From: | Tzeentch tzeentch666@*********.net |
---|---|
Subject: | The Matrix: Feedback on existing material |
Date: | Sat, 4 Dec 1999 15:20:01 -0800 |
http://www.amurgsval.org/shadowrun/megapulses.html. My comments are located
in the >>>>>[ ]<<<<< boxes. The original author is
apparantely
Talks-With-Cats. No offense or infringement is intended.
----------------------------------------------------------
Computing in Shadowrun
The nature of computing in Shadowrun is, at best, inobvious to anyone
familiar with modern computers, and at worst nonsensical. Why is a computer'
s power measured in megapulses (Mp), with memory and computing power
apparently being identical? What is a megapulse anyway? Why are so many of
them involved in Shadowland posts? And why is it so hard to make a secure
computer system? Here's my take on the matter.
>>>>>[Hmm? I always assumed that your MPCP was your decks "power"
and Mp
were memory units.]<<<<<
What is a "pulse"?
The basic technology of Shadowrun computing is the quantum pulse element,
usually just called a "pulse". Its basic nature is that it's a cluster of
quantum dots and wires (or maybe some funky proteins embedded in a
crystalline matrix, if you want computing to be completely optical in
nature) that can be reconfigured into a different set of logic elements on
the fly. If you want it to turn into a group of flip-flops, it can become
memory. If you want it to be an Arithmetic and Logic Unit instead, it can
switch to that in the middle of your program.
>>>>>[So some sort of "rod" logic as detailed in The Diamond Age?
"Pulse" is
not very descriptive however. "Configurable" processors are nothing new,
there has been research into using field programmable arrays for just such
an use. Viruses can be a LOT nastier in this enviroment if the very system
can be configured in this manner. And as you note later on the optical chips
are not burnable/configurable without special hardware.]<<<<<
Quantum pulse elements are so much faster and more convenient than other
forms of computing technology that the old silicon stuff has become quite
obsolete. With the universality of quantum pulse elements, you have access
to processing that's as parallel as you want it. All programs are compatible
with other computers, as long as the computer is big enough to run it.
Programs from older computers can easily be run on an emulator.
>>>>>[I'm leery of assigning even future computer technology too many
elements of "quantum" computers. And while reconfigurable systems may be
more efficient and faster you will have a hard time beating silicon or
gallium arsenide, or even plastic circuitry for cost. Don't expect the
future Whizz-O'-Matic blender to have a quantum "pulse" based chipset. As
for universal compatibility, that I don't buy at all. You'll still have
architectual differences, variant OS's and even incompatibility caused by
other installed software.]<<<<<
Naturally, programming and debugging with such elements is incredibly
difficult without computer assistance. People seldom write compilers from
scratch any more; the equivalent of assembly language for quantum pulse
elements makes programming for massively parallel RISC machines look simple.
>>>>>[So not only cost but development overhead comes into play. If I
wanted
to make an embedded system using this "pulse" technology I could very well
have to develop the necessary infrastructure from scratch. And if it really
is that hard and complex I'll just use an inferior, but vastly cheaper
technology.]<<<<<
When someone refers to a megapulse as a measure of size, that's the number
of bits that can be stored by 1,048,576 (220) quantum pulse elements
configured for uncompressed memory storage only. (A megapulse of computing
logic can usually store more than a megapulse of raw data, since it can be
configured as mixed memory and compression logic.)
>>>>>[Hmm. Ok.]<<<<<
So why do Shadowland posts take up so many megapulses?
The computing environment of Shadowrun is a descendent of the applet-filled
Java world currently growing on the Internet right now. Everyone has their
favorite word processor or other editing program; some people post by
speaking to the computer, some by sending in the bits over a direct neural
interface, some use keyboards. When you ship your message up, it could be as
small as the Unicode or even ASCII representation of your post, or it could
be a multimedia extravaganza with an animation of your Matrix icon speaking
your post out loud in your voice (or a synthesized one) and all the software
needed to run such a thing independently (instead of making someone download
a viewer). Naturally, when it gets converted to text for a paper sourcebook,
you wind up with the usual posts we're familiar with. This is, by modern
standards, horribly wasteful, but memory and storage are cheap, and the same
phenomenon that makes Microsoft Word take up 30MB on your hard drive today
hasn't gone away.
