[Cryptography] DoJ/FBI's "nuclear"/Lavabit option

[Henry Baker]

At 01:38 PM 3/17/2016, Jerry Leichter wrote:
>> When IBM was faced with anti-trust litigation from the U.S. Govt in the late 1960's, it famously delivered the subpoena'd discovery documents in a number of moving-van-type semi-trailer-truck-fulls of boxes.  I think it took years for the govt to wade through the documents....
>
>None of this would work today to do anything but get the company fined or its people put into jail.  The courts have established rules about "E-discovery" which require, for example, that if information that you have to supply is available in a machine-readable format, you have to deliver it that way.
>
>The rules, in fact, can impose some big costs.  Some of the rules around e-mail are particularly strict and intrusive.  And after Enron, the rules about data retention are pretty sharp-edged.
>
>"E-discovery software" is a thing these days....

It's time for a digital version of Al Gore's "Lock Box":

Encrypt discoverable data using random keys of carefully selected lengths, and then destroy the keys.  (This is an improvement upon Ben Franklin's key escrow system: "3 can keep a secret, so long as 2 of them are dead".)

The key lengths are carefully chosen to match the expected computing power available 5-10 years in the future, but at a quite substantial price.

In the case of something as valuable as Apple source code, which might be worth $100 billion today, but only $50 million in ten years, an appropriate key length should be computable.

If in 10 years someone wants to put up $50 million to decrypt -- knock your socks off.

Such a digital lock box closely approximates Comey's/Vance's warrantable safe; the only problem is that it might take 10 years to drill into such a digital safe.

This is essentially what DoD is *already* doing with their classification codes -- trying to make sure that the encryption lasts as long as the secrets it's trying to protect.  Ditto for the entertainment industry.

There's not much difference between this scheme and "paging out" your data base onto really slow tape drives and storing the tapes in some salt mine that could take weeks to sort through (assuming that you can still find an operable tape drive to play it back).  Oh wait!  That's NASA's backup strategy!

(NASA could have a better strategy: send a digital repeater (bent tube) into interstellar space, and then start shuttling the data back and forth like the mercury delay lines of old.  Once the repeater gets a couple of light-years out, they can set up a digital "lock box".)