27 expanded posts ยท every known parent and child
The only change the future post-September PGP 2.6 messages will have is a change in the version number byte from 2 to 3. PC's little hack not to check version numbers will work, but as a patch it's not the most robust. It would be more robust if it checked for the range [2..3]. Another thing a patched 2.3 release would have to do to be fully indistinguishable is to generate new version numbers itself after the given date. Eric
You wrote: | Another thing a patched 2.3 release would have to do to be fully | indistinguishable is to generate new version numbers itself after the | given date. While I understand that people prefer the 2.3 code because of its availablility outside of the US, and speed advantages, I think that its important to remember PGP has not really caught on in the US because of questions about its legality. I'm trying to push for the widespread use of PGP 2.5 here at the Brigham & Women's hospital where I work. I can't push for version 2.3 for legal reasons. There is a significant advantage to pacthing both the US-legal and world legal versions of PGP. Both are useful & neccessary to the future of strong encryption. I've heard that 2.5 is available outside of the US. If this is so, would it make sense to make 2.5 the version which is patched and enhamced as the standard? Adam -- Adam Shostack adam@bwh.harvard.edu Politics. From the greek "poly," meaning many, and ticks, a small, annoying bloodsucker.
Adam Shostack says: > You wrote: > | Another thing a patched 2.3 release would have to do to be fully > | indistinguishable is to generate new version numbers itself after the > | given date. > > While I understand that people prefer the 2.3 code because of > its availablility outside of the US, and speed advantages, I think > that its important to remember PGP has not really caught on in the US > because of questions about its legality. I'm trying to push for the > widespread use of PGP 2.5 here at the Brigham & Women's hospital where > I work. I can't push for version 2.3 for legal reasons. People overseas want to be able to use this program, too. There are 250 million people in the U.S., which constitutes under 1/20th of the Earth's population. Quit being provincial. This discussion is about what the other 4.75 billion people have to do to interoperate with the brain-damaged MIT stuff. Perry
Date: Tue, 24 May 1994 13:06:22 -0400
From: "Perry E. Metzger" <perry@imsi.com>
There are 250 million people in the U.S., which constitutes under
1/20th of the Earth's population.
These statistics are somewhat misleading given that the vast majority
of users that are on the net are in the U. S. I suspect that the same
is true for computer users in general, but I'm much less certain.
I agree that this legal silliness is unfortunate, but I don't think
that it's especially terrible that Adam would like to be able to
advocate PGP use at work without putting himself at risk.
I think it's great that patches are coming out to bridge the gaps
between 2.3 and 2.6 from either direction. It's also nice that the
newer versions of PGP appear to have a propensity toward travel :-)
RickRick Busdiecker says: > There are 250 million people in the U.S., which constitutes under > 1/20th of the Earth's population. > > These statistics are somewhat misleading given that the vast majority > of users that are on the net are in the U. S. I suspect that the same > is true for computer users in general, but I'm much less certain. > > I agree that this legal silliness is unfortunate, but I don't think > that it's especially terrible that Adam would like to be able to > advocate PGP use at work without putting himself at risk. You've misunderstood. The point is only that overseas users, technically speaking, do not have access to 2.[56], and might want patches. I didn't say anything about whether Adam should be running 2.[56] on his machine. Perry
> Rick Busdiecker says: > > There are 250 million people in the U.S., which constitutes under > > 1/20th of the Earth's population. > > > > These statistics are somewhat misleading given that the vast majority > > of users that are on the net are in the U. S. I suspect that the same > > is true for computer users in general, but I'm much less certain. > > > > I agree that this legal silliness is unfortunate, but I don't think > > that it's especially terrible that Adam would like to be able to > > advocate PGP use at work without putting himself at risk. > > You've misunderstood. The point is only that overseas users, > technically speaking, do not have access to 2.[56], and might want > patches. I didn't say anything about whether Adam should be running > 2.[56] on his machine. That is a snotty answer to avoid answering the question, Perry. Non-US/ Canadian users weren't supposed to have access to PGP in the first place, so what's the problem? If they want it, they can probably get it from the place where they got PGP 2.X in the first place. -- Ed Carp, N7EKG/VE3 ecarp@netcom.com 519/824-3307 Finger ecarp@netcom.com for PGP 2.3a public key an88744@anon.penet.fi If you want magic, let go of your armor. Magic is so much stronger than steel! -- Richard Bach, "The Bridge Across Forever"
Ed Carp says: > > You've misunderstood. The point is only that overseas users, > > technically speaking, do not have access to 2.[56], and might want > > patches. I didn't say anything about whether Adam should be running > > 2.[56] on his machine. > > That is a snotty answer to avoid answering the question, Perry. What question precisely is it that I'm not answering? I was unaware that any question had even been asked. Adam said that he didn't think patches were useful. I simply noted that there are lots of people outside the U.S. who might want them. Hell, there are lots of people inside the U.S. who might want them. No one was implying by the distribution of such patches that Adam should be running any software on his computer he doesn't feel comfortable with. Perry
From: Adam Shostack <adam@bwh.harvard.edu>
Date: Tue, 24 May 94 12:55:36 EDT
There is a significant advantage to pacthing both the US-legal
and world legal versions of PGP. Both are useful & neccessary to the
future of strong encryption.