>>>>>[Soo...bad programming is the cause of the inflated Mp sizes shown in
the Shadowland posts..as opposed to say..the author not paying attention to
what he was writing when making the book.]<<<<<
The difference between data and executable in Shadowrun is almost
nonexistent. Sensibly paranoid deckers keep a suite of trusted programs for
reading the myriad different data file formats, rather than risking the
possibility that the file handed to them could be hostile. Most cyberdecks
have a diagnostic mode that allows the decker to freeze the deck into
read-only mode and run comparison utilities on their personal computer,
making sure that a virus hasn't snuck into the software. (The notion that a
virus can sneak into firmware- seen in the "Escher loop" in Lone Star and
the "Worms" IC in VR 2.0- lacks believability. If you could do that in
software, you wouldn't need optical chip encoders.)
>>>>>[Well isn't that the very thing your proposal of configurable logic
allows?]<<<<<
Mainframes hooked up to the Matrix tend to be pumping around large amounts
of executable content as various utilities running on the mainframe talk to
their creators' sites to pick up the latest updates and access data that's
not worth keeping at the local site. Decking, in my view, is the art of
corrupting the data streams going into a mainframe in order to gain access
to the system.
>>>>>[I keep hearing thin clients are dead. Guess Sun and Oracle won the
computer wars in the Shadowrun universe. Seems generally more efficient to
use fat clients maintaining their own executables, and grabbing updates as
they become available. What do you mean corrupting the data streams exactly?
Sneaking in your data with the other data that is entering their system?
Sounds like spoofing.]<<<<<
If that rationalization isn't strong enough for you, here's an extension of
the notion:
Bellcore did a study back in the late 1980's that showed that with a
sufficiently high-speed network, it could be more efficient to have data in
constant circulation, rather than being provided on request by a server: if
the data goes by faster than the request-response cycle, and you have
bandwidth to spare, you might as well just have the data being pumped by.
Computers on the network could then simply grab whatever chunk of the
operating system or the Encyclopedia Britannica they need at the moment.
>>>>>[I'd be interested in the premise of this study and what exactly they
were applying this towards. And as anyone who uses corporate networks know
you can never seem to have enough bandwidth - every time you get more it
gets saturated shortly thereafter.]<<<<<
The Matrix could be a phenomenon of this nature: the routers handling Matrix
traffic might be keeping data "pumped" throughout the network, and the
biggest cost of taking your mainframe off the Matrix is that you'd have to
buy gigapulses and gigapulses of memory to store all the software your
mainframe might need, rather than just picking it up off the Matrix. Entire
data havens could hide inside that incredible flow of traffic, being quietly
rebroadcast along with all the other information.
>>>>>[Oracle must be having a field-day in 2060! Essentially you are saying
everyone is hooked up to the Matrix backbone and getting everyone elses
data. If you wanted to watch a simsense show you just snatch it out of the
datastream and watch it. I don't buy this both from the data integrity issue
(cracking software just got a lot easier!!), piracy and spoofing
software.]<<<<<
Why Does My Encephalon Need a Skillsoft?
Requiring an encephalon to be slotting a decking skillsoft in order to get a
bonus during decking doesn't make a great deal of sense to me. To keep the
same game balance while improving believability, I suggest that the
skillsoft be replaced with a program of the same rating and space
requirements as a specialized decking skillsoft, with the purpose of
properly utilizing the encephalon's capabilities during decking. In
particular, this program would interface to the deck and make it possible to
provide extra information at the level of skills and memory: instead of
having to read statistics about whatever the decker was perceiving, they
would simply know them.
Making Sense of Decking
"The Crystal Wind is the Storm, and the Storm is Data, and the Data is
Life."
-the Player's Litany
Decking, as represented in Virtual Realities 2.0, doesn't always make a
great deal of sense when you contemplate how computers actually work. This
problem is endemic to most cyberpunk, in fact: William Gibson's Neuromancer,
which is perhaps the classic decker book, was written by a man who wasn't
even computer-literate at the time.
>>>>>[You have to admit VR2 does a pretty good job
though.]<<<<<
I take my inspiration from an out-of-print work of cyberpunkish fiction
titled The Long Run, by Daniel Keys Moran. His work contains the only
believable depiction of decking I've seen in fiction. (Daniel Keys Moran's
day job is, of course, being a computer programmer.)
>>>>>[Which may or may not mean much.]<<<<<
There are some different terms in Daniel Keys Moran's universe of the
Continuing Time. Garden-variety deckers are called webdancers, the Matrix is
called the Crystal Wind, and the top-notch folk like FastJack are called
Players.
