As clarified previously, it's not the test that's the concern, it's the timeframe, which remains ambiguous. I wouldn't make a player roll to walk a mile, but I'd at least have mental timeline on how long it'd take him.

I wasn't suggesting he'd abuse anything. Just, that an AI is a complex and complicated enough character, that without at least the GM being reasonably experienced handling Matrix actions, I wouldn't advise allowing an AI PC.


A meat hacker, even if traced right back to his or her commlink, can simply turn it off and drive away. Or worst case, physically throw the commlink off a bridge, and go get another one.

An AI can't usually do either of those with their home node. Not with nearly as much ease, at least.

With Black ICe, the meat hacker can't disconnect without a complex action and test, while the trace gets the equivalent of 2 free hits against the hacker. If the hacker loses combat, the trace is a guaranteed thing, and the hacker is at least knocked out, if not dead. Basically, as it seems now, the AI can get away from any combat where it doesn't get killed in the first pass, where the hacker has no such luxury. I'm not suggesting that how it seems is how it actually is, only that it's the way I understand it currently.



I'm one of those types of people who jumps right into the thick of things to learn. Might not be the best way, but Im figuring by the end of this campaign, the group will have at 2 people who understand the Matrix, and that's two more then we have now.

Logging out without wiping the Access Log - which might be Encrypted, making it take quite a while - leaves you open to potentially being found. Also, the system may retain a Restricted Alert against the AI's Access ID for some time - perhaps even hours - making it very very hard for the AI to hack back in.

Duh. Just encrypt the damn log. lol. <3 You guys are great.

Note, it may only take an action or two for the AI to decrypt the log - but that's another action or two for the IC to run attack on him, too. And makes him decide "take the hits so I can run without trace, or, load up attack and actually try and take this annoying S.O.B. down?"

Unless you find me a rule for longer times (it might exist), the complex action deletes the file in question. The old %-bar moving up while the bad guys try to get the file is very dramatic, but does not jive with the general disregard for file sizes.

SR4a suggests a simple action per turn until the edit is complete for larger tasks, and a single complex for simpler ones. I'm not sure which "Delete Everything" falls under.

I mean that it can take as much or as little time as you want. Technically with the removal of memory accounting (SR1-3, and thank god its gone) data is insignificant size, unless you want it to be (like encryption in the core book vs encryption in unwired). Do whats best for the story. You can even throw some real world at it with having it do a drive wipe which should take a fair amount of time but recovering it today is theoretical. (ie too much time an resources to bother). You can throw in as many tests (or as few) as the story needs. Its a terrible answer I know but thats what comes with abstraction.

[edit] but for a large corporation in the real world loosing a data is more than going in a deleting it (you got backups on backups on backups, 3-2-1 or better), but this is tron so do whats best for the story.



While I agree with you about avoiding the real life stuff. Its a kerberos (or whatever they use in the future) key. Microsoft has some pretty good documentation on technet for their implementation of them. Look up Active Directory SIDs to see the big picture.

Multiple Edits are for generating content - a delete requires little user input.

Please note this is my opinion and not from any books:

For the deltree - type action...

I would have it take (System) Turns for a communit or similar device or (System * max users) Turns for a Nexus. The delete action would not require constant supervision once started, however without supervision it could be interrupted by a Spider before completion. I would also say that with Admin level access it shouldn't require a test, but with any lower access perhaps a Hacking + Edit (2 for Security Access, or 4 for User Access).

-D

I STRONGLY suggest waiving the 'successive extended test -1' rule that Anniversary edition introduced, but only in regards to Tracing. It basically broke Tracewars, in that the track test is extended but defending against it is NOT. A tracewar (trace user vs redirect trace) should be a hack-off, but it shouldn't be one-sided.
If someone is just changing their access ID repeatedly, it does make it harder to find them, but if they're doing it in the middle of combat they're basically shooting themselves in the foot.

For one, see passive traces in Unwired. For second, spoofing your access ID to a new one just happens to disconnect you from all nodes, and resubscribing takes either a simple or complex per(i forget which). It will basically wreck your PAN configuration, unless its smartly networked.


Doing it Wrong. Black IC only works against users in VR, and very specifically doesn't work against Agents/Sprites, etc. (231, no effect on X...). An AI is basically a better, self-aware agent, and also a Program, so it claims immunity. Do keep in mind that if someone is in AR mode, it doesn't work either.
Otherwise, you're doing it right in terms of tests.
You basically want Attack to damage any wierd programs like sprites or agents or stuff black ic doesn't affect.


Slightly fuzzier territory. This is expanded upon in Unwired, where users/spiders/ic can take actions to boot specific users from the system. Yes, it took unwired to introduce Kicks. I think you need the Account, but I'm not sure. Also relevant is the Alert section from the same book - if an Alert is successfully triggered against a user, they lose all access rights AND everyone can see you. From the faq:




Edit is indeed the program for altering files, but there is a distinction between levels of Access Rights that is rather relevant here. If you have good user rights, you will just succeed automatically without a test. If you don't, you will have to roll against system+firewall to attempt it.
The corrupt program and action from unwired is also extremely relevant, and there is also a virus which can delete everything on a system in the same book, but takes a few minutes to work.