Date: Tue, 24 May 1994 13:58:34 -0400
From: "Perry E. Metzger" <perry@imsi.com>
Adam said that he didn't think patches were useful.
Hmmmmm....
RickPerry: | > I agree that this legal silliness is unfortunate, but I don't think | > that it's especially terrible that Adam would like to be able to | > advocate PGP use at work without putting himself at risk. | | You've misunderstood. The point is only that overseas users, | technically speaking, do not have access to 2.[56], and might want | patches. I didn't say anything about whether Adam should be running | 2.[56] on his machine. Technically, they never had access to v1, either. As I said in my first message, I've heard 2.5 has already found its way out of the US. If that is the case, then the non-US users have access to 2.5. If they do have access to 2.5, then could we discuss the technical merits of patching 2.5 v. patching 2.3? Benefits of starting with 2.3: * widespread use * no RSA code * faster? 2.5: * clearly legal in the USA * single code base for future modifications Its my opinion that the single code base, developed outside of the US, based on 2.5, is the way to go. Patching 2.3 is worthwhile, but does not address all (potential) users of PGP. Patching 2.5 does (again, assuming that its been exported), and as such, I feel it is a better way to go. Adam -- Adam Shostack adam@bwh.harvard.edu Politics. From the greek "poly," meaning many, and ticks, a small, annoying bloodsucker.
Adam Shostack says: > Technically, they never had access to v1, either. As I said > in my first message, I've heard 2.5 has already found its way out of > the US. If that is the case, then the non-US users have access to > 2.5. I wouldn't know where to find 2.5 outside the U.S. Besides, there are other scenarios in which one would want such patches. Here are just a couple. 1) You have a friend with an old PGP who wants to send you mail and who can't get a new PGP. Old PGP will read old PGP generated files, but new will not read old. 2) You have a friend eight months from now who only has old PGP and who you would like to send new PGP to. He knows your old-form signature but can't read the new one. The patch is simple enough that he can verify it himself. You can send it to him and then send him a signed copy of the new PGP. In any case, I see no reason to oppose people posting patches. This is the last time I'll post on this topic. Its getting old fast. Perry
> 1) You have a friend with an old PGP who wants to send you mail and > who can't get a new PGP. Old PGP will read old PGP generated files, > but new will not read old. Wrong, Perry. Go read the announcement again. 2.6 will read old messages, but after sept 1 it will start generating incompatible messages that old versions cannot read. -derek
Derek Atkins says: > > 1) You have a friend with an old PGP who wants to send you mail and > > who can't get a new PGP. Old PGP will read old PGP generated files, > > but new will not read old. > > Wrong, Perry. Go read the announcement again. 2.6 will read old > messages, but after sept 1 it will start generating incompatible > messages that old versions cannot read. Ahem. This is only true if the pkccompat mode was on. I have plenty of text around that 2.5 will not read. This is not tragic, but the situation might arise. In any case, I don't understand why anyone would rationally oppose the distribution of Pr0duct Cypher's patches -- you don't have to use them if you don't like. Perry
Perry writes: | In any case, I don't understand why anyone would rationally oppose the | distribution of Pr0duct Cypher's patches -- you don't have to use them | if you don't like. I wasn't opposing them; I was suggesting that patching 2.5 would be more productive in the long run than patching 2.3 Adam -- Adam Shostack adam@bwh.harvard.edu Politics. From the greek "poly," meaning many, and ticks, a small, annoying bloodsucker.