Images in the Crystal Wind
Despite numerous depictions of deckers responding to computers in
milliseconds, this is not actually the case. There is simply no way that a
human being, whose thoughts take place at the speed of ions diffusing across
semipermeable membranes in a complex network of of wet tissue, can keep up
with the speed of the computer world. A decker needs a program to intercede
for him between the computer world and that of human reaction times. This is
the job of the Master Persona Control Program. The MPCP is a program the
represents the decker in the Matrix; it carries as much of the decker's
decision-making capability as they can get down into code. On its own, an
MPCP following verbal orders should be capable of a great deal of work
without even a decker directly attached. With the decker present, the MPCP
becomes an extension of the decker's will, a customized interface to a world
where the speed of light is an important limit on reaction time.
>>>>>[No computer system would run at anywhere near lightspeed. Even a
fully
optical computer would have switches and logic gates, as well as chips that
run at considerably lower speeds then light. Then you have distances
involved between components, response delay of programs and the speed that
the OS can switch between tasks.]<<<<<
The MPCP, under ordinary circumstances, never gets used as anything other
than a way of integrating the other persona programs (which are basically
libraries).
>>>>>[I've always looked at the MPCP was more a piece of firmware that
coordinated your Persona programs. You don't have a CPU as such since a
cyberdeck is assumed to use several specialized processors. Much like a high
end console system of today.]<<<<<
Datastarve
Decking is supposed to be addictive. However, there is nothing in the source
material to suggest why anyone would stay in the Matrix when they weren't
breaking into systems- and there's no way anyone would spend all their time
doing that, as they'd be bound to screw up eventually.
>>>>>[Information is addicting. Anyone who spends a large amount of time
can
agree with that.]<<<<<
So why do deckers spend all their time out there on the Matrix? Part of
their time will be spent in working on code for various projects, of course.
Another part will be satisfying their curiosity. The World Wide Web of 1997
is already addictive: it is very easy to waste hours chasing down
interesting information with search engines and sites filled with links, and
taking part in USENET newsgroups.
>>>>>[Exactly. Especially in the future Matrix where you can be and
experience literally whatever you can imagine!]<<<<<
Deckers should be information addicts to some degree or other. They may be
in the habit of wandering around with an unobstrusive fiber optic cable
snaking into their datajack, feeding information into their encephalon and
display link, or be nearly impossible to tear away from the data feed. They
should feel naked if they can't glance up at the sky, wonder about the
weather, and download real-time pictures from a satellite and check them
against the forecasts from the meteorology services, or are out of range of
being paged when one of their automated filters finds something of great
importance. (For instance, a decker once surprised by an earthquake might
set up a filter to monitor global seismology stations and page him
immediately if the magnitude and distance are right for him to notice it;
such a filter might get there in advance of the first crustal waves.)
>>>>>[Exactly. I'd do this right now if I could.]<<<<<
Icons
The whole notion that there needs to be a correspondence between iconography
in the Matrix and the purpose and power of the programs the icons represent
is nonsense. Computers don't care about icons. Human beings may demand that
software conform to certain standards in order for them to have a better
chance of understanding it- but why is a decker going to make life easy for
them? A powerful attack program may manifest as a huge flaming sword to
comply with iconography, but it could just as easily be a petite derringer,
a poisoned dart shuriken, or completely invisible. A deck's Sensor module
would pass the important information on to the decker, but a sarariman doing
his job would probably have no idea what was going on.
>>>>>[Icons were always a disposable concept, I always assumed your Sensor
suite was the one rendering programs as wolves or doors or whatever. Fitting
it into the metaphor of the Matrix to quite another poster.]<<<<<
What are those utilities doing?
It may help the computer-literate to consider what the parts of a cyberdeck
are doing at a programmatic level.
The Bod program is a hybrid descendent of Purify and virus-checking
software. Its job is to perform huge amounts of error checking, enforcing
memory protection, perfoming constant checksumming on code while it runs,
and more: all the work of preventing a decker's running programs from being
corrupted.
>>>>>[Which one wonders why you can't separate the memory space these
programs are running in totally. Or better yet running them in firmware.
Unless they require input and output there is little reason to make them
immune to normal corruption methods.]<<<<<
A deck's Evasion code is concerned with making it difficult to find the
decker's programs in order to do anything to them. Its techniques include
spawning extra processes with similar characteristics to the important ones
the decker is running, shifting executing code to different spots in the
machine's address space, and leaving false trails in all the locations that
monitor system resources.
>>>>>[I doubt it's one "uber" program. More likely it's a suite
of software.
