Help - Search - Members - Calendar
Full Version: Thoughts on Encryption/Decription
Dumpshock Forums > Discussion > Shadowrun
Pages: 1, 2, 3
Smokeskin
Sengir and Rotbart, we're talking about data being recorded by the drone's sensors. Are you saying that the sensors record at some phenomenal rate that can't be kept on storage? Also, the data is being transmitted. Are you saying that transmitters and receivers in SR operate at such a high rate that you can't keep even hours transmission in standard storage? That sounds very strange. Perhaps you're confusing full human sensory recordings with that coming from a few drone sensors.

Since you've chosen to make feeble arguments against only full simsense rigging, obviously the worst case, I assume that you've accepted how much sense one-time pads make for normal data transmission?
JoelHalpern
Very few uses of Encryption have the property that there is a daily rendezvous with a reliable source of one time pads.
Note that just because the drone goes back to base does not usually meet the needs. Corporate drones are often managed by remote riggers, or at least backed up by them.

In general, as has been said in this thread, the problem with one time pads is that key distribution becomes a continuous, high volume, activity. Master key distribution is the single riskiest part of the system, since if the enemy gets that without you knowing, you are completely compromised, without a clue.

On top of that, given that mathematical crypto as we know it has been declared broken, we don't know what they are using. (We don't really know what is broken either, since that depends upon the math that I would not ask the Devs to specify to a sufficient degree.)

Also, for playability, we mostly have to assume that many normal practices either don't work or are not practiced. Given that playability trumps authenticity, we let it go.

(No, I am not a cryptanalyst, cryptologist, cryptographer, nor a designer of security systems. But I do work with a lot of this stuff in my protocol design work. )
Sengir
QUOTE (Smokeskin @ Jan 3 2010, 12:25 AM) *
. Perhaps you're confusing full human sensory recordings with that coming from a few drone sensors.

What makes you think they are that much different? Sure, the emotions are missing, but it makes up for this with constant control data

QUOTE
I assume that you've accepted how much sense one-time pads make for normal data transmission?

You simply didn't bring up that point any more, but if you insist I will repeat myself: What good is an an encryption that requires the secure exchange of a message to securely exchange a message of egual size?
Draco18s
Just like to point out:

Hours upon hours of full human sensory recordings fit on a datachip small enough to fit in your ear.

Or have we forgotten about BTL?
Sengir
Those have been subject to lossy compression, see Unwired wink.gif
Draco18s
QUOTE (Sengir @ Jan 2 2010, 07:16 PM) *
Those have been subject to lossy compression, see Unwired wink.gif


Even so, lossy compression doesn't compact things down as small as you think it does.

Besides, you still have to have somewhere to extract it to in order to "read" it.
Sengir
QUOTE (Draco18s @ Jan 3 2010, 01:18 AM) *
Even so, lossy compression doesn't compact things down as small as you think it does.

UW says it does. Page 185ff to be exact.
Draco18s
QUOTE (Sengir @ Jan 2 2010, 07:20 PM) *
UW says it does. Page 185ff to be exact.


Reading over that makes me wonder how the matrix could be capable of allowing a rigger to jump into a drone at all, if the data is so massive that nodes can't handle it. And the DIR-X are compressed, jumping in isn't.
Smokeskin
QUOTE
QUOTE
Perhaps you're confusing full human sensory recordings with that coming from a few drone sensors.

What makes you think they are that much different? Sure, the emotions are missing, but it makes up for this with constant control data


I'm very sure that the few cameras, microphones, accelerometers and whatever other sensors a drone has it on, hours upon hours of input from these can easily be stored in standard storage. Unwired says that raw simsense recordings cannot. Ergo they are different. Control data is going to be much less than sensory data, there are very few control elements on a drone, or any vehicle for that matter.

Why did you dodge the question of how you could transmit the data, but not store it?

QUOTE ( @ Jan 3 2010, 12:49 AM) *
You simply didn't bring up that point any more, but if you insist I will repeat myself: What good is an an encryption that requires the secure exchange of a message to securely exchange a message of egual size?


Because the two can be separated in time. It is practical and easy to exchange OTPs through direct links at the start of the shift, mission briefing, drone refuelling. During operations, it is not practical or easy to exchange "a message of equal size" through a direct link, but having secure wireless communications is very convenient.


I don't think we need to talk about this anymore, you're clearly not making any effort in thinking these things through for yourself.
nezumi
Why are people suggesting properly implementing OTPs in the game, but leaving regular encryption intentionally hobbled? If you want real encryption in SR, change the rules to permit real encryption. If you implement OTPs as actual encryption, you're just back to the original question of fun/realistic without ever bothering to actually think about it.

