Re: How to crypt for 1,000,000 years into the future?

From: Donald Hines (Donald.C.Hines_at_mail.sprint.com)
Date: 07/09/03 

Date: 9 Jul 2003 08:49:09 -0700

Thomas Moore <TMoore@h8spalace.com> wrote
>It's actually for a journal that I keep. I have to record
>some things that I'd rather remain unknown until my death.

THIS is a much easier problem. We can break it down to 3 basic
issues.

1: Keeping the secret until you die.
2: Making certain the secret can be decrypted after your death.
3: Making certain someone bothers to actually decrypt it.

The problem with using secure encryption right now and just hoping
someone breaks it after you die is that, unless you just happen to be
a very important and interesting person, number 3 above will probably
be violated and no one will actually go to the trouble to decrypt the
journal.

If your information is likely to be so interesting that number 3 is
not an issue, then we have a difficulty with number 1 and your journal
will be read before you die. Because if you make the encryption easy
enough so that:

>I expect it to be decrypted not long after I die.
  
then it will probably be decrypted not long after someone sneaks in
and photocopies your journal. (And if you aren't worried about
physical security, why bother encrypting it?)

I think there ARE some practical solutions.

This is actually a place where key escrow would be useful. Pick three
people you trust and give each of them one third of the key, with
instructions to only reveal it after you are dead. Security would be
enhanced if the key holders do not know who the other key holders are.

OR, you could also just hire a law-firm to keep the secret papers with
instructions to only reveal them after you are dead.

Another option would be simple physical security instead of
cryptographic security. Purchase a Safety Deposit Box at a bank and
put the relevant journal entries in there. Keep the key on your
person. When you are dead the key will be found and the estate will
track down the safety deposit box.

Now all 3 of the above solutions require an element of trust in order
to ensure your security. The key holders, or the law-firm, or the
bank MIGHT betray you and reveal your secret. There is a very
complicated way to avoid all that. Put a time-lock on your journal.

Rivest, Shamir, and Wagner have a paper on time-lock crypto available
at:
http://theory.lcs.mit.edu/~rivest/RivestShamirWagner-timelock.ps
(Google will translate this to text for you)

The essential concept is that you encrypt the journal using standard
strong crypto. THEN you encrypt the key using a mathematical function
that can only be solved by dedicating a single computer to the job for
a long and predictable period of time. You just set up the time-lock
to take longer then your expected lifespan to decrypt (taking into
account probable improvements in computing technology)

The problem with the time-lock solution is that it puts an extreme
strain on our issue number 3. Is anyone actually going to be
interested enough in your journal entries to dedicate 50 years of
computer time to decrypting them?

Donald Hines

Relevant Pages

  • Re: Anything to this mumbo jumbo I found while surfing the net?
    ... > We believe there is one very simple rule in encryption - if someone can ... someone else will be able to decrypt it. ... > navigate inside the Virtual Matrix is created. ... > security at low computational cost. ...
    (sci.crypt)
  • Re: How to prevent importing tables
    ... why in-built encryption doesn't even matter if they otherwise have ... off-topic but is all to do with security and "extraction". ... > sufficient permission to open the database, ... then they don't even need to decrypt it! ...
    (microsoft.public.access.security)
  • Re: OT: Password management
    ... My task is not to deal with security issues they have already addressed, ... how do I reverse the encryption of a reversible password? ... The question deals with how to decrypt ... IME the term "reversible password encryption" doesn't refer to any one ...
    (microsoft.public.vc.mfc)
  • Re: EFS is a joke!
    ... > You seem to be missing a couple of basics of security yourself. ... And if he could do that, why use EFS at all? ... they can't decrypt the file and see your information. ... multiple layers of encryption for security. ...
    (comp.security.misc)
  • How do I Use DPAPI to Encrypt and Decrypt Data (C#/VB.NET)?
    ... Use DPAPI to Encrypt and Decrypt Data ... The code below demonstrates how to call Data Protection API (DPAPI) ... In addition to encryption and decryption, ... public static string Encrypt ...
    (microsoft.public.dotnet.framework.aspnet.security)