307 digit number factored
leichter_jerrold at emc.com
Fri Oct 12 11:27:43 EDT 2007
| > AFAIK, the only advantage of ECC is that the keys are
| > shorter. The disadvantage is that it isn't as well
| > studied.
| On past performance, elliptic curves are safer than
| integers. From time to time, integer based asymmetric
| encryption is abruptly and surprisingly weakened by
| advances in discrete log algorithms. This is just not
| happening with elliptic curves.
"Past performance does not predict future results."
I don't think this is a particularly strong argument. A
reasonable counter-argument is that we've been doing
number theory over the integers for hundreds of years,
while intensive work on computations over elliptic curves
goes back, what, 20 years at most?
Ultimately, we have no scientific basis for judging the
relative vulnerability of factoring, RSA, discrete
logarithms over integers, discrete logarithms over
elliptic curves, and so on to "breakthrough attacks".
(We can pretty well quantify their vulnerability to
*known* attacks as technology changes, and we have some
rough ideas about "evolutionary" attacks which don't
change the fundamental constraints but give you an
extra factor of 10, say.)
| The cost of computing power is going down faster than
| the cost of communication.
I don't see that at all. There are multiple domains to
consider. In terms of local communication, Ethernet is
continuing to deliver factors of 10 speedup every couple
of years. 10Gb/second Ethernet cards are approaching
the $100 mark. That's an astoundingly fast interconnect.
If you're talking about long-distance communication,
about the only use case I know of where *speed*, as
opposed to *latency*, is a big deal these days is in
moving really big datasets. It's hard to come up with
useful *computations* for which even something like DSL
isn't fast enough that latency completely dominates the
cost to the algorithm of communication.
In any case, how often do you send *keying material*?
The difference between 100 bytes and even 10K bytes is
lost in the noise for that rare operation on any
Meanwhile, single-stream CPU speed has more or less
stalled. Everyone is going for parallelism. Thats's
fine for some problems, not helpful for others. So in
fact it's the lack of progress in CPU speed that makes
for a stronger argument for the smaller operations of
EC than slower communications speeds!
| The size of sufficiently
| safe asymmetric encryption based on integers is growing
| considerably faster than the size of sufficiently safe
| asymmetric encryption based on elliptic curves.
| Thus the advantage of elliptic curve encryption
| continually increases, will become overwhelming in the
| near future - and a large part of that continually
| increasing advantage comes from unpredictable
| improvements in factoring and discrete log over the
| My intuition is that because elliptic curves are
| considerably less orderly than the integers, there is
| less scope for discovering fast discrete log methods. We
| are continually discovering improvements to finding
| discrete logs over the integers. It has been a long
| time since any such has been discovered for elliptic
| curves, long enough to give a plausible hope that no
| further such will ever be discovered.
I don't see that there's anything you can really back
that statement up with. Mathematics takes time. It also
tends to move in fits and starts, because it has all kinds
of unanticipated interconnections. An advance in one field
may be found to have completely unexpected applicability to
another. The proof of Fermat's Last Theorem - based on a
large number of deep results in many different fields, few
if any what you would think of as "number theory over the
integers" in any obvious sense - came together to enable
progress after hundreds of years. This is hardly an
isolated example - it's just a very widely known one.
I quite agree the elliptic curve techniques let you get
by with much smaller parameters, which has any number of
advantages. We have no particular reason to believe that
these techniques are *weaker* than techniques over the
integers, but I don't see that we really have any evidence
that they are *stronger* either. As an engineering choice,
sure, go with the cheaper computation - there's no known
loss in doing so. But don't try to convince yourself that
you've bought extra inherent security that way, because you
really don't know.
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com
More information about the cryptography