Talk about apples and oranges!
Sengir
QUOTE (Smokeskin @ Jan 3 2010, 10:46 AM) *
I'm very sure that the few cameras, microphones, accelerometers and whatever other sensors a drone has it on, hours upon hours of input from these can easily be stored in standard storage.

And what makes you so sure about that?

QUOTE
Why did you dodge the question of how you could transmit the data, but not store it?

Because you didn't ask? A CCTV cam can transmit images all year long, without a single byte of internal storage.

QUOTE
Because the two can be separated in time. It is practical and easy to exchange OTPs through direct links at the start of the shift, mission briefing, drone refuelling.

I people see each other daily, sure. The whole point of telecommunications is that that you don't have to.


QUOTE
I don't think we need to talk about this anymore, you're clearly not making any effort in thinking these things through for yourself.

Well Mr. Thinker, then here's one for you: OTPs are clearly superior to everything we have right now and apparently key exchange is no problem...so why does everybody rely on the slow, wasteful and potentially insecure (who knows, maybe the NSA has found an attack on RSA long ago?) public-key algorithms? Those things are nothing but a crutch, and not even a comfortable one...


@Draco: The same way the Matrix allows you to control a drone on the other side of the planet without any latency wink.gif
Smokeskin
QUOTE (nezumi @ Jan 3 2010, 03:23 PM) *
Why are people suggesting properly implementing OTPs in the game, but leaving regular encryption intentionally hobbled? If you want real encryption in SR, change the rules to permit real encryption. If you implement OTPs as actual encryption, you're just back to the original question of fun/realistic without ever bothering to actually think about it.

Talk about apples and oranges!


I don't think anyone is suggesting to actually use OTPs in game. The discussion is about SRs backstory of why there is no effective encryption begin completely bogus, even if someone did find a way to break mathematical encryption.

I'd probably go with something magical if I had to explain why OTPs don't work. Maybe there's some sort of magical threat that are attracted to random strings, using an OTP carries a high threat of getting a visit from a chaos spirit. Or some artefact of technomancy causes XORing with a random string to produce a random output (that's why people hate technomancers, they took away our privacy!).
Tymeaus Jalynsfein
QUOTE (Smokeskin @ Jan 3 2010, 07:55 AM) *
I don't think anyone is suggesting to actually use OTPs in game. The discussion is about SRs backstory of why there is no effective encryption begin completely bogus, even if someone did find a way to break mathematical encryption.

I'd probably go with something magical if I had to explain why OTPs don't work. Maybe there's some sort of magical threat that are attracted to random strings, using an OTP carries a high threat of getting a visit from a chaos spirit. Or some artefact of technomancy causes XORing with a random string to produce a random output (that's why people hate technomancers, they took away our privacy!).



I think that people are indeed missing the point of Encryption in the game... It is not meant to mimic real world occurrences... Unbreakable encryption would totally eliminate the Hacker as an Archtype... loads of No Fun for anyone...

If you cannot see that the Encryption is clearly designed for slowing people down (sometimes to the point of inability, but usually not... Higher rated Encryption slows for far longer than lower rated Encryption), then you really are looking at it the wrong way... Is it realistic... No, but any other solution removes teh immediacy of hacking usefulness... There are options in Unwired for those who like to make things even harder... Strrong Encryption Intervals can be from 1 Turn to 12 Hours; and this will indeed slow down a hacker even more... implementing this would make a Hacker more like what you get in Neuromancer, but that is not a Good Thing in that the Archtype is truly no longer a viable character and we are back to the Decker of Versions past... again, no fun...

The other is dynamic encryption, and, while not exactly as useful as Strong Encryption, it can make a system almost impossible to crack if the rating is very high, as the dynamic encryption will tend to score as many hits as "most" Decryption attempts... again, it may be somewhat exciting at first, but in the end it eliminates the Acrhtype, and again we are back to Deckers...

Now... in the end, we have encountered both such systems in the game, but they are not ubiquitous by any means... they were designed for tailored and sculpted systems that took much research and additional resources so that we could social engineer the key (which we have been working on in game for about a year now for that "Big" Run)... takes a while, but once you have it, the system is then vulnerable... Why are these schemes not included with each and every system? Because it would reduce the options for the Hacker to just one... Social Engineering... Way Boring and all to predictible...

My 2 Nuyen anyways...

Keep the Faith
ZeroPoint
A few things about SR encryption to think about.

First off, with all forms of encrypted transmission, you can't capture any of the data until you have broken the encryption.
This is definitely not like real world encryption. Sounds a lot more like quantum cryptography to me.
But quantum cryptography isn't really an encryption scheme, its more of a method of transmitting encrypted data. We would still have to implement some method of encryption. And in SR with all standard forms of encryption being broken, unfortunately that's all that is left. But when combined with quantum cryptography, now any would be listeners can't just snoop it, they have to try to decrypt it, alerting the sender that an encryption attempt is being made so that they can stop transmission, change channels, change keys, or any number of other things. Making encryption a lot more useful.

