 Agent Smith Problem, The problem and how to solve it |
  |
 Feb 3 2007, 09:45 AM
Post
#26
|
|
|
Moving Target  Group: Members Posts: 870 Joined: 2-October 06 From: Athens Ga Member No.: 9,517  |
Yes Irony abounds!
Personally I think that making agents only able to run from a system makes the agent back into what I think that they were designed for - an interface. Basically you ask your agent to do something instead of doing it yourself. Treat it more like a Pilot program that drives your connection around the matix for you. It makes sense in the future where literacy is down and people are wanting the easiest way to do anything. We are moving away from requiring much in the way of computer skill NOW. If your agent is just a way to talk to your commlink then all these problems go away. You have no problem with lots of agents. You have no problem with people giving hacking programs to their agent and it doing all the work for them. It returns things to the mode of a skilled person subverting the system. Agents go from an imbalencing loophole to a cute toy. |
 |
 
|
|
Feb 3 2007, 10:01 AM
Post
#27
|
|||
|
Moving Target Group: Members Posts: 745 Joined: 2-January 07 From: Los Angeles, CA Member No.: 10,510 |
This is kind of what I've tried to do with them, actually. The only 'agents' in my Custom Hacking Rules are viruses (which I haven't written up yet), and the lowest wrung of viruses are "rats", which are minor agents used to perform various tasks for users in the Matrix. Rats are so abundant in the Matrix, that it's really hard for legitimate users to use public nodes anymore and so rats become the standard. Meanwhile, the Matrix and its associated nodes work hard to innoculate themselves against these rats once they've realized its a rat and not an actual person, so each rat has to be custom programmed and can only be released "into the wild" once and lasts about a day before the Matrix has adapted itself. Due to their inefficiency and their restriction to unprotected nodes, the rats take about a day to fulfill most tasks. Basically, they're little "toys" for players. They're one-use agents that perform a function on their own as a sort of time-saving device. They take longer to work, but the character doesn't have to do it directly and can do other stuff. Rats are easy to order online from a multitude of vendors. You can use the Software skill to make your own. |
||
|
|
|
||
|
Feb 3 2007, 10:07 AM
Post
#28
|
|
|
Moving Target Group: Members Posts: 870 Joined: 2-October 06 From: Athens Ga Member No.: 9,517 |
cetiah, are you going to put all this up on a web page at some point? It will make it easier to follow I think. Frankly I tend to skim a bit when I'm looking at too many posts. I would like a format that would be easier to read. Plus you can just point people to it and they don't have to go through back posts trying to findit all.
Just a suggestion |
|
|
|
|
Feb 3 2007, 10:11 AM
Post
#29
|
|||
|
Hoppelhäschen 5000 Group: Members Posts: 5,807 Joined: 3-January 04 Member No.: 5,951 |
The problem is that drones can be batch commanded. Which tells us that it is technically possible, and, once you know that, the rest is onply an implementation issue. |
||
|
|
|
||
|
Feb 3 2007, 10:50 AM
Post
#30
|
|||||
|
Runner Group: Members Posts: 3,314 Joined: 26-February 02 From: Lisbon, Cidade do Pecado Member No.: 185 |
As Serbitar has already noted, you can issue batch commands to multiple agents too. In both cases such commands apply to agents/drones subscribed individually. Drones were intended to function just like Agents (though Pilots do indeed combine Agents with a specialized OS per the FAQ) in terms of issuing commands. As such the same limitations apply to both. However, if I've missed some rule about issuing commands to non-subscribed agents or drones, subscribing groups of agents, or not having to individually subscribe agents/drones/whatever to command them on another system you have subscribed please feel free to point it out. In all cases the subscription list limit (of the commlink you are using) on the number of systems/drones/agents/whatever is the cap on how many individual systems/drones/agents you can command (individually or collectively) at one time. Currently it's the cap on how many different links to autonomous systems your commlink can sustain. |
||||
|
|
|
||||
|
Feb 3 2007, 11:07 AM
Post
#31
|
|||
|
Mystery Archaeologist Group: Members Posts: 2,906 Joined: 19-September 05 From: The apple tree Member No.: 7,760 |
Just a minor nitpick here. Why in a world where everyone is on line 24/7 would you have internet cafes? Why do I need to go somewhere to log on when I'm always logged on, all cafes are in effect Internet Cafes. It has just occured to me that there must be some way of allowing more than 2*rating users to subscribe to a node. Otherwise how would Internet gaming like Dawn of Atlantis and other mmorpgs work? |
||
|
|
|
||
|
Feb 3 2007, 11:14 AM
Post
#32
|
|
|
Running Target Group: Members Posts: 1,468 Joined: 5-December 06 From: Somewhere in the Flooding, CalFree Member No.: 10,215 |
Extremely expanded bandwith and response, along with an ultra system rating and multiple multiple linked nodes (like a room full of servers) that function in conjunction with each other allowing a multitude of players to play at once?
