Bandwidthoracle
Oct 10 2005, 02:17 PM
I've been reading the extended test times, and I have come up with a strange problem:
Creating an OS, no matter the rating, is measured in years, since successes no longer reduce time, doesn't this mean that even the most gifted team in the world working on a rating 1 OS, is going to take a whole year? If I'm reading this wrong I hope someone will point it out.
blakkie
Oct 10 2005, 03:09 PM
That is the downside to specifying a larger Interval instead of a higher Threshhold for an Extended Test, yes.
EDIT: BTW they could 1/2 the time by making a Rush Job of it.
calypso
Oct 10 2005, 03:12 PM
In theory, you could double the threshold and halve the interval, without affecting the difficulty couldn't you? This would give you finer granularity and allow for the Super Team to make an OS in a reasonable time?
Calypso
Bandwidthoracle
Oct 10 2005, 03:43 PM
In SR3 on of the PC's in the group had a "Super coder" character, essentially this character's best quality was the ability to produce code fast. In SR4 the PC's kinda refuse to code, when they realise it could be a six year endevor to produce a useable program.
Actually, come to think of it, don't most of the extended tests have no ability to shrink?
Eagle
Oct 10 2005, 04:08 PM
Yeah there seems to be something wrong with those times. It would take 2 or more months to put together a rtg 6 encyrption program. In real life, the majority of work putting it together is the encryption formula, which is public domain. Putting it into software is relatively simple, certainly not 2 months work.
Decryption would take much longer because of trying to circumvent in several different ways, but these holes are consistently patched but that could be simulated by SoTA, ie software decays at 1 point/month. So requires less time to write but more time upkeep.
I would think that software should be easier make but require much more maintenance. This is especially true when programming consists of mainly of re-using current code but upgrading it to new requirements.
Eyeless Blond
Oct 10 2005, 04:29 PM
Also with more people you're very likely of netting a Critical Success, which you can rule divides into the base time or somesuch. Of course that requires truly massive dice pools being thrown at the problem to reliably program something quickly.
This is actually one of the bigger objections Ellery had to SR4 back when the static TN was first announced, back in May I believe?
Chandon
Oct 10 2005, 04:40 PM
In real life, writing non-trivial software takes a team of people a reasonable amount of time. The concept of a single person sitting down and writing a useful and complete software program in a couple weeks is mostly a myth.
Azralon
Oct 10 2005, 04:43 PM
QUOTE (Chandon @ Oct 10 2005, 12:40 PM) |
In real life, writing non-trivial software takes a team of people a reasonable amount of time. The concept of a single person sitting down and writing a useful and complete software program in a couple weeks is mostly a myth. |
Sure, but you can gloss over that easily by declaring the tools & APIs available in 2070 are superspiffy.
Kagetenshi
Oct 10 2005, 05:29 PM
QUOTE (Chandon) |
In real life, writing non-trivial software takes a team of people a reasonable amount of time. The concept of a single person sitting down and writing a useful and complete software program in a couple weeks is mostly a myth. |
Define "useful and complete"?
~J
hobgoblin
Oct 10 2005, 05:32 PM
question is, what is trivial in SR?
sure one can trow together a nice simple editor in something equivalent to visual basic in a matter of hours. but isnt mostly tools like that incorporated into the os these days? or maybe leased of the net as needed?
how long did it take linus torvalds to write the first release of the linux kernel?
how long have it taken microsoft to write the next windows version?
how many people are working on the cups printing system used by most linux distros?
what can be defined as a trivial software to write?