There are still problems with trying to rationalize that, but its the best I can think of at the moment.

Static file encryption is still gonna be mostly useless. But for that at least you will be able to rely on the system security.

Also, the problem with OTP methods is that they still rely on random strings. There is no such thing as random in the computing world. And the larger the random string, the more likely you are that your random string is gonna break itself as suddenly it starts repeating (given that it might be after a few trillion digits, but if your transmitting terabytes of data on a regular basis that's gonna happen a lot). Which basically means once someone knows how your random strings are seeded its not really gonna take very long to break your OTP. If there is some magic equation that can break any and all PSK and PKI encryption methods, then its not gonna be that hard to break an OTP.

And lets not cry doom over encryption being unbreakable. If all encryption was always unbreakable and everything was encrypted, there are still so many other ways for a hacker to be a hacker and get the job done.
kzt
QUOTE (Tymeaus Jalynsfein @ Jan 3 2010, 08:27 AM) *
I think that people are indeed missing the point of Encryption in the game... It is not meant to mimic real world occurrences... Unbreakable encryption would totally eliminate the Hacker as an Archtype... loads of No Fun for anyone...

No it doesn't. We have strong encryption now and people still break into our systems. if we encrypted them to hell it wouldn't matter. Encryption isn't some frigging magic wand that mysteriously makes systems secure.

Tymeaus Jalynsfein
QUOTE (kzt @ Jan 3 2010, 01:53 PM) *
No it doesn't. We have strong encryption now and people still break into our systems. if we encrypted them to hell it wouldn't matter. Encryption isn't some frigging magic wand that mysteriously makes systems secure.


The point is, it is no longer Shadowrun, it is Neuromancer... and the hackers do not crack the encryption in a timeframe that is even playable... so, as such, the archtype is now relegated to the Decker of old, which many people never used as a playable character... why, you ask? Because it was NO FUN...

It is a game, not a reality simulation... people just need to get over the disparities and play...

Keep the Faith
kzt
QUOTE (ZeroPoint @ Jan 3 2010, 12:27 PM) *
Also, the problem with OTP methods is that they still rely on random strings. There is no such thing as random in the computing world. And the larger the random string, the more likely you are that your random string is gonna break itself as suddenly it starts repeating (given that it might be after a few trillion digits, but if your transmitting terabytes of data on a regular basis that's gonna happen a lot). Which basically means once someone knows how your random strings are seeded its not really gonna take very long to break your OTP. If there is some magic equation that can break any and all PSK and PKI encryption methods, then its not gonna be that hard to break an OTP.

OTPs are typically created via specialized hardware that relies on actual random quantum events, such as thermal noise or nuclear decay. See Hardware Random Number Generators. They are really random. There is a reason why OTPs are considered unbreakable, it's because they are part of the class of problems mathematicians call undecidable and hence can't be decisively answered.

"Think about it. If I give you a message encrypted with a one–time pad and then give you the key which I claim decrypts it, do you have any way to prove that I’m right? Do you have any way to prove that I’m wrong? All keys are equally likely, all decryptions are equally likely, and there is absolutely no mathematical way to tell when you get the right key as opposed to any of the umpteen billions of wrong ones.

"This is an undecidable cryptosystem."
kzt
QUOTE (Tymeaus Jalynsfein @ Jan 3 2010, 02:03 PM) *
The point is, it is no longer Shadowrun, it is Neuromancer... and the hackers do not crack the encryption in a timeframe that is even playable... so, as such, the archtype is now relegated to the Decker of old, which many people never used as a playable character... why, you ask? Because it was NO FUN...

You keep ignoring that encryption just doesn't matter to hackers. What encryption does is allow electronic communication to be used for non-trivial purposes. Like fund transfers and not having everyone listen to your calls and perfectly simulate people calling you.
Draco18s
QUOTE (Sengir @ Jan 3 2010, 09:43 AM) *
And what makes you so sure about that?


Because its not wet simsense, which includes emotional data and dozens of other things that your brain does.

QUOTE
Because you didn't ask? A CCTV cam can transmit images all year long, without a single byte of internal storage.


I believe we're talking EXTERNAL storage here.

QUOTE
I people see each other daily, sure. The whole point of telecommunications is that that you don't have to.


Drones aren't people and still have to be serviced by someone. In person. SR doesn't have repair droids.