|
|
|
|
|
Feb 3 2007, 11:28 AM
Post
#33
|
|||||||||
|
Hoppelhäschen 5000 Group: Members Posts: 5,807 Joined: 3-January 04 Member No.: 5,951 |
No. What I'm talking about is this:
Technically you can have 100 Agents search the Matrix for Infos and only take up one subscription slot. |
||||||||
|
|
|
||||||||
|
Feb 3 2007, 11:44 AM
Post
#34
|
|||||
|
Running Target Group: Members Posts: 1,498 Joined: 4-August 05 From: ADL Member No.: 7,534 |
Thats why I think the agent smith problem is nto a problem when used by hackers. Stealth is everything. |
||||
|
|
|
||||
|
Feb 3 2007, 11:47 AM
Post
#35
|
|||
|
Running Target Group: Members Posts: 1,498 Joined: 4-August 05 From: ADL Member No.: 7,534 |
Ill think I will incoprorate my -2 dice rule for everything. being attacked, being traced, being percepted, being whatever atht he same time. |
||
|
|
|
||
|
Feb 3 2007, 11:50 AM
Post
#36
|
|||||
|
Running Target Group: Members Posts: 1,498 Joined: 4-August 05 From: ADL Member No.: 7,534 |
Yes, I totally Agree. thats Why I used a max of 12 agents in my calculation. (well, that was before reading Rotbarts objection). I also said that I dont consider it a real problem, because of the stealth issue. Im with you all the way. (Apart from my opinion that the subscription rule is totally nuts, see the matrix cafe example) My only point left is agent smith on corps side. They dont have to care about stealth at all. Its their node. |
||||
|
|
|
||||
|
Feb 3 2007, 11:54 AM
Post
#37
|
|||
|
Great Dragon Group: Members Posts: 5,537 Joined: 27-August 06 From: Albuquerque NM Member No.: 9,234 |
So assume Aries has 12 major data centers and 6 heavy duty security hackers on duty at 3 am on New Years day. And suddenly 8 centers get hit by a huge deluge of well programed automated attacks, that don't look completely automated because the guy who programmed them was clever. What do the security guys do? How much attention do they likely have for seemingly minor crap going on in the other 4 centers for the next few minutes? Would you like the ability to make the security guys go watch the shinny stuff while you go about your illegal business? |
||
|
|
|
||
|
Feb 3 2007, 11:56 AM
Post
#38
|
|
|
Running Target Group: Members Posts: 1,498 Joined: 4-August 05 From: ADL Member No.: 7,534 |
I personally dont have a problem with this clever use of agents. I dont think it is unbalancing things to much.
