Hmm, know something... I can't find any reference to crashing due to response degradation anywhere... if anything the system is going to be REALLY slow.

But the example in text in the RAW are very clear... for every System Programs run, response degrades by 1.
It gives the example of 10 programs run on a System5 reducing response by 2. (so 5-9 was -1, and 10 is -2)

My train of logic goes as follows... response degrades by 1 to 0. System is limited by response, so now System goes from 1 to 0. So you can effectively only run rating 0 programs. At this point, you're pretty limited, Lets say you're using it to browse, in which case the normal role is "Data Search + Browse", well browse is 0 so you're only rolling your data search skill dice.

But Unwired, does give you two ways out of this problem. Ergonomic programs which don't count against the limit. And Optomization allows the program to run at higher than system.

So I don't think this is a huge problem.

According to the Devs, this doesn't happen.

Rotbart... link or cite please. Back that up.

What's the point of degrading response if things limited by response don't degrade with it?

What happens when response hits 0? Can it even go negative?

There's an awful lot of programs which don't care about response... Just for arguments sake... we could load down a single response 6, system6, node w/ tons of rating 6 browse agents? So we drop the response of the node to 1 we're running so many. What's the effect, they all operate normally? That's the logical extension of what you've just asserted. (system limits program rating... but if response degradations don't degrade system capability, then programs aren't limited by response degradation).

See the FAQ.

You suck in Cybercombat.

Even less than that. Multiple agents running from one link suck in cybercombat. Without multiple agents running on too little hardware, no hacker should ever suffer from Response degradation. If need be use some of your subscriptions for programs running on another node. "One size fits all" calls for the ability to run some core programs on a rating 3 node anyway (Optimisation).

Thanks for the link... I missed that in the errata.

That was what I noticed though, response only matters for cybercombat/hacking/decryption really.

My thought was, for a largescale single commlink data mining operation. Pick out agents/software which doesn't care about response. Hide it behind a good firewall node. Then load it to the gills w/ say 16 rating 6 agents, running 16 rating 6 browses. Maybe try and run a setup like that while you're trying to track someone through publically accessable surveillance/traffic channels. You could get more agents working for you if you take program options of course. Who cares that you've dropped the response from 6 to 1... you've got your army all running at full rating.

Fitting in a discussion of clusters: May I ask what you want to achieve by not using a low, single-digit number of agents? As a search does take only minutes (searching the full matrix), the benefit of more than say 3-4 browse agents is very small.

Because I thought you'd require a nexus or a cluster to host an appreciable number of daemons on a single node.

Frankly I think it's a little broken that I can put 35 daemons on a single node all rating 6, all operating at 6, despite the fact they have the node loaded... but thems the rules. I'm not going to argue it. Even the optomization option seems a bit wrong... I probably would have made it hardware specific... this program can run better when run on this hardware, which would limit it's offensive potential (unless say you were intentionally targetting executives w/ their fairlight caliban's for example, which does seem pretty worthwhile). But that's going off the topic.

Anyhow, minutes may be too long. Lets say you want to actively check the publically broadcast SINs of everyone in the area looking for low rating fakes. You have 20 people per minute, sometimes more sometimes less. That means at 1 minute per, you need at least 20 agents doing the data searches. Or lets say you're trying to track a vehicle/pedestrian using low-sec or publically available traffic/cafe security cams. You want to have a small army of agents spread out and look at them all and run the program to do the facial/vehicle recognition routine.

The point is these are all tasks which don't care one wit about response. Response has gone from a vital statistic in my book to one which rarely matters except for combat and hacking . Sounds kinda silly, it doesn't matter that the program is sluggish and unresponsive it still does it's job just as well as if it was the only thing in memory.

But in any case, this has been one of the most informative threads I've seen in a while, thanks to all for the informative feedback, interchange, and corrections.

and i guess a quick explanation is that most stuff done works not at speed of machine, but speed of comprehension.

the faster the user can comprehend whats going on, the faster it gets done.

as in, when doing a search, the search system may find things at a insane rate, but the human mind still needs to see the patterns and recalibrate the parameters to sort out the noise.

while a agent in SR4 has more in common with a SK from SR3, its still not a true AI in that it can do leaps of logic (guesswork) that its not coded to look for.

but when it comes to cracking and combat, you want automated systems that does its thing as fast as possible, to outpace the opponents similarly automated stuff.

still, it could be a indication that the matrix metaphor used in SR4 is not "set in stone" and therefor become subject to interpretation and reinterpretation as its forced to take into account new ways people come up with for using it.

hell, its been rewritten 3 times in 4 editions or there about. in the same space the magic system have had maybe 2. and even then most of the magical foundations have stayed the same, only specific cases have been moved around or removed/inserted...

we have gone from "dungeon crawl" to "metaplanes" and now "AR". and in those times the numbers and titles being used have been shuffled, removed, added, fully replaced and whats not.

the SR2 deck had some 7 stats, the SR4 comlink have 3.

the SR2 host had a infinite number of "rooms", the SR3 host had 5-6 stats, the SR4 node share its stats with the comlink (the comlink is a node).

SR2-3 had a long list of special IC. SR4 have them as agents with different loadouts. the same agents that a hacker makes use of for offense.