QUOTE
Well Mr. Thinker, then here's one for you: OTPs are clearly superior to everything we have right now and apparently key exchange is no problem...so why does everybody rely on the slow, wasteful and potentially insecure (who knows, maybe the NSA has found an attack on RSA long ago?) public-key algorithms? Those things are nothing but a crutch, and not even a comfortable one...


Because the hassle of one time pads is so exorbiently high that we use a "good enough" solution called the Public Key Method. Which in the fucking real world is neigh unbreakable. This is not true in ShadowRun.

QUOTE
@Draco: The same way the Matrix allows you to control a drone on the other side of the planet without any latency wink.gif


Everyone knows the matrix is backwards. It was written with a "you are in the node" rather than the way the internet works of "the node sends data to you." They've kept this backward stuff around, despite its painful wrongness. Oh. And in ShadowRun if you have a hacker at node A exploring node C through node B and jam/crash/remove node B, nothing happens. The hacker is in node C and his body is at node A, node B is irrelevant.

One (two) more reason(s) why the Matrix rules need an update.
tete
QUOTE (Tymeaus Jalynsfein @ Jan 3 2010, 09:03 PM) *
Because it was NO FUN...


IMHO that wasnt it at all, I've played several deckers through out the years and GMed for more. The reason most people didnt play them was the go get a pizza problem or misconceived idea that the rules were horribly complex. In reality the only suckage was the system map(1e-2e)/host(VR2.0-3e)/node(4e). By having multiple interaction points you have now created a minigame that no one else is a part of. Its the same as having the face have to talk to multiple npcs to finally get the desired results. By limiting the interaction points you take a solo game back into an acceptable aside. Also if you were to have the multiple points and the decker/hacker did his thing but its going to take a couple hours in game you can now bounce back and forth between players so they dont have to get a pizza. By having a 12 point matrix map that takes 10 seconds in game to go through you now have to play the whole thing out as a GM without interacting with the other players because their characters had enough time to walk to the fridge and grab a beer even though out of game it took 2 hours to complete all the rolling.

Ideally if you want Deckers/Hackers in your game you would have 1/2 dozen or less interaction points with each interaction point taking 30 minutes or so of in game time, that way you can bounce between parties and still have a reasonable aside for the Decker/Hacker.
Tymeaus Jalynsfein
QUOTE (tete @ Jan 3 2010, 05:00 PM) *
IMHO that wasnt it at all, I've played several deckers through out the years and GMed for more. The reason most people didnt play them was the go get a pizza problem or misconceived idea that the rules were horribly complex. In reality the only suckage was the system map(1e-2e)/host(VR2.0-3e)/node(4e). By having multiple interaction points you have now created a minigame that no one else is a part of. Its the same as having the face have to talk to multiple npcs to finally get the desired results. By limiting the interaction points you take a solo game back into an acceptable aside. Also if you were to have the multiple points and the decker/hacker did his thing but its going to take a couple hours in game you can now bounce back and forth between players so they dont have to get a pizza. By having a 12 point matrix map that takes 10 seconds in game to go through you now have to play the whole thing out as a GM without interacting with the other players because their characters had enough time to walk to the fridge and grab a beer even though out of game it took 2 hours to complete all the rolling.

Ideally if you want Deckers/Hackers in your game you would have 1/2 dozen or less interaction points with each interaction point taking 30 minutes or so of in game time, that way you can bounce between parties and still have a reasonable aside for the Decker/Hacker.



Which in my opinion was no fun... I was often part of that Pizza Crowd...

Keep the Faith
tete
I just don't feel 4e improved this in anyway its still a multiple node map with microseconds of game time. You still either keep it to a single or few nodes or house rule the time it takes up. Both of which people have been doing since 1e.
Sengir
QUOTE (Draco18s @ Jan 4 2010, 12:46 AM) *
Drones aren't people and still have to be serviced by someone. In person. SR doesn't have repair droids.

It does, see Lockheed Vulcan in Arsenal wink.gif
Besides, I was talking about actual people. If Mr. J wants to send me a job offer, but has to send a physical messenger first (with escort to make sure the message remain confidential), that defeats the whole purpose of telecommunication...


QUOTE
Everyone knows the matrix is backwards. It was written with a "you are in the node" rather than the way the internet works of "the node sends data to you." They've kept this backward stuff around, despite its painful wrongness. Oh. And in ShadowRun if you have a hacker at node A exploring node C through node B and jam/crash/remove node B, nothing happens. The hacker is in node C and his body is at node A, node B is irrelevant.

I'm afraid you are confusing the 'trix and Astral Space here. A hacker's consciousness stays inside his own body, which is why Math SPUs or Rigger equipment even do anything in the matrix. And killing somebody while in VR does not leave a lingering ghost like it does with an astrally projecting mage.
Smokeskin
QUOTE (Sengir)
Besides, I was talking about actual people. If Mr. J wants to send me a job offer, but has to send a physical messenger first (with escort to make sure the message remain confidential), that defeats the whole purpose of telecommunication...