|
|
|
|
|
Feb 3 2007, 06:00 PM
Post
#39
|
|||
|
Moving Target Group: Members Posts: 745 Joined: 2-January 07 From: Los Angeles, CA Member No.: 10,510 |
It's up, already. I made a post in this thread specifically covering nothing but rats. I don't know how to reference a particular post in a thread, but it's the 16th post in the thread. Cetiah's Custom Hacking Rules Eventually I'll have a PDF. Hopefully by the end of the month, but we'll see. Real life has really been distracting me from my Shadowrun life this week. |
||
|
|
|
||
|
Feb 3 2007, 06:35 PM
Post
#40
|
|||||
|
Runner Group: Members Posts: 3,314 Joined: 26-February 02 From: Lisbon, Cidade do Pecado Member No.: 185 |
Thank you. I stand corrected. I had overlooked that exception for drones. The intent to parallel agents and drones does seem compromised, but I'll recheck with the author. His intent was for the subscription list to function as a limiter. Nonetheless, until something official specifically states otherwise I suggest you chalk the exception up to some uniqueness of the mixed agent/OS Pilot. |
||||
|
|
|
||||
|
Feb 3 2007, 06:43 PM
Post
#41
|
|
|
Hoppelhäschen 5000 Group: Members Posts: 5,807 Joined: 3-January 04 Member No.: 5,951 |
The problem is that this limiter applies to everything, yet multiplexing is only described as possibility for drones.
The very basic is that there is no seperation between subscription (your persona is there) and connection (only mundane data)... i.e. limiting important stuff and not limiting network traffic. |
|
|
|
|
Feb 3 2007, 06:59 PM
Post
#42
|
|||
|
Prime Runner Group: Banned Posts: 3,732 Joined: 1-September 05 From: Prague, Czech Republic Member No.: 7,665 |
Is it unbalancing to send hundreds or thousands of zombie PCs after a corp you are targetting? Perhaps not. But it is completely impractical to resolve game mechanically. This is a real weapon in the computer hacker's arsenal. But it's not practical to roll dice for every hijacked credstick, coffee pot, and desktop that is sending cascading spam bombs as part of a DoS attack on a tertiary node. In Shadowrun, Hackers should be pulling this sort of thing all the time, and yet there's no possibility of me actually resolving that even once. It involves the rolling of hundreds of dice, and the end result is always going to be the same:
As written, the DoS attack always works, and actually generating it (to determine, for example, whether it takes one combat round or two) takes the entire evening and is about as much fun as generating Anti-Aliasing by hand. ---- So, what needs to be done? First, the idea that you should keep track of individual streams of code running on individual processors around you is horse shit. We can't be expected to do that, because of the afore-mentioned "rolling dice hundres of times" problem. We can simulate that with a more abstract ruleset based on processes and influence. Influence In 2071, a Hacker compromises computers and makes them run as parallel processors to assist in compromising larger and more secure computers and so on. At any given time, you are going to be throwing down with perhaps hundreds or even thousands of separate service calls to computers all around you - assuming that they are available. Your Influence is a flat bonus you get on Hacking based on how many computers you have compromised and how many your system can coordinate. Your Influence can never be higher than the System rating of your Commlink, because your processes have to be managed by a central unit or the parallel processors stop working on problems from different angles and just become a mess. Your influence is otherwise capped by how many computers you can communicate with that are influenced by you. In practice, we assume that a Hacker is usually in communication with enough zombie computers to fill out their Influence pool. But if a Hacker is in prison, or in a dead zone, or in a remote facility where general matrix communication is reserved or blocked, then a Hacker will have to deal with just the devices on his person. Very roughly, Influence increases by 1 for every ten times as many compromised devices there are (after all, you're only using the spare CPU cycles on fairly modest machines like jacket music players). So yes, when in his element, an elite Hacker with a system of 6 on his commlink really will have backdoors in one million devices. That's not even unreasonable, because a lot of those devices are just lights and toilets and such. Processes: Monitored and Unmonitored Hackers multitask like mad. Literally every single thing they are doing