QUOTE (eagle) |
In real life, the majority of work putting it together is the encryption formula, which is public domain. Putting it into software is relatively simple, certainly not 2 months work. |
while encryption is mostly public domain in real life (baring patents and military stuff), it dont have to be so in SR. it may well be that encryption software from the megacorps, atleast the stuff available on the open market, are black box compiled code.
sure you may say that ensuring that encryption works need peer review. but do they have to care about that? maybe some high profile encryption labs can see how the stuff works under a NDA so that they can put up their support of the product or not.
rember that SR is ruled by corps. if its bad or potentialy bad for the bottom line then its not done. and releasing encryption algorithms so that others can use them is not good for the bottom line.
this is why shadowrunners have a job market.
lets say we kill the patent system and instead extend copyright to cover anything, even physical products. so if megacorp A release a lease a product and then megacorp B release one that is suspisiously similar then its shadowrunners to find evidence and the corp court to make a ruling.
mfb
Oct 10 2005, 05:36 PM
QUOTE (Chandon) |
In real life, writing non-trivial software takes a team of people a reasonable amount of time. The concept of a single person sitting down and writing a useful and complete software program in a couple weeks is mostly a myth. |
that's true, if you're talking about developing new software, eg something no one's ever done before. if you're just making a new version of something others have already done, a few weeks is a suitable amount of time.
RunnerPaul
Oct 10 2005, 05:54 PM
QUOTE (Azralon) |
Sure, but you can gloss over that easily by declaring the tools & APIs available in 2070 are superspiffy. |
And that glossing can be matted back down by pointing out that OS design in 2070 would by its very nature be insanely complex, and therefore, the advantage of having a superspiffy development suite means you get to write your OS in years instead of centuries.
hobgoblin
Oct 10 2005, 05:58 PM
QUOTE (mfb) |
QUOTE (Chandon @ Oct 10 2005, 11:40 AM) | In real life, writing non-trivial software takes a team of people a reasonable amount of time. The concept of a single person sitting down and writing a useful and complete software program in a couple weeks is mostly a myth. |
that's true, if you're talking about developing new software, eg something no one's ever done before. if you're just making a new version of something others have already done, a few weeks is a suitable amount of time.
|
i dont think i could reproduce ms office in a matter of weeks even tho its been done before
![nyahnyah.gif](http://forums.dumpshock.com/html/emoticons/nyahnyah.gif)
lets just say that for every feature needed the complexity of the code squares...
Chandon
Oct 10 2005, 06:07 PM
Another important point is that the rules in the book describe work by a single programmer, not a team. The origional poster referred to operating systems, and I can't come up with a single example of a single programmer writing a sufficiently non-trivial operating system to get a rating of 1 on a 1-6 scale since Minix - and that came on 5.25" floppy disks.
calypso
Oct 10 2005, 07:00 PM
QUOTE (Chandon @ Oct 10 2005, 01:07 PM) |
Another important point is that the rules in the book describe work by a single programmer, not a team. The origional poster referred to operating systems, and I can't come up with a single example of a single programmer writing a sufficiently non-trivial operating system to get a rating of 1 on a 1-6 scale since Minix - and that came on 5.25" floppy disks. |
So you throw a hundred programmers at the problem and it's completed in 1/100th the time, right???
![rotfl.gif](http://forums.dumpshock.com/html/emoticons/rotfl.gif)
Calypso
EDIT: *secret handshake* "Mythical Man Month"
Azralon
Oct 10 2005, 07:01 PM
QUOTE (RunnerPaul @ Oct 10 2005, 01:54 PM) |
QUOTE (Azralon @ Oct 10 2005, 11:43 AM) | Sure, but you can gloss over that easily by declaring the tools & APIs available in 2070 are superspiffy. |
And that glossing can be matted back down by pointing out that OS design in 2070 would by its very nature be insanely complex, and therefore, the advantage of having a superspiffy development suite means you get to write your OS in years instead of centuries.
|
I hear ya. If the RAW say years, then I'd say that's clearly the case.
They could have said building new programs is approximately as difficult as narrating what you want to a Star Trek holodeck (now there's a toolset for you). But they didn't, so here we are.
Vector
Oct 10 2005, 07:03 PM
QUOTE (calypso) |
So you throw a hundred programmers at the problem and it's completed in 1/100th the time, right??? ![rotfl.gif](http://forums.dumpshock.com/html/emoticons/rotfl.gif)
Calypso
EDIT: *secret handshake* "Mythical Man Month" |
No, but at the same point you should be able to break it up into smaller pieces that can be worked on in parallel. I would bet that most software projects of reasonable size can be broken into parallel steps.
