1995-08-26 - SSL CHALLENGE: I told you so….

Header Data

From: monty.harder@famend.com (MONTY HARDER)
To: CYPHERPUNKS@toad.com
Message Hash: 01f7da6c00d1705435952ebfa004176b60323edcfcccc9aad4560d10f80f0f1f
Message ID: <8AFD417.000300036F.uuout@famend.com>
Reply To: N/A
UTC Datetime: 1995-08-26 00:42:54 UTC
Raw Date: Fri, 25 Aug 95 17:42:54 PDT

Raw message

From: monty.harder@famend.com (MONTY HARDER)
Date: Fri, 25 Aug 95 17:42:54 PDT
To: CYPHERPUNKS@toad.com
Subject: SSL CHALLENGE: I told you so....
Message-ID: <8AFD417.000300036F.uuout@famend.com>
MIME-Version: 1.0
Content-Type: text/plain


PT> I've been looking at the allocated list of keys, and I notice a possible
PT> problem:  One user has allocated 50,000 segements:
PT>
PT> 0c2b-cf7a NOACK 0c2b 50000 Joe Thomas <jthomas@ogi.com>

                                  -o-

PT> I can't contact the server to request keyspace anymore - I get a
PT> '500 Server error'
PT>
PT> It looks like 12 hours on a P5/90 are going to waste (could have done
PT> 90 segments)

                                  -o-

RL> The bottleneck at the server is pretty awful.  My clients are spending
RL> most of their time in timeout, waiting for keyspace.


<rant>

  As I watch you fumbling for keyspace IN REAL TIME....

  This is a great time for me to say "I told you so" WRT the doling of
keyspace. It is far better to sign up your volunteers, CHECK THE
ALLOCATION, and hand out starting keys, before there is even a challenge
to brute.

  The quad-coverage scheme I outlined before allows sufficient
protection against dropouts (whether accidental or malicious), so that a
single person who asked for some keyspace, but was unable to actually
sweep it, would not hurt the effort.

  Those who fail to ACK (once it is certain that the client does in fact
issue it - should have some parallel means of acking) should have their
allocation, based on my evolving formula

        available_mips = mips * idle_rate * .5**(unacks/sweeps)

for the next challenge reduced, reflecting previous unreliability. Set
up a sign-up period, add up the total available mips, figure the share
that each person should get, and assign keyspace accordingly via direct
Imail to each participant.

       The next level of abstraction is to not only give a start..end
     and direction, as I previously outlined, but supply a scaling
     factor in the challenge release. That way, any late entries can
     automagically be inserted into the keyspace, by squeezing everyone
     according to the final total of available_mips.

  Once the proverbial clock is running, there should be no need to
discuss who will do what. That is what the huddle is for. Or, to string
along the analogy, I guess Steve Young should just go up to the line,
take the ball from center, and then give hand signals to Jerry Rice
while a 300-pound defensive lineman is zeroing in on him, because the
linemen are standing around wondering whom to block.

  Geez. It would be better not to even have a damned keyserver at all,
but to just have everybody pick a random start location, than to do it
this way.

</rant>


 * A child said "Mommy, why is the man with the crown naked?"

 * A child said "Mommy, why is the man with the crown naked?"
---
 * Monster@FAmend.Com *    





Thread