Adam Shostack says: > Perry writes: > | In any case, I don't understand why anyone would rationally oppose the > | distribution of Pr0duct Cypher's patches -- you don't have to use them > | if you don't like. > > I wasn't opposing them; I was suggesting that patching 2.5 > would be more productive in the long run than patching 2.3 Again, as I've noted, there are people who will need, for whatever reason, to fix their old 2.3a (or pre-2.3a) system so that it will interoperate. Assuming that 2.5 finds its way overseas, it is not an unreasonable code base for FUTURE development. However, what we are talking about is not new development but retrofits. Perry
I've been out of the literature for quite a while now so pardon me if this is a dumb question. Do any of you know of any public key cryptosystems based on the graph isomorphism problem? Last I heard there weren't any. But I think I've found one. j'
Jay Prime Positive says: > I've been out of the literature for quite a while now so pardon me > if this is a dumb question. Do any of you know of any public key > cryptosystems based on the graph isomorphism problem? Last I heard > there weren't any. But I think I've found one. There was a powerful result a while back concerning public key systems based on NP complete problems -- in particular, I recall that there was a large class of them that were flawed -- the original knapsack problem based public key system suffered from the defect from the limited amount my neurons will disgorge. Sadly, I can't remember the details any longer. Anyone else have a vague recollection on this? It would be cool to hear about your graph isomorphism based system in any case. I have heard of zero knowledge systems based on graph isomorphism, but never public key systems. By the way, there is a neat paper circulating in samizdat form from China about public key systems based on compositions of finite automata. However, I'm more or less obligated not to spread it about until the paper has been published (sigh). Its quite tantalizing, though. Perry
Date: Tue, 24 May 1994 18:11:51 -0400 From: "Perry E. Metzger" <perry@imsi.com> There was a powerful result a while back concerning public key systems based on NP complete problems Hmm. Was it the set of 'super increasing' knapsack problems? It would be cool to hear about your graph isomorphism based system in any case. I only worry that if I publish, it could be patented. And I don't want the algorithm to end up in the hands of the software patent folks. Especially if they will be making money off it, and I wont. Solutions? Perry
Jay Prime Positive says: > I only worry that if I publish, it could be patented. And I don't > want the algorithm to end up in the hands of the software patent > folks. Especially if they will be making money off it, and I wont. If you publish, only you could patent it. There is only danger if you don't publish, in which case others can independently make the same discovery and patent it. Perry
>
>
> Jay Prime Positive says:
> > I only worry that if I publish, it could be patented. And I don't
> > want the algorithm to end up in the hands of the software patent
> > folks. Especially if they will be making money off it, and I wont.
>
> If you publish, only you could patent it. There is only danger if you
> don't publish, in which case others can independently make the same
> discovery and patent it.
>
> Perry
But you would then need to file within one year of the publication
date I think.
BartI only worry that if I publish, it could be patented. And I don't want the algorithm to end up in the hands of the software patent folks. Especially if they will be making money off it, and I wont. If you publish, only you can patent. One must be the 'true inventor' (or some similar term of art) in order to file a patent on an invention. As someone pointed out, a system can be re-invented; then that person is also a true inventor and can patent. Publication is protection against patenting. This is one of the main reasons behind such publications as the IBM Technical Journal--the publication of results not worth patenting themselves, but definitely worth preventing others from patenting. Publication of a result precludes this. Eric
> From: jpp@jpplap.markv.com (Jay Prime Positive) > cryptosystems based on the graph isomorphism problem? Last I heard > there weren't any. But I think I've found one. Interesting. Have you tested it against the known methods for the isomorphism problem? Van Leeuwen* references an O(n log n) average-case algorithm, and ones that are pseudopolynomial w.r.t. degree, genus, and treewidth. There are also methods based on "signatures" (hash functions on graphs, basically); there's an O(n^2) expected-time perfect signature, and an O(n) (worst-case?) one with exponentially small failure rate. These might provide attacks, though none solve the general problem. * (in Handbook of Theo. Comp. Sci., Vol. A) BTW, the graph isomorphism problem is not known to be NP-complete, and van Leeuwen comments that there is some theoretical basis for expecting it not to be. Disclaimer: I don't know much about graph theory, I'm just getting paid to do it. :-> Eli ebrandt@hmc.edu
