAW rules are so vague that it can be interpreted two different ways. Just because you interpret shitty rules differently doesn't make your view point any better.

They seriously need to come out with FAQs.

windows 2000 or later, if you screw around a bit with the settings.

linux, with the SElinux stuff added. and btw, thats a NSA creation...

basicly, admin today as a account just like any other. its just that by default its the master controller of all things on the os. but if you play around with the settings, it can be just another account...

What is the other interpretation?
The only interpretation consistent with RAW regarding this aspect is:

Agent = IC
comlinks and nodes suffer response penalties from running too many agents/IC

There is NO other interpretation which is consistent with RAW, and on first look, I see no difficulties with these rules. Thats why I think they are meant that way. Only on second thought, they are flawed (DoS attack, weak matrix hosts, you name it . . .)

And I think we do not only need a FAQ, but we would have needed unwired as the first book to be published. It is a severe mistake not to do this.

BTW: I think it is really sad that not even one of these "rule guys" or "freelancers" is commenting on this. Were the rules written by Rob Boyle only? Where are the guys writing unwired at the moment?

God I hope not, the constantly changing rules from version to version is bad enough.

If you want an official rules clarification, you email the official people. Rob doesn't want to read Dumpshock because it could create issues with claims of copyright and intellectual property.

Well...that, and no offense intended to anyone, but it's been shown that some of the devs can't handle it here.

Seriously. If you can formulate the debates weve been having here into answerable questions, just email it to Rob and hell probably answer you.

(sorry for lack of apostrophes, my keyboard is being wierd)

That may be an understatement.

Hahaha nice.

I'd still wager even a *large* infrastructure setup in 2070 wouldn't occupy nearly a tenth of the space it does today. Considering you can implant a commlink in your head with zero concern of heat (go-go-superconductors) really means you can pack more in less space than before.