online could be automated, but they are totally hardcore and do it better hands on. And they get their hands on all of it, all the time. A commlink can orchestrate a number of separate processes equal to the System rating before it experiences slowdown. But a Hacker can only monitor one process with a Simple Action (thus, two processes can be monitored per IP). During Cybercombat, for example, the monitored processes might be expected to be Attack and Defense. At other times, it might be something else. In any case, a monitored process uses the Hacker's own Attribute + Skill, and all the other processes being run just use the Hacker's Agent rating as their dice pool. --- It's a start. A start on a system that does not expect me to roll four hundred fucking dice just because I'm smart enough to have figured out cutting edge 1992 hacking skillz like zombie PCs and Denial of Service attacks. -Frank |
||
|
|
|
||
|
Feb 3 2007, 07:35 PM
Post
#43
|
|||
|
Immortal Elf Group: Members Posts: 11,410 Joined: 1-October 03 From: Pittsburgh Member No.: 5,670 |
the biggest problem is, it makes sense to allow it. using giant armies of hacker agents is one method actual hackers use to do their thang. i like FrankTrollman's idea. two problems i have with it: first, it creates a difference between rigging and directing agents. no real getting around that, i guess. second, it counts only the number of devices infiltrated, with no regard to their power. what about corporate hackers, who can draw a lot of Influence from high-end security servers? |
||
|
|
|
||
|
Feb 3 2007, 07:57 PM
Post
#44
|
|||
|
Prime Runner Group: Banned Posts: 3,732 Joined: 1-September 05 From: Prague, Czech Republic Member No.: 7,665 |
I'm going to talk about the second problem first, because I think it is easier to handwave. The world of computers is full of all kinds of things. Indeed, while many Hackers are going to be using thousands upon thousands of zombie wristwatches and portable freshness indicators, the security technician on the other side is running sevice calls to a massive server cluster in space. He really seriously does not have to worry about being in Matrix link with a car full of churro warmers - that shit is totally beneath him. From a game mechanical standpoint, his influence bonus is the same. His commlink can and will orchestrate the same number of service calls before it slows down whether it's making those calls to one server cluster or to a physical cluster of ten thousand dancing lawn gnomes in a warehouse. There is a degree to which it makes sense, and a degree to which it doesn't. But it's fairly game balanced, so I'm willing to handwave futuristic technology that much. Exchange rates for more powerful devices could be anything you wanted, but I have no problem with an account at ZO being literally a million times better than a backdoor into the MP3 player on the doorbell. -- The first statement is harder to get around. It's an unfortunate reality that drones exist in reality and programs work only in virtual reality. The difference is that while it makes sense to keep track of drones in terms of real space, programs exist only in virtual space. And that virtual space unfortunately does not have a 1:1 correspondance with real space, even though it would be really cool if it did and the existance of AR means that it almost does. The primary difference is shared location. A drone can be expressed locatively with a set of cartesian coordinates. It has a single location, expressable with an X, a Y, and a Z. A computer program has only a Topological location. There are things that a program can interact with directly, there are things a program can interact with after first interfacing with another system, and so on. JUst because a program is running on the same hardware as another program does not mean that it is necessarily topologically close to that program, and so on and so forth. While a drone can be thought of as playing chess, being in a specific space and moving in predictable patterns, the program is playing Chutes and Ladders - its relative closeness to other things is contingent on future events and often not at all obvious. Certain concessions have to be made in the rules to that difference, or that difference will destroy the rules. It's unfortunate, but I think unavoidable. -Frank |
||
|
|
|
||
|
Feb 3 2007, 08:42 PM
Post
#45
|
|
|
Immortal Elf Group: Members Posts: 11,410 Joined: 1-October 03 From: Pittsburgh Member No.: 5,670 |
yeah. i like the rules more than i dislike their flaws. they're easier to handwave than, say, radio brains...
