Encryption too weak ?, Broken Options

[Serbitar]

[Crusher Bob]

change to rating^2? or maybe 2^rating? The base time could use some fixing as well...

rating^2 gives the following difficulties:

1..1 (3 dice needed)
2..4 (12 dice needed)
3..9 (27 dice needed)
4..16 (48 dice needed)
5..25 (75 dice needed)
6..36 (108 dice needed)

2^Rating gives to following difficulties:
1..2 (6 dice)
2..4 (12 dice)
3..8 (24 dice)
4..16 (48 dice)
5..32 (96 dice)
6..64 (192 dice)

assuming program rating 6 (easy to get, relatively cheap)
and the following skill ratings, limiting rolls to skill rating, here are the total dice rolled:

skill 1: 7 dice (7x1)
skill 2: 16 dice (8x2)
skill 3: 27 dice (9x3)
skill 4: 40 dice (10x4)
skill 5: 55 dice (11x5)
skill 6: 72 dice (12x6)

This lets someone will skill 6 crack rating 5 encryption almost all of the time using the first system (in ~18 seconds) and need plenty of luck to crack rating 5 encryption using the second system...

Of course the elapsed times are too fast as well.

[Serbitar]

In my example above, the rating of "decrypt" would be the "skill" rating "response" would qualify as the attribute. It is funny that neither encrypt nor decrypt use the "electronic warfare" skill, although the utility is linked to the skill.

I would change this "decrypt + response" into "decrypt + electronic warfare" for streamlining. Response is always in the equation as it is constraining utility ratings . . .

rating^2 sounds reasonable. A base timr of "enc rating" combat turns might be ok.

[Crusher Bob]

another possibility is to make the retest times for decryption very long, so that many things can be decrypted, if you don't mind waiting for a few hundred years... Of course this makes good encryption esentailly unbeatable, and since runners will almost always be going against people with good encryption this will remove the 'ability' to decrpyt whatever they want. Thankfully this is not 'game breaking' as both social engineering and the rubber hose method of decryption will usually have good odds of breaking the encryption.

[Rotbart van Dain...]

[Magnus Jakobsson]

To me, too many dice rolls to accomplish a single thing seems boring. I would keep the system as it is, and limit the number of rolls to 4. That's not even changing the rules - the amount of rolls is supposed to be the GM's decision. Rolls = Dice Pool is only a suggestion (see page 58).

Assuming you got the best decryption program possible for you (that is, equal to your Response rating), you would on average be able to break encryptions one level higher than your program. Here's the table:

Decryption (and Response) 1 = 2,6 hits (beats Encryption 1)
2 = 5,3 hits (beats Encryption 2)
3 = 8 hits (beats Encryption 4)
4 = 10,6 hits (beats Encryption 5)
5 = 13,3 hits (beats Encryption 6)
6 = 16 hits (beats Encryption 8 )

If you want better encryption, a smoother curve and faster gameplay, you could limit the number of rolls to 3. Then you can't even count on beating encryptions of the same rating as your decryption program. Sounds to harsh for me, but here's the table:

Decryption (and Response) 1 = 2 hits (beats Encryption 1)
2 = 4 hits (beats Encryption 2)
3 = 6 hits (beats Encryption 3)
4 = 8 hits (beats Encryption 4)
5 = 10 hits (beats Encryption 5)
6 = 12 hits (beats Encryption 6)

Allso, The Electronic Warfare Specialization "Encryption" has got to be the most worthless Specialization in the book...

[Rotbart van Dain...]

Until Unwired, possibly.

[Lebo77]

The poor effectiveness of encryption could be due to advances in P vs. NP problems. The ability to solve NP -hard problems in polynomial time would make decryption of most modern schemes trivial given the massive computer power availible to 2070 electronics.

Assuming Moore's law continues to hold, the computing power availible come 2070 will be about 9x10^12th (or 2^43rd) times as powerfull as what is availible now. With a highly efficient prime number sive, public-key encryption would be easy to crack.