In fact, I'm just going to pretty much recommend that you read the first 90 pages of unwired. Everything about things like matrix topography, how systems are set up, data transfers, expanded rules on alerts and subscriptions, the security example setups.... basically goes a huge long ways towards making it easier to wrap your brain around the matrix without making it more complicated. There's information there that's basically missing from the core book (my favorite obscure text is the one that lets you go above your subscription limit in exchange for program lag, which actually makes pans kind of feasable), and I'm not even talking about crap like echoes or new gear. Just clarifications.

The developers actually left the tools to solve this problem easily. Most people just forget about them.
They're called Node Scripts.

The files that do things automatically on nodes, like ejecting your smartgun's ammo clip while it's empty.
You can pretty easily have a script that 1)stops logging, 2)deletes your jack out timestamp 3) Self-delete, sending a command to restart logging. and run it with Administrator authority.


Just so you know, Decryption takes SIGNIFICANTLY longer than that. Multiple Combat Turns is rather long when all combatants involved get 3 passes minimum.

Try that when one of your players who programs flight sims for a living is the hacker........

The problem with node scripts, Udoshi, is that they're *too* powerful. Have you tried using them in Eclipse Phase? Talk about god mode. :/ But yes, you would expect people to be using 'cleanup' exits like that.

I GM'd EP before I GM'd SR.

Fell in love with the universe. The char creation left a lot to be desired.

Yeah, it… poses unique challenges. :\ Honestly, the most fun we had with it was the very first game, no knowledge of the setting, just handed us charsheets. I forget the name of the scenario, you wake up alone on the haywire space station. But anyway. I'd want to limit scripts in *some* way, for balance (not 'realism', oh well).

Eh. If it's a rating 3 or 4 Encrypt program, a threshold of 6 or 8 shouldn't take longer than 2, MAYBE 3 combat turns for an AI equipped to handle it. And you only need to use one Complex Action to start the process - it's automated from there. Going full-defense and loading up a Medic program might be enough to buy time for that decryption to work.

Same. I played it once at a con, had a blast. Bought the core book and sunward right away. No one was playing it in my area, so I started GMing it, and never got to play it again.

I am not familiar with Eclipse Phase, sadly.(i haven't had the time to dig into it, but I've heard its good) how do they work in that system?

I'm of a mixed mind about nodescripts. On one hand, they make sense for a lot of things (gridguide, searchbot agents, smartguns, even vehicle sensors and stuff) but they can be rather powerful if they aren't limited at all. (personally, I tend to say there's a limit of auto-actions a node can take by itself in any given turn, so there's a bit of a limit to 'oh the node does a billion things at once')

On the other hand, a lot of hacking these days is script based anyway, and, at the very least, an exit script is probably hacker 101.

Maybe if those scripts were treated as a kind of malware / virii, in terms of mechanics? Something that acts as a temorary Agent sort of thing; the higher it's rating, the more complex the instructions you can put into it (so the "typical exit script" described here would be R1 suitable) ...?

Yeah, powerful. They work about as you'd expect, which is the issue. You prepare a bunch of things ahead of time, waste X time uploading the script, and then it just goes nuts. It's not so much the number of actions *at once*, just the 'free' actions. One thing that (IIRC?) EP does is require you to 'code' the script, which at least requires some effort and gives you a limitation.

That's interesting, Pax. Hm.

I could have sworn there was a sidebar dealing with scripts and suggested guidelines on how elaborate they are, but I can't find it. I DID discover its a software+logic test to make, though.

I think a more easy solution is to treat a node kind of like a pilot - it has its own initiative and limit of actions a turn, but can only spend them on scripted actions.

I think the better solution is more along Pax's lines, which I like. Something along the lines of putting each hit on creating a script towards a dice pool, in the case it needs to roll something, which also makes multi-command scripts harder to throw together.

Ah, there *is* a test for it? Then that's like EP after all. Yeah, I'm still interested in the other suggestion.

Well. there IS a test for it.

With absolutely no details or guidelines on how it actually works.

How do EP scripts work?

Well, I think this will give you an idea, even if you don't know the rules system:

Okay, a little bit stream-of-consciousness here ... brainstorming an idea.

Have the player define the script s/he wants, step by step, in list format. (Or as a flowchart, if they want to go that far.)

Now, go through the list. Every action that would not normally require a roll, is worth +1 point of complexity. Every If-Then decision point, is worth +2 points. Every action that would require a roll to achieve, is worth +3 points. The Script has a Program Rating equal to half it's Complexity, round up; the programmer can freely increase the complexity, without adding actual instructions, in order to improve that rating.

Coding the Script requires a Software+Logic extended test, with a threshold equal to the script's final Complexity, and an interval of ... does 1 day sound good?

Using the script requires uploading it to the node, then commanding the Node to run it. Any die rolls the Script needs to make use a die pool of it's rating, plus the Response of the node it is on.

...

Does that sound like a workable starting point?

Sounds a little complex. It probably shouldn't be able to perform any logic, just attempt tasks in order.

For consistency, it'd be nice if you could make the Software test based on Rating, and then the number of tasks is based on Rating (instead of vice-versa).

Probably the node should be making the tests, not the script? We might want to examine actual examples to decide.

Hey, Ryu, what book are you referencing? I can't figure out what RW is (the only thing I could think of was "Running Wild," but that doesn't make any sense).

While I'm not sure how Ryu made that typo, the rules being referenced are in runner's companion, on those page numbers.

Ah, I should have figured. Thanks!

Answer already found. Thank you sir!