Oh you were? But you said

QUOTE (Sengir)
...and half an hour before the end of the drone's duty cycle some runners come in and the resident spider realizes he can't jump into the drone because the OTP is nearly "drained". Ooops.
I might also point out the Unwired section on simsense, which says that raw simsense recordings are HUGE and require "expensive storage", so putting sufficiently large pads into every drone and maglock would be complicated to say the least.


It seems you were talking about drones and maglocks. You just weren't making any sense, so your point was as usual easily refuted, and then you tried to make it seem like you were talking about something else to save face, but fell right on it instead.


QUOTE (Sengir)
QUOTE (Smokeskin)

I'm very sure that the few cameras, microphones, accelerometers and whatever other sensors a drone has it on, hours upon hours of input from these can easily be stored in standard storage.

And what makes you so sure about that?


What makes me sure that hours of camera footage and audio recordings can be kept in standard storage? That you can even doubt this shows exactly why I said I was done discussing this with you.
Sengir
QUOTE (Smokeskin @ Jan 4 2010, 12:05 PM) *
Oh you were? But you said

This:
QUOTE
QUOTE

QUOTE

QUOTE

QUOTE

What good is an an encryption that requires the secure exchange of a message to securely exchange a message of egual size?

Because the two can be separated in time. It is practical and easy to exchange OTPs through direct links at the start of the shift, mission briefing, drone refuelling.

I[f] people see each other daily, sure. The whole point of telecommunications is that that you don't have to.

Drones aren't people and still have to be serviced by someone. In person. SR doesn't have repair droids.

It does, see Lockheed Vulcan in Arsenal wink.gif
Besides, I was talking about actual people.



QUOTE
What makes me sure that hours of camera footage and audio recordings can be kept in standard storage?

Not quite:

In general, a standard vehicle sensor package (Capacity 12),
will contain the following sensors:
• Atmosphere Sensor (taking up 1 Capacity)
• 2 Cameras (front and back, taking up 2 Capacity)
• 2 Laser Range Finders (front and back, taking up 2 Capacity)
• 2 Motion Sensors (front and back, taking up 2 Capacity)
• Radar (taking up 5 Capacity)


Plus the sensors which make the rigger actually feel acceleration, tilt, or engine data as if those were senses of his own body.
StealthSigma
QUOTE (Smokeskin @ Jan 1 2010, 05:42 PM) *
The reason one-time pads aren't used in the real world is because public key encryption is practically unbreakable. That isn't true for SR.


PK encryption is practically unbreakable because most people don't have the computing resources to brute force the encryption. This will -not- change in SR. One of the major reasons we use current strength keys is due to limitations of computing power. So while SR available hardware will be able to easily break modern encryption key sizes, key sizes will increase thus increasing the required processing power needed to break it. Moore's law is a benefit to BOTH sides of the encryption battle.

--

QUOTE (Draco18s @ Jan 1 2010, 06:46 PM) *
There's already a disconnect there as to how the people who make fake SINs manage to do so while the PCs can not. The cost and interval implies that takes a few days to a couple weeks for these organizations to crank out a new SIN for someone (would you charge $4000 for 9 months worth of highly illegal work? No. Would you charge $4000 for a week's worth of highly illegal work? Hell yeah you would).


That depends.... how many fake SINs can I make at once?

--

QUOTE (ZeroPoint @ Jan 3 2010, 03:27 PM) *
First off, with all forms of encrypted transmission, you can't capture any of the data until you have broken the encryption.
This is definitely not like real world encryption. Sounds a lot more like quantum cryptography to me.
But quantum cryptography isn't really an encryption scheme, its more of a method of transmitting encrypted data.


Quantum cryptography is a method of determining encryption keys. It allows you to create symmetric keys without having to go through the rigors of distributing them. It would make asymmetric encryption obsolete and unnecessary. It won't be as secure as physically transporting the key and using a chain of custody, but it will be good enough to replace any public key schemes.

QUOTE (ZeroPoint @ Jan 3 2010, 03:27 PM) *
We would still have to implement some method of encryption. And in SR with all standard forms of encryption being broken, unfortunately that's all that is left. But when combined with quantum cryptography, now any would be listeners can't just snoop it, they have to try to decrypt it, alerting the sender that an encryption attempt is being made so that they can stop transmission, change channels, change keys, or any number of other things. Making encryption a lot more useful.


The encryption technique based of prime numbers -may- be broken. However their are encryption techniques which have been mathematically proven to be unbreakable.