Azralon
Oct 10 2005, 07:05 PM
QUOTE (calypso) |
So you throw a hundred programmers at the problem and it's completed in 1/100th the time, right??? |
I say this is a programmer myself:
HA!
hobgoblin
Oct 10 2005, 07:07 PM
QUOTE |
No, but at the same point you should be able to break it up into smaller pieces that can be worked on in parallel. I would bet that most software projects of reasonable size can be broken into parallel steps. |
bingo, there is a reason why you have both a compiler and a linker
hobgoblin
Oct 10 2005, 07:11 PM
QUOTE (Azralon) |
QUOTE (calypso @ Oct 10 2005, 03:00 PM) | So you throw a hundred programmers at the problem and it's completed in 1/100th the time, right??? |
I say this is a programmer myself:
HA!
|
the proverb "more chefs, bigger mess" comes to mind
calypso
Oct 10 2005, 07:17 PM
QUOTE (Vector) |
No, but at the same point you should be able to break it up into smaller pieces that can be worked on in parallel. I would bet that most software projects of reasonable size can be broken into parallel steps. |
Sort of. As you parallelize a task, the coordination required to do so begins to eat up the gains from the parallelization. I forget how the graph looks exactly, but at some point, adding people causes the task to take longer.
Calypso
mfb
Oct 10 2005, 07:19 PM
QUOTE (hobgoblin) |
i dont think i could reproduce ms office in a matter of weeks even tho its been done before |
depends. are you going to try to reproduce the bugs hidden features, too?
Azralon
Oct 10 2005, 07:32 PM
Brook's Law, for those who wish to Google for more info. The graph looks something like plotting an inverse exponent.
As far as parallel tasking goes, it's insightful management when you can compartmentalize subroutines cleanly into different simultaneous projects. It's a friggin' miracle when the Systems Designer(s) can hook all of those different subroutines together successfully since they evolved under (sometimes radically) different conditions.
Yuck. I've started to sound like one of those guys who insist they're martial arts experts and the rules should work differently because blah blah I'm a lethal weapon blah.
Bandwidthoracle
Oct 10 2005, 08:30 PM
I figured most hacker apps wouldn't be profesonally-polished, gumdrop-interfaced, idiot-proof code. I mean, does my IC really need a help file? I think a lot of things like that could be made in a week, given you where not doing proof of concept.
I wonder if you could be alloud to allocate hits to time instead of success, or divide time by the number of hits over success, of course that really only works on your last roll (so that second one is out)
I want my shadow-coder PC back
hobgoblin
Oct 10 2005, 09:28 PM
writing the manual may take time. but writing the exploit code may allso take time. if your only writing for known exploits then you risk it being patched on 90%+ targets before your done (if they bother at all) so its a bit of a race.
write for unknown exploits and you have to find it first, the hard way
![nyahnyah.gif](http://forums.dumpshock.com/html/emoticons/nyahnyah.gif)
thing is, why bother running the shadows when one can sitt on ones ass and code all day for those that do? and why should they do the running when they can do the same?
in the end, file the times under game balance.
Chandon
Oct 10 2005, 11:56 PM
QUOTE (mfb) |
that's true, if you're talking about developing new software, eg something no one's ever done before. if you're just making a new version of something others have already done, a few weeks is a suitable amount of time. |
Which is exactly why Mozilla (aka Netscape 6) took only a couple of weeks to reach release quality rather than five years. I mean, all a web browser does is display a well specified and reasonably simple class of formatted text documents.
QUOTE (calypso) |
So you throw a hundred programmers at the problem and it's completed in 1/100th the time, right??? |
No, but if a project took a team of 10 skilled programmers 2 years, that's a strong indication that the project can't be duplicated by one programmer in two weeks.
hobgoblin
Oct 11 2005, 12:08 AM
chardon, did you forget the sarcasm tags?
the mozilla code was based on the 4.5 or there about netscape code. alltho most of it was trown out as time went on. hell, they redid the whole gui code so that its a mix of xml based stuff, with some javascript and some sort of binding code. basicly the whole window content is renderd by the gecko engine.
