[Cryptography] Plan to End the Crypto War

james hughes hughejp at me.com
Fri Jan 8 22:01:39 EST 2016

> On Jan 8, 2016, at 3:49 PM, Bill Frantz <frantz at pwpconsult.com> wrote:
> On 1/8/16 at 10:46 AM, hughejp at me.com (james hughes) wrote:
>> My personal conjecture is that split golden key is no better than single golden key because any human process where the split key works, that process can be implemented around a single golden key.
> One advantage a split key has is more security against insider attack. You need N of M insiders inorder to get at the data. If those insiders have to be inside different organizations, all the better.

Nice discussion. Thanks. 

For the internal escrow of a single organization to make sure there is is sufficient collusion for a rare one time event, OK. Makes sense but this is not in this case. 

In just the USA, the FISA court issued 1,588 warrants in 2013 or about 4/day. Assume they are not going down, and assume that this is related to the populations size. That is 1 warrant for every 200k people/yr. Assuming that people in the USA are just as bad as the rest of the world, based on a world population of 7.3B with 1 in 200k “bad", that is ~100/day. Lets test this conjecture: 

With key split: 100 times every day (365/year) the 5 of the 10 people give their part to the (validated) requestor. What are they going to do now? The requestor has the single golden key. The part holders must watch how the golden key, is used, and destroyed to ensure it was used for the purpose it was intended and not stolen. Stolen once, BZZZZ, game over. Do you think we can hide the combiner inside millions of lines of complicated code and expect that it stays secure? Would this need to be repeated for each cell phone operator? or even each cell phone builder? We can’t even reliably build a random number generator.

Without key split: Have a room with 5 locks. all 5 keys need to be present. Yes, they can be different organizations. The five people must now watch the person wielding the golden key. The five people make sure the room is locked (and the 5 alarms set). Stolen once, BZZZZ, game over. Same.

Bottom line, why not put the data in a room with a lot of locks, guards, guns, gates, and process and just do the last step making sure the warrant is followed? Wasn’t that what has already tried? The last step is the problem the rest are red herrings. How is any of this math magic making things better.

Another issue. Given that it is nation states that “need” this recombined golden key and to be able to secretly wield it, the nation states will be a huge amount of time and money put into getting it. Assume they get it, what do we do? Rekey everything?

We are quibbling about the dividing line between math and human process. No one can have control over the golden key or the re-combination of the split golden keys without an incredible amount human process oversight and trust. I believe my conjecture still stands, key escrow (split or not) is ultimately the same. 

Worse case, waving the flag of magic math to politicians provides a fig leaf of “being a better solution” opening the door to "BZZZ, game over". 

There is no simple solution. This is a really gnarly problem, and Shamir Secret Sharing (rebranded as Chaum) is not the silver bullet.

BTW, this still does not work even if single split golden key for each phone. It is the human/geopolitical problems that will fail, not the math.

On Jan 6, 2016, at 8:35 PM, Henry Baker <hbaker1 at pipeline.com> wrote:
> Chaum recognizes, for any single company or government.  So he's given the task to a sort of council system.  When PrivaTegrity's setup is complete, nine server administrators in nine different countries would all need to cooperate to trace criminals within the network and decrypt their communications.  The result, Chaum argues, is a new approach that "breaks the crypto wars," satisfying both the law enforcement agencies who argue that encryption offers a haven for criminals, and also those who argue that it's necessary to hobble mass spying.

Three basic classes of people use cryptography. Plain people living normally, criminals where there would be a broad consensus of them being “bad”, and governments. Imagine how this works. The USG says Snowden is bad, China, Russia say he is good. What happens? Nothing. Seems like this is a UN security council where any one country can veto. This may be what we want (an escrow system that is unusable) but it is not what will make governments happy. 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160108/8cc570d6/attachment.html>

More information about the cryptography mailing list