QUOTE (ZeroPoint @ Jan 3 2010, 03:27 PM) *
There are still problems with trying to rationalize that, but its the best I can think of at the moment.

Static file encryption is still gonna be mostly useless. But for that at least you will be able to rely on the system security.

Also, the problem with OTP methods is that they still rely on random strings. There is no such thing as random in the computing world. And the larger the random string, the more likely you are that your random string is gonna break itself as suddenly it starts repeating (given that it might be after a few trillion digits, but if your transmitting terabytes of data on a regular basis that's gonna happen a lot). Which basically means once someone knows how your random strings are seeded its not really gonna take very long to break your OTP. If there is some magic equation that can break any and all PSK and PKI encryption methods, then its not gonna be that hard to break an OTP.

And lets not cry doom over encryption being unbreakable. If all encryption was always unbreakable and everything was encrypted, there are still so many other ways for a hacker to be a hacker and get the job done.


It's actually not even necessary for encryption on a static file to need to be broken by PCs. Given the storage space available in a data steal a hacker can just dump all the files and deliver them to Johnson, unless the Johnson is paying for the files to be decrypted. Let the corps with their significantly more vast computing power go through the rigors of breaking the encryption. It still boils down to the whole at rest versus in motion issue of defending data. Data in transit has always been more vulnerable than data sitting on a server and there's no reason why this would change. I'm willing to suspend disbelief for encryption on data transmissions to be broken in minutes by PCs. I just cannot cross that barrier for data that is at rest.
Smokeskin
QUOTE (Sengir @ Jan 4 2010, 03:09 PM) *
Plus the sensors which make the rigger actually feel acceleration, tilt, or engine data as if those were senses of his own body.


Oh yeah, that's a lot of data, because it makes the rigger feel it, just like wet simsense! It can't just be a number for each of the six degrees of freedom and then something at the rigger's end generates the simsense signal - of course all drones pack the hardware and software to turn simple sensor input into something a human can feel.
Smokeskin
QUOTE (StealthSigma @ Jan 4 2010, 03:27 PM) *
PK encryption is practically unbreakable because most people don't have the computing resources to brute force the encryption. This will -not- change in SR. One of the major reasons we use current strength keys is due to limitations of computing power. So while SR available hardware will be able to easily break modern encryption key sizes, key sizes will increase thus increasing the required processing power needed to break it. Moore's law is a benefit to BOTH sides of the encryption battle.


All wrong.

First, the problem scales differently for the encryption and decryption size. I can't remember how it scales, but the decrypting is exponential while the encrypting is polynomial - as computers get faster, if we devote the same processor time to encrypting, crackers will have to spend more time breaking the encryption.

The main reason for this is that prime factorization can only be done in exponential time with known methods. No one has proven that prime factorization can't be done faster though - and in SR it specifically mentions that someone figured out a clever, fast way to do this. This is also reflected in the rules, where encryption is completely trivial to break, very much unlike today.
Draco18s
QUOTE (Sengir @ Jan 4 2010, 05:24 AM) *
I'm afraid you are confusing the 'trix and Astral Space here. A hacker's consciousness stays inside his own body, which is why Math SPUs or Rigger equipment even do anything in the matrix. And killing somebody while in VR does not leave a lingering ghost like it does with an astrally projecting mage.


No I'm not. I am well aware that the hacker is still in his own conscious body the whole time.

HOWEVER, the rules for how matrix hacking works make me think that the original creators had No Fucking Clue how the internet worked and thought that a hacker would actually be in the system.

I mean, how else does:

1) sitting in a printer's IO port cause malformed printouts (see: Psychotrope where this happens)
2) hacking into a node give you access to the CPU, SPU, Data, or one of several other things single physical machine, but not at the same time? Because each is a different node?
3) opening and reading a file happens instantly, but downloading a copy takes time?
4) lack of lag
(I'm sure there are more)

work?
StealthSigma
QUOTE (Smokeskin @ Jan 4 2010, 11:05 AM) *
All wrong.

First, the problem scales differently for the encryption and decryption size. I can't remember how it scales, but the decrypting is exponential while the encrypting is polynomial - as computers get faster, if we devote the same processor time to encrypting, crackers will have to spend more time breaking the encryption.

The main reason for this is that prime factorization can only be done in exponential time with known methods. No one has proven that prime factorization can't be done faster though - and in SR it specifically mentions that someone figured out a clever, fast way to do this. This is also reflected in the rules, where encryption is completely trivial to break, very much unlike today.


Exponential growth is significantly faster than polynomial growth. In fact, if encryption is indeed polynomial growth, then decryption (being exponential growth) will begin to lag behind encryption in requisite key sizes. The fact that encryption is polynomial does nothing to break the point that an increase in computing power will not make it easier to break encryption.

