[RC5] First impressions

Matt Chatterley matt at mpc.dyn.ml.org
Tue Mar 3 08:43:34 EST 1998

On Mon, 2 Mar 1998, Paul Ross wrote:

> General moan at top of email, more specific questions at end :)

Hopefully this will remain on topic (my first post, too!)
> Forgive me for sounding naive, but I am :) I stumbled across the actual 
> workings of d.net a couple of days ago, being familiar with the concept for 
> some time, and got quite excited about actually being able to be part of this. 
>  Having three P133s, a P266, a P300 and very shortly two more P333's available 
> - each with about 80% idle processor time - I felt I could contribute a fair 
> whack to the project. So hi ho, downloaded, installed and let's see what 
> happens.

I've been running RC5/DES on the two home business machines here for some
time (currently both p133s, one soon to upgrade to dual K6-233), and have
chewed through a fair few blocks. It's troublesome porting in/out buffers
to the one which has no 'net connection, admittedly. However, its fun, and
I have an active interest in cryptography, not to mention the other
beneficial statements and arguments that d.net presents very pursuasively,
just by existing.
> A couple of times, the list has mentioned attracting new people to the project. 
> How this 'sexy' project would attract more participants, therefore even the 
> non-sexy projects would benefit in the long run, etc, etc. Has anyone thought 
> about scaring people off, though?

This seems a fairly real danger, now its mentioned. I hadn't heard of
Golomb rulers before the mention here (although, being a maths/physics
student, I promptly looked up and absorbed all available information on
them, then sequentially forgot it again - calculus probably chased it
away). Whatever d.net starts doing, by spreading out into new areas, as
long as it is done carefully, we can attract more people by becoming more
visible, though - collecting lots of specialised groups doing their own
projects, who will be willing to contribute to DES-III and so forth in
return (we should really aim to *slaughter* DES-III, now we've proven the
theory on DES-II).
> Someone (I deleted the mail, I forget who - sorry) has already brought up 
> something I'd noticed straight away - how even though the DES contest was over, 
> their machine was still downloading and cracking DES blocks. A tad worried that 
> my servers were plugging away for nothing, I tried the stats pages, to see if 
> I'd made a dent - result? Stats pages offline.

The stats pages always tend to be a bit behind, something I accept because
it seems reasonable! I can imagine it being quite a lot of work to keep
those pages upto date and running (particularly when it requires analysis
of logs). The stats are far more reliable now than they used to be. On the
note of folks still cracking DES blocks, I think there were bound to be
hiccups with the changeover (it was still very annoying though - I had to
fetch 'end' DES blocks and repeatedly feed them to the offline gui client
until it realised DES was over), but next time.. we can aim to wipe out
those hiccups. With luck.
> vvv relevant bits vvv
> Now I realise that I may have joined this project at an unsteady time, in that 
> the DES key has 'only just' been cracked, and the switch back to the original 
> project is still ongoing. Is this the case? If so, how about a couple of more 
> user friendly pages on the web site, explaining such? I'm sure there's at least 
> a few people who've looked over the site and thought 'naaaah. Too much hassle.' 
> when trying to make sense of the information held within.

Possibly. I think the changeover will take a little while more as people
update offline clients and such, but we should already be going like the
clappers at RC5 compared to before DES-II; the increase in power during
DES-II was quite significant. :)
> Then, of course, that brings up another question, maybe one more suited to 
> debate than a general moan over the quality of the website ;) Does d.net want 
> to recruit people who are likely to delete the software halfway through a 
> keyblock and never return the results, leaving that block to wait until the 
> re-issue at the end?

Summary from a recent (and far too lengthy) debate: Any block is as likely
to contain the winning key as any other block. It doesn't matter what
people do with buffers, as long as they aren't returning duped blocks.
> Another question, based on issuing blocks - how is it allocated...start at the 
> beginning and continue until the end, then stop? I know the correct key is 
> chosen at random, but averages are it'll be more towards the middle than either 
> end...you can guess where I'm heading.

Start at the beginning and work our way through, slowly but steadily - in
a way, this is about brute force being used to break a lock, rather than
sneakily picking it. Statistically the odds of any given block containing
the winning key are the same (a rectangular distribution, or somesuch).
> Discussion invited, flames to be directed to my personal address :)
> I know I'm a lengthy soab, but it's just my way of saying 'hi' ;)

Hi, and welcome aboard!

	-Matt Chatterley
Spod: http://user.super.net.uk/~neddy/spod/spod.html
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