> Date: Tue, 24 May 94 17:08:05 PDT > From: Eli Brandt <ebrandt@jarthur.cs.hmc.edu> > > Interesting. Have you tested it against the known methods for the > isomorphism problem? Van Leeuwen* references an O(n log n) > average-case algorithm, and ones that are pseudopolynomial w.r.t. > degree, genus, and treewidth. There are also methods based on > "signatures" (hash functions on graphs, basically); there's an O(n^2) > expected-time perfect signature, and an O(n) (worst-case?) one with > exponentially small failure rate. These might provide attacks, > though none solve the general problem. > * (in Handbook of Theo. Comp. Sci., Vol. A) No I haven't tested it against any known GI algorithm. Your references are all very interesting and I will investigate them. If you had a publisher handy, along with the city the publisher is in, I would happily phone them up and get a copy. But if not, I can operate a card catalog. > BTW, the graph isomorphism problem is not known to be NP-complete, > and van Leeuwen comments that there is some theoretical basis > for expecting it not to be. No, I didn't expect GI to be NP-complete at all. I expect rather that P < GI < NP. That is one of the reasons that GI is an interesting problem. Especialy because (as you point out) GI is amost always in P. In any case, my PK cryptosystem is not interesting except for the new complexity point. (Although, the general construction may be interesting.) I can prove that my cryptosystem has a level of security which is reduceable to GI, and GI to it. (The reduction is only in polynomial time. I will try to see about getting the slow parts down to O(n) time.) PGP will almost certainly never include my PK system as an alternative to RSA. For one thing it needs a k^3 to 1 expantion in communication costs for a security parameter of k. For another the 'fast' decrypt routine requires O(n^3) in the number of nodes in the graphs. But there is no known GI algorithm which is O(n^3) in general. (And if there is one for *my* graphs, then I will give you a polynomial time algorithm for all of GI.) > Eli ebrandt@hmc.edu
> > Technically, they never had access to v1, either. As I said > in my first message, I've heard 2.5 has already found its way out of > the US. If that is the case, then the non-US users have access to > 2.5. If they do have access to 2.5, then could we discuss the > technical merits of patching 2.5 v. patching 2.3? > > Benefits of starting with 2.3: > > * widespread use > * no RSA code > * faster? > > > 2.5: > > * clearly legal in the USA > * single code base for future modifications > > Its my opinion that the single code base, developed outside of > the US, based on 2.5, is the way to go. Patching 2.3 is worthwhile, > but does not address all (potential) users of PGP. Patching 2.5 does > (again, assuming that its been exported), and as such, I feel it is > a better way to go. > Perhaps, but I think that many folks still do not have the answers they are looking for -- primarily, can you use the secring generated by 2.3 or below with the new (2.6) release, if you so desired? - paul
> From: "Perry E. Metzger" <perry@imsi.com> > People overseas want to be able to use this program, too. There are > 250 million people in the U.S., which constitutes under 1/20th of the > Earth's population. You dropped the part of his message where he said that he believed v2.5 was available abroad. If it's not, I don't think it will take long. And once it's escaped from this little prison state of ours, overseas users incur no risk in using it: they can't even be Noriega'd, since they've broken no law in any country. Eli ebrandt@hmc.edu
> The only change the future post-September PGP 2.6 messages will have > is a change in the version number byte from 2 to 3. PC's little hack > not to check version numbers will work, but as a patch it's not the > most robust. It would be more robust if it checked for the range > [2..3]. Agreed. > Another thing a patched 2.3 release would have to do to be fully > indistinguishable is to generate new version numbers itself after the > given date. Is "indistinguishability" the point or "interoperability"? If the latter, then no change to generated version numbers should be necessary/desired. I believe that 2.6 plans to read previous versions just fine. Cort.
> Another thing a patched 2.3 release would have to do to be fully > indistinguishable is to generate new version numbers itself after the > given date. Is "indistinguishability" the point or "interoperability"? Reference is not advocacy. I was speaking of what was necessary to ensure indistinguishability. If that is your goal, then this is directly relevant. If not, then it may be beside the point. The change in version numbers seems to have two effects, both of which I addressed. Use these statements as they are appropriate to your goals. Eric P.S. The "you" is the general "you".