So I'm not sure where your disagreement lies, aside from being a pedant with a minute detail which only serves to reinforce the point, rather than weaken it.
kzt
QUOTE (Smokeskin @ Jan 4 2010, 08:05 AM) *
First, the problem scales differently for the encryption and decryption size. I can't remember how it scales, but the decrypting is exponential while the encrypting is polynomial - as computers get faster, if we devote the same processor time to encrypting, crackers will have to spend more time breaking the encryption.

The main reason for this is that prime factorization can only be done in exponential time with known methods. No one has proven that prime factorization can't be done faster though - and in SR it specifically mentions that someone figured out a clever, fast way to do this. This is also reflected in the rules, where encryption is completely trivial to break, very much unlike today.

Decryption isn't an exponential function. Symmetric key encryption is exponential. 2^4=16, 2^8=256. Not sure how public key encryption works.

I think decryption is linear, but I may be wrong.

There are multiple approaches to the concept of public key encryption, not just prime factorization. The one NIST and the NSA like is based on elliptic curves, though there are other approaches like "discrete logarithms". In fact there seem to at least 6 different approaches that seem workable.
Smokeskin
QUOTE (StealthSigma @ Jan 4 2010, 04:38 PM) *
Exponential growth is significantly faster than polynomial growth. In fact, if encryption is indeed polynomial growth, then decryption (being exponential growth) will begin to lag behind encryption in requisite key sizes. The fact that encryption is polynomial does nothing to break the point that an increase in computing power will not make it easier to break encryption.

So I'm not sure where your disagreement lies, aside from being a pedant with a minute detail which only serves to reinforce the point, rather than weaken it.


Yeah, that's what I'm saying - as processing power increases, breaking encryption becomes relatively harder. You said that increases in processing power would be equal gain for both sides, which is wrong.
StealthSigma
QUOTE (kzt @ Jan 4 2010, 01:17 PM) *
Decryption isn't an exponential function. Symmetric key encryption is exponential. 2^4=16, 2^8=256. Not sure how public key encryption works.

I think decryption is linear, but I may be wrong.

There are multiple approaches to the concept of public key encryption, not just prime factorization. The one NIST and the NSA like is based on elliptic curves, though there are other approaches like "discrete logarithms". In fact there seem to at least 6 different approaches that seem workable.


I was talking about encryption times against cracking. Since key size growth is typically exponential, the time to crack the encryption is also exponential. Decryption is a trivial process.

So even if the time spent encrypting is exponential, it's still keeping pace with cracking. If encrypting/decrypting have a polynomial growth, then as key size increases, the amount of additional computing power necessary to encrypt will be less than the computing power necessary to crack a key in the same period of time. An increase in processing speed is only a benefit to crackers IF encryption key size does not also grow. If key size is also growing with processing power then cracking encryption is more the domain of locating the key rather than brute forcing. Cracking encryption will be the domain of stealing the key rather than trying to brute force or look for patterns to try to discern the key.

Public keys also grow exponentially. I may be wrong, but 2024bit asymmetric encryption is about as strong as 256bit symmetric encryption. The next asymmetric key size will be 4048bit, followed by 8096bit and so on.
Smokeskin
QUOTE (kzt @ Jan 4 2010, 06:17 PM) *
Decryption isn't an exponential function. Symmetric key encryption is exponential. 2^4=16, 2^8=256. Not sure how public key encryption works.

I think decryption is linear, but I may be wrong.

There are multiple approaches to the concept of public key encryption, not just prime factorization. The one NIST and the NSA like is based on elliptic curves, though there are other approaches like "discrete logarithms". In fact there seem to at least 6 different approaches that seem workable.


It isn't about the type of function, it is about how the processing power needed to solve the problem scales with the problem size. The time needed to decrypt grows exponentially as the key length grows, while encrypting only grows polynomially. If for key length x encryption takes x^2 time, and decrypting takes 2^x time, then as you ramp up key length, decrypting soon starts taking MUCH longer.

I didn't know there were public key encryption schemes not based on prime factorization, I'll look into that. Good to learn new stuff.
Sengir
QUOTE (Draco18s @ Jan 4 2010, 04:07 PM) *
HOWEVER, the rules for how matrix hacking works make me think that the original creators had No Fucking Clue how the internet worked

Sure, that's the beauty of 80s SciFi

Then again, who would want to calculate lag, packet loss and everything for his drones?



PS: And RSA has a runtime of O(log n) for an input of n bits length, both for encryption and decryption (with the correct key).
Draco18s
QUOTE (Sengir @ Jan 4 2010, 02:31 PM) *
Sure, that's the beauty of 80s SciFi

Then again, who would want to calculate lag, packet loss and everything for his drones?