one possibility, just to add some tactical choice to the mechanic, is making it easier to target someone, the larger their Influence pool. deckers using Influence have a lot of connections, so it's easier to slip an attack program through their defenses. it might also make them easier to trace. you should be able to drop Influence any time you want, but raising it should take a while--an hour per point, say. alternatively/in addition, Influence could decrease your ability to sneak around. a million connections will make a lot of noise; maybe subtract Influence from Matrix stealth rolls (i forget the exact mechanic for staying out of sight in the Matrix)? |
|
|
|
|
Feb 3 2007, 09:49 PM
Post
#46
|
|||||
|
Runner Group: Members Posts: 3,314 Joined: 26-February 02 From: Lisbon, Cidade do Pecado Member No.: 185 |
It is today. Different tactics will definitely prevail in a world where hacking will get you a bullet in the brainpan rather than a slap on the wrist. |
||||
|
|
|
||||
|
Feb 3 2007, 09:55 PM
Post
#47
|
|
|
Immortal Elf Group: Members Posts: 11,410 Joined: 1-October 03 From: Pittsburgh Member No.: 5,670 |
eh. hack jobs that would get you a bullet in the brainpan in T6W will get you a decade or more in prison today. have you seen the sentences spammers are getting?
besides which, even if that were true, the likelihood of dying doesn't stop you from pulling out your million-agent army, any more than characters are prevented from doing anything else that might get them killed. there's no rule in SR that says you can't walk into a Lone Star office and start shooting the place up. you'll die if you do, sure, but that doesn't mean you can't do it. if you want hackers who use the million-agent trick to get caught and killed, there needs to be rules for it. |
|
|
|
|
Feb 3 2007, 10:04 PM
Post
#48
|
|
|
Hoppelhäschen 5000 Group: Members Posts: 5,807 Joined: 3-January 04 Member No.: 5,951 |
For anyone interested in the result of Zombie Wars - search for Bluefrog.
|
|
|
|
|
Feb 3 2007, 10:08 PM
Post
#49
|
|||
|
Great Dragon Group: Members Posts: 5,537 Joined: 27-August 06 From: Albuquerque NM Member No.: 9,234 |
While visits from the police are not as bad as visits from an Aztechnology hit team, the principles in avoiding the visit from the Azies are the same. You absolutely minimize the ties between the actual agents and your system, with multiple layers of cutouts. Essentially every part of your agent network runs on someone else's compromised hardware and you control it via a series of hops through other people's networks. Your personal system just has a link to the master control unit, which ideally would be something that would have a huge amount of normal traffic from across the world running through it. So you don't have a million connections, you have one, to say a system run by media company that provides you and 3 million other people the current headlines. This computer doesn't actually control the network, it controls command and control node(s) that do that, with subordinate C&C nodes as required when you network gets big, which it will (It takes a LOT of compromised computers to make money at 6 cents/day/PC, which I've been told is the going rate). This soft of setup works fine as long as you are using you network as some sort of blunt instrument and have written/stolen/purchased some decent agents and know how to properly task them. Essentially you have an army of kiddie scripters, or spambots, or a huge DOS network or whatever you want to make it do today. |
||
|
|
|
||
|
Feb 3 2007, 10:22 PM
Post
#50
|
|
|
Immortal Elf Group: Members Posts: 11,410 Joined: 1-October 03 From: Pittsburgh Member No.: 5,670 |
yeah, i realize the logic is kinda reversed from reality. however, decking in general basically reversed from reality. rather than distributing your activities as much as possible, like a hacker would in real life, an SR hacker's activities are combined into a single icon. you could try and handwave that as simple metaphor, but i think too much of the game mechanics rest on the icon being an actual online entity, rather than a metaphorical aggregate of a hacker's activities.
so, rather than tracing a million different connections, IC/security deckers only have to trace one connection--that of the icon. with Influence, the paths available for them to follow (and there's no reason they can't follow a bunch at once) increase exponentially. |
|
|
|
|
|
Lo-Fi Version | Time is now: 14th June 2026 - 12:22 AM |
Topps, Inc has sole ownership of the names, logo, artwork, marks, photographs, sounds, audio, video and/or any proprietary material used in connection with the game Shadowrun. Topps, Inc has granted permission to the Dumpshock Forums to use such names, logos, artwork, marks and/or any proprietary materials for promotional and informational purposes on its website but does not endorse, and is not affiliated with the Dumpshock Forums in any official capacity whatsoever.