[Chandon]

Lebo77:
It's true that math advances could make cracking public key encryption much easier. On the other hand, private key cryptography couldn't be broken so easily unless there have been developments that give everyone infinate processing power (such as some of the dream predictions for quantum computing). Even with infinate processing power, it's still possible to make unbreakable crypto by ensuring that multiple keys can decrypt the file to different valid plaintexts.

In the context of Shadowrun, breakable encryption might make things more fun (although having to hack into where the key is stored can be fun too).

A rule like "If the rating of your encryption program doesn't exceed the rating of the encryption, the extended test interval is 1 hour." might be interesting.

[Crusher Bob]

Well, you do lose the 'story mechanic' of 'needing X time' to decrypt the file. It either happens in seconds or not at all. Also it makes sitting down and breakingthe encryption preferable to any number of schemes to obtain the encryption key.

Personally, I'd prefer to keep encyrption 'hard' as more of the party is likely to be involved in 'key theft' that just the hacker.

[hades]

[Crusher Bob]

In general, secure encryption algorytms are public. It's just that there are no efficient ways to break them, so it dosen't matter if you know someone is using xyz encryption angorythm, you can't wait for the universe to grow cold to find out what was said.

Usings Moore's law, some quick google results, and a bit of guessing, a PGP message encrypted with a 1024 bit ket would take a pretty good 2070 desktop ~ 9 minutes to crack. A PGP message encrypted with a 2048 bit key would take roughly 17,000 years to crack. This is, of course, assuming no 'improved' attacks against PGP are found in the intervening time.

(For a bit of fun, this is assuming that a 2070 desktop computer can do a about ~9E20 useful operations per second.)

So, the short version is that encryption that is 'pretty good' today could still possible give headaches to our great grandchildren...

[hades]

[Crusher Bob]

This, of course, assumes that no polynomial time algorithm exists for finding large prime factors exists. If such an algorithm were discovered, most, if not all, of current encryption schemes (barring one time pads) would become crap almost overnight.

[Shadow_Prophet]

[hades]

[Crusher Bob]

I was making no comment about future encryption schemes. It was an addendum to my post about cracking PGP. In that the difficulty of cracking PGP lies in finding large prime factors of numbers. This current algorithms to do this take 'super' polynomial time, this is why PGP would take so long to crack using current algorithms.

[warrior_allanon]

stupid question, and it only applies to text encryption, but what would the result be if you ran each leter through a letter number interface filtering the sequence through 5 dead languages and one non-sensical language, would that make it harder to crack you think

[Vaevictis]

[Shadow_Prophet]

[Crusher Bob]

Maybe your are refering to a breaking of RSA-129 in 1994?

Linky

[Shadow_Prophet]

[Crusher Bob]

There's also the cracking of DES in ~1999.

Linky

[NightRain]

More likely you're talking about distributed.net and their breaking of either RC5 or DES II

http://encyclopedia.thefreedictionary.com/...Distributed.net

[Vaevictis]

He can't possibly be referring to any of those based on his statement of the "best encryption algorithm out there today."

DES-56 was considered near obsolete by most experts by the time I learned about it in 1995. At that point in time, they were basically saying don't use it -- it could fall apart at any time.

RSA-129 is not relevant; modern public key systems are suggested to use a minimum of 1024 bits, which is a number that is 308 digits in length. People these days often go for a 2048 bit key, which is a truly obscene number.

DES-II was just DES-64, iirc. Not even close to SOTA at the point it happened.

As far as cracking encryption goes, given current methods and the rate of improvement of technology and methods, modern day encryption generally provides "good enough" protection for encrypted data for about 20 years after the introduction of the technology. Often times, simply increasing the bitsize of the key can increase the lifetime of the method.

Remember, each bit size increases the keyspace by twice as much as the previous bit. The original DES used a 56 bit key, basically a 16 digit number. DES-64 was a 19 digit number. Modern symmetric key systems use at least 128 bits -- a 38 digit number. That's a huge difference in keyspace sizes.

That's why I was suprised by his statement that someone broke SOTA encryption in a matter of days. If that was the case, they either got lucky and it was a unique incident (that can happen on brute forces, it's just very unlikely), or someone proved P=NP or invented a practical quantum computer or something like that. If that's the case, that is *crazy* stuff.