[RC5] Block order?

Denie Andriessen D.Andriessen at pink.nl
Sat Dec 26 18:45:10 EST 1998


First.
Merry Christmass (or what's left of it)

Starting at a different place from EFF could favour our chances.
About the finishing the whole keyspace.. Not that I do not believe that
the system is not capable of doing things like that. I have the feeling
it's a very good and stable, and well contempled piece of software.
Great job guys ! (and girls ?)

But; there's some things that could be of interest in finishing a keyspace
during this short, and so cheap (?) competition :
As there are a large number of people not uploading finished blocks every
time they finish one, there's a possibility to get some nice data out of the
whole 'system' for example:

What's the average/shortest/largest (or ofcourse the whole distribution) of
the
  time a block is 'away' from the keyserver ?
What is the distribution of the number of times a block is beeing recycled ?
  do some blocks get recycled more than one ? How many of them ?
Can you get an 'efficiency' or 'accuracy' from an email (I think the system
knows
  how many blocks it has send out, does it also know to whom ? How long does
  it take before the blocks get returned from that email ? Number of blocks
that
  get lost from that email? Number of block beeing returned by two machines
?
  How are they put in the stats ? Do two people 'get' the block ? The first
one that
  retrieved the block, or the first one that calculated the block, or the
first one that
  returned the block ? If the keyserver starts recycling block, how many
machines
  are supposed to work on the block simulatiously ? How much 'loss' is this
?

I think from this data some nice information could be gained:
Optimum number of block to download ? Optimum blocksize ? Where is it 'best'
to send recyled block to ? A distribution sceme for the recycled blocks ?
(If you know
client only connects twice a week, why send him blocks anyway ? (especially
usefull
is running more projects at a time) Not to lose time on the other projects ?
Can you
specify after some time for each client for which competition it is most
usefull to run ?

As you see from some of the (undoubtably hundereds) of question that
can/could
be answered by finishing a whole competition, you could gain a lot of
information
about the whole network, and maybe it would even give information about how
to be able to tune your tactics in a future competition...

Suppose: We start at some other place in the keyspace (preferable all blocks
random from the keyspace) to have some less disadvantage from EFF. What is
the
advantage in a 24Hrs competition if 50% of the block is usually more than 12
Hrs
away from the keyserver ?

Well, that's a lot of questions, and I think I could go on for a while...

Of course I don't know what information is gathered at the keyservers, etc,
but
if you give it a tought, I think everyone can see there is some potential in
finishing
a whole keyspace...

Well, only if the keyservers store some of the information that could be
gathered
If this is not the case, I would propose to increase the amount of
information that
can be gathered from them, and when that is so, to finish a whole keyspace..
Maybe even without a competition, because it would/could give a LOT of
information
about the whole 'system'!

Right, now some some Gluh-wine.
Cheers,

Denie Andriessen

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list