Some commercial, some custom, some old. None of the Persona "programs" are
just one program - but a constellation of programs and untilities. An
example of a program in your Evasion suite would be the software that munges
your datatrail]<<<<<
The Masking code attempts to cloak the decker from discovery. In addition to
shifting the appearance of the user's icon to an appearance appropriate to
the local host, it also camouflages the characteristics of the decker's
utilities to make them appear to be legitimate programs running on the
system. This includes activities such as talking to license servers,
accessing the appropriate support files for a given utility, and inserting
the user's code into a running utility program to steal its resources.
>>>>>[Your icon makes no difference except to yourself I would imagine. If
you buy our original discussion on this you would be seen as a "default" to
other people unless you broadcast yourself. Most people would not even see
you. Why would you want to offload your decks programs onto the server to
begin with? If you wanted to get it to run some malicious code I could see.
I would not mind seeing Masking and Evasion be combined into one stat in
fact.]<<<<<
Sensors make sense out of the chaos of the computer world. Any good decker
knows that icons are nothing more than an illusion wrapped around the true
reality that happens in the computer; frequently, these illusions are
dangerously deceptive. The deck's Sensors provide annotations to other icons
in the area, and supply vital statistics about the system as the decker
operates it. A very large part of the Sensors suite is the heuristics for
knowing which information to feed the decker at the right time. This is the
major place where an encephalon helps out with decking, as the sensors can
simply feed the information directly to the decker's short-term memory
without requiring additional time for conscious processing.
>>>>>[Uhm. Ok. I've always assumed your sensors do the interpreting
anyones.
No annotation of existing icons would be needed. Otherwise the Sensor suite
is an advanced version of existing network diagnostic tools.]<<<<<
So what's that IC doing, then?
FASA's treatment of IC is occasionally lacking in an understanding of
computer technology. Here's what I've done to make sense of it:
Killer IC is a straightforward, half-tamed descendent of Core Wars, designed
for the modern computing environment. It attempts to scramble code, free
allocated memory, kill processes and threads, deny permissions, and
similarly cause a decker's programs to become unstable and crash. A Data
Bomb is just specially triggered Killer IC, and Tar Baby IC is just Killer
IC targeted against programs that attempt to do a particular thing.
Crippler IC is designed to attack the sorts of code that are used in the
appropriate deck attributes. Acid IC targets Bod by setting its own hostile
callbacks on requests for memory protection and attempting to separate the
Bod code from the programs it protects. Binder attacks Evasion by invading
the decker's address space and leaving haphazard data trails that make it
obvious where the decker is. Marker breaks Masking by blocking access to
license servers and protecting other utilities running on the system from
invasion. Jammer goes after Sensors by intercepting requests for system
information and returning hostile code or nonsensical data.
The nature of quantum pulse elements is such that it is possible to burn
them out through executing an appropriate series of instructions. (One of
the major parts of compiler design is searching for possible errors that
could lead to this sort of problem!) This makes it possible for IC to
actually damage a decker's burned-in utilities. Blaster does exactly this,
and the different Rippers are just like the Cripplers above, but sending
code on down to the deck to attack the firmware of the routines they're
attacking.
>>>>>[Blah, I don't buy this at all.]<<<<<
Scramble is a complete misnomer for that kind of IC, and Decrypt is
definitely not the right name for the program that defeats it. Scramble
doesn't work through encryption; it works through destruction. Decrypt
should be named Unscramble, and the description of everything makes perfect
sense as long as you ignore any references to actual encryption.
Sparky simply doesn't make sense. I sincerely doubt there are any deckers
who allow anything other than a fiber-optic cable into their head, and it's
far too easy to (a) build a cyberdeck whose power supply can't be affected
by the code running on it and (b) run your actual visualization code on a
protected processor that won't transmit such a thing to the user. I just don
't believe Sparky, as a concept, would survive one round of the SOTA.
Substitute Blaster for Sparky where you find it.
>>>>>[But it's no problem to have the underlying "pulse" logic
burned
out?]<<<<<
Tar Pit is explained quite well in the source material.
Worms cannot truly invade the MPCP; if it were possible to put code into an
optical chip through running programs that interact with it, you wouldn't
need optical chip burners to create cyberdecks. What Worms do is target
particular security checks in your MPCP and burn them out, and then hide in
the "shadow" that this leaves.
>>>>>[It would be easier to say they don't exist then come up with an even
less plausible explanation. By your reasoning "formatting" the deck and
reinstalling would remove the Worms.]<<<<<
>>>>>[Ken]<<<<<