and i have no clue as to the number of people that worked on it at any one time...
and how long did it take the kde people to write the khtml rendering engine that power konqueror? kde is a bit funny that way as stuff is made up of kparts. so you take your ftp kpart and put it inside a file system gui and presto, you have a ftp client. no wonder even a save dialog can understand ftp:// links when using that desktop
![wink.gif](http://forums.dumpshock.com/html/emoticons/wink.gif)
(and dont forget fish://, basicly ssh based file transfer).
mfb
Oct 11 2005, 01:36 AM
QUOTE (Chandon) |
Which is exactly why Mozilla (aka Netscape 6) took only a couple of weeks to reach release quality rather than five years. I mean, all a web browser does is display a well specified and reasonably simple class of formatted text documents. |
re-read. i'm not talking about developing new features, i'm talking about slamming out a program for your own use that is already on the market. it's not like every decker who wants a new sleaze program has to start from scratch; a quick search of the Matrix will net him dev notes and all kinds of other data from other people who've created what is basically the same program.
hobgoblin
Oct 11 2005, 03:46 AM
unless its been ripped out of there by corp techs on a mission...
Chandon
Oct 11 2005, 04:04 AM
mfb: If that's true, then you might as well just grab an open source sleeze program off Shadowland. On the other hand, the book implies that people don't tend to share software or software concepts, so if you accept SR4 at face value neither of those are really an option.
Actually, the way I'd probably run that sort of thing right now is that before you can get in on the good shared hacking programs you need to demonstrate that you're A.) good enough to contribute and B.) not going to tell the corps about the exploits you're using. I'd say that's at least as hard as finding a good initiatory group for a mage.
blakkie
Oct 11 2005, 04:16 AM
As far as i can see RAW has no way for you even share code with
yourself between projects.
![frown.gif](http://forums.dumpshock.com/html/emoticons/frown.gif)
Building a bigger, badder version of a given program? Back to square 1.
mfb
Oct 11 2005, 04:17 AM
QUOTE (Chandon) |
mfb: If that's true, then you might as well just grab an open source sleeze program off Shadowland. |
you might, at that. and then use your programming skill to improve it to your own specifications.
hobgoblin
Oct 11 2005, 04:27 AM
question realy is, how simple is it to find said info.
i would expect info on everything from browsers to ascii editors to be very available.
but most of the software that a hacker uses are software that have security defeating code buildt into it.
i would guess that this is illegal to share in SR. or atleast attempted controled, with anyone found to be spreading it affected by laws that make the dmca look like a bad joke.
sure, you may find some discared code or notes on some shadowland node. but chances are that the stuff have allready been scanned by a corp and patched up in their latest products.
secrecy is key, both for the corps, and for the people breaking into their systems. alone it dont help much, but used as an extra layer in a layerd secuirty plan, it helps buy response time.
rember that the matrix of sr where buildt to corp specs. and with corps in tight control. i have a feel that the open way of talking about security hole that we see today is not accepted in sr.
Vector
Oct 11 2005, 12:56 PM
QUOTE (hobgoblin) |
in the end, file the times under game balance. |
They are decent game balance. But they also erradicate (or nearly so) the concept of the hacker who writes all his own software. Page 240 "Commlink Customization" brings up the same idea that has been written into the fluff of the rules in each of the core books... the hacker who doesn't write his own code is an amatuer. An idea that is much easier to handle in concept than execution.
Software does take time to write, even if you are simply upgrading code for which you have the source. That said, the rules just don't let the hacker upgrade his programs or write new ones in a time frame that allows it to be done during a campaign on a regular basis. Not sure what a solution to that would be though.
Eyeless Blond
Oct 11 2005, 02:01 PM
QUOTE (hobgoblin @ Oct 10 2005, 08:27 PM) |
sure, you may find some discared code or notes on some shadowland node. but chances are that the stuff have allready been scanned by a corp and patched up in their latest products. |
Hahahahaha! You're not serious, are you? Just to take a modern example, Microsoft has several hundred documented security flaws that they as yet haven't even bothered trying to plug A) because they're too busy putting band-aids over their swiss-cheese software and B) because they'd rather make a new, even more bug-filled version of windows and force everyone to pay them for it instead of giving everyone free patches. The only time you see a patch with MS is if the cript kiddies get ahold of a flaw and spread a virus to tens of thousands of computers; even then it'll take awhile to see an update.