My point is, it worked fine in the 80s. Its now the Twenteens. Even my grandmother understands how a computer fetches internet data (my grandfather on the other hand is constantly amused at how his watch talks to him, but he's never been technically minded and he's losing what little mind he had).

Also, packet loss need not be calculated. The modern internet has already evolved to a point where packet loss is insignificant until it exceeds about 40% (SR terms: increased chance of glitching) and at around 75% you might as well be disconnected (dumped). There, simple rules for dropped packets.

Lag: base it on the distance (number of nodes) away the hacker is (for interacting with the real world--cybercombat already Makes No Sense so we can ignore any effect lag might have, and other operations would be completely unaffected). -1 die per 2 intermediate nodes (eg, comlink - one - two - three - four - drone is a -2). Which would only come up if the hacker is out of signal range of the device. I'd also put in a flat -3 for the satelite uplink due to sheer physical distance (satellite phones are NOT REAL TIME, there's a good 3 second delay there). Interestingly this might bring a drones pilot ability back up to on par with the average security rigger who might be four or six nodes away (seriously Pilot 3 + Targeting 3 = a dice pool of 6 = a kid with a shotgun is more dangerous).

Might also include an additional -1 for all nodes that has 0 signal for all matrix tests (routing your hack attempt through the vending machine is not going to be pretty). Mainly because I am of the opinion that hacking cyberware is one of the least effective things you can do in combat.

Anyway, the "number of nodes" thing would only come up rarely ("I hack from the van! Across town!") or doing really bizarre stuff ("I log into my cybereye, which I rolled down the hall, hit the vending machine to relay the signal to the fire alarm....")
Sengir
QUOTE (Draco18s @ Jan 4 2010, 09:28 PM) *
Also, packet loss need not be calculated. The modern internet has already evolved to a point where packet loss is insignificant until it exceeds about 40% (SR terms: increased chance of glitching)

Insignificant because the underlying protocols retransmit lost packets, but that takes time...
Also remember that our modern internet is still a mostly wired affair, where packet loss, if it occurs, is nearly always caused by congestion and not actual problems in the transmission (in fact the TCP flow control mechanisms are built on that aussumption). In wireless environments, it's a whole new game.
Draco18s
QUOTE (Sengir @ Jan 5 2010, 05:45 AM) *
Insignificant because the underlying protocols retransmit lost packets, but that takes time...
Also remember that our modern internet is still a mostly wired affair, where packet loss, if it occurs, is nearly always caused by congestion and not actual problems in the transmission (in fact the TCP flow control mechanisms are built on that aussumption). In wireless environments, it's a whole new game.


Yes, and you can still assume that the hardware takes care of it for you. Voila, the whole issue is eliminated and SIMPLICITY OF THE RULES reigns supreme.
Sengir
Like I said, retransmissions are usually done automatically, but the point is that they take time. So worse connection = more lag. Also, your idea requires a map of all nodes in the area, plus constantly doing all the routing and roaming stuff to determine how many hops the signal takes... trust me, that's already complicated enough if a machine does it ^^
Draco18s
No, it doesn't, as I said twice. The only time it comes up is if there is not a direct connection from the hacker to the target, based on signal rating. At which point you assume the shortest possible connection that can be made and just need to count the number of nodes. The only time you would use anything different is when the player specifies their route, which also incurs the "node has 0 signal" penalty, if applicable (which only ever comes up in the assumed shortest route if you're hacking cyberware).

The assumed shortest route would be based on distance and an average signal rating. No books on me, so distance divided by whatever the distance is for a R3 or R4 signal.
Sengir
QUOTE (Draco18s @ Jan 7 2010, 12:15 AM) *
No, it doesn't, as I said twice. The only time it comes up is if there is not a direct connection from the hacker to the target, based on signal rating.

Which we can translate as "always" wlog. The Matrix is a mobile ad-hoc network, direct connection would be the exeption and not the rule. Even if two nodes are in each other's maximum signal range, power control will often force them to use far less power (you don't want two guys to drown out the whole network) and multiple hops.

QUOTE
At which point you assume the shortest possible connection that can be made and just need to count the number of nodes.

That assumes a perfect routing through a constantly fluctuating net, already nothing short of magic. Additionally we are talking about a packet-switched network, so the route your data takes can be different for each packet, because the best route is constantly changing with nodes entering or leaving the net, moving around in it, link congestion fluctuating and so on.


In short, your proposal for realism assumes completely unrealistic conditions ("I've got a private channel for all my drones, I can send at whatever power level I like"), or drives complexity though the roof. wink.gif
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Dumpshock Forums © 2001-2012