The fact that you see so many updates is a testament to how buggy the software is and how many people are writing viruses today. I suspect you'll se a little less than that when people get more heavy-handed about swatting such people. Right now they get off with a few months in jail or a fine, but after Crash 1.0 they'd likely get extradited to corp property and shot.
Eyeless Blond
Oct 11 2005, 02:05 PM
QUOTE (Vector) |
They are decent game balance. But they also erradicate (or nearly so) the concept of the hacker who writes all his own software. Page 240 "Commlink Customization" brings up the same idea that has been written into the fluff of the rules in each of the core books... the hacker who doesn't write his own code is an amatuer. An idea that is much easier to handle in concept than execution. |
On the other hand, it's not generally possible to write code that's any better than what's available at chargen anyway. It costs less than 100,000Y to buy every program at rating 6, along with nearly the best Commlink. It hardly matters how difficult it is to program; there's nothing left *to* program!
Jaid
Oct 11 2005, 02:18 PM
[edit] ahhhhh nuts... obviously spent way too long on writing this out =P [/edit]
so start off with rating 6 for all programs, and insert the writing of said programs into your backstory. resources do not necessarily represent
![nuyen.gif](http://forums.dumpshock.com/html/emoticons/nuyen.gif)
spent, you know, it can also represent time put into things using skills, for example.
under core rules, i don't recall being able to buy anything higher than 6, so you should be fine. the only thing you would start with at less than 6 is agents, and considering you're basically programming an almost-AI, as far as i can tell (are these basically the same as semi-autonomous knowbots, by any chance?), i should ruddy well hope that takes you a while.
as far as coding anything higher than 6, assuming it is allowed in your campaign... well, it's supposed to be rare. if you could code it in two weeks with hardly any effort behind it, that wouldn't sound terribly "rare" to me.
Vector
Oct 11 2005, 03:10 PM
I think the OS and Agents would certainly be things you would want to improve after char gen. But the point about buying all the progs at rating 6 is helpful, but doesn't help in game. Has anyone done the math to find out how long it would take to code every program at rating 6?
hobgoblin
Oct 11 2005, 03:13 PM
dont know how SR4 agents are represented, but the SK of SR3 was something like a robotic brain, only able to roam the net. a agent was supposedly a step down from a SK, and it would take teams of people months or years to write a good SK.
Vector
Oct 11 2005, 03:21 PM
I agree Agents are watered down SKs at best. AIs are unbelievably powerful SKs with an X-factor thrown in.
Eagle
Oct 11 2005, 03:24 PM
There's something fundamentally wrong with a system where the default purchase is all software at level 6, the maximum possible. There seems to be no tradeoffs and the cost is low, so even with 300BP characters the hackers has rating 6 software. Something very wrong there.
blakkie
Oct 11 2005, 03:35 PM
That's actually pretty expensive to do for the Hacking programs (6K/pop) so 90K there, and another 6000
![nuyen.gif](http://forums.dumpshock.com/html/emoticons/nuyen.gif)
for Firewall and System. Virtually nothing for the normal programs, 4200
![nuyen.gif](http://forums.dumpshock.com/html/emoticons/nuyen.gif)
to cover all of them. The best Agent you can get is 10
![nuyen.gif](http://forums.dumpshock.com/html/emoticons/nuyen.gif)
, same for IC and Pilot. So around 130K gets you set up best you can. Another 5
![nuyen.gif](http://forums.dumpshock.com/html/emoticons/nuyen.gif)
for the best starting Commlink (done via upgrades). So basically within reach for the serious hacker, and the only thing he'll ever need to improve is the last point on his commlink for another 12
![nuyen.gif](http://forums.dumpshock.com/html/emoticons/nuyen.gif)
and those pesky Agents/IC/Pilots.
Yes, there is something wacked there.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please
click here.