Cryptome does good stuff, by posting interesting material from the Federal Register and from whistleblowers. It also shows its crypto-anarchy roots – which are often childish in nature. When it leaves the topic of crypto politics and addresses the design space of ciphers and cryptanalysis, it leaves impressions that do a disservice to its wider cause. It becomes the “Fox News” of crypto, leading opinion in ways that are just boring (and mostly wrong). In short: it’s unfair-minded, while pretending otherwise. It’s a rant forum (like this is a rant). It becomes a rallying point for ranters.
http://cryptome.org/0005/ssl-broken.htm
SSL was not broken at birth. And, it did not attempt to do what IPsec attempted to do. It is not a layer 3 security protocol after all; but the result of years of previous design efforts to distinguish what a layer 4 security protocol could do (that a layer 3 service can not). That some designers with a “US orientation to their thinking” never wanted a split between layer 3 and 4 should not go unobserved, even today. Even now, they evidently continue to want a universal security layer (tied to the internet service providers). Such is the intenet however, that the battle between host-based or router-based architectures will never wane.
Yes SSL was a design of the times it lived in. What design is not? And yes, in those times the issues were about ending the cold war’s years of crypto supression so as to fashion a space of crypto enablement for public services. And, this included government services. SSL was never about the PGP brigade re-losing against the Fascists forces in the Spanish civil war, enabling the Russian refuseniks, or Arab Springs. It was and still is about e-commerce, first and formost; and the securing of such as nntp message board posts, second. (These days, replace nntp by wordpress blogs…) By focussing on business and common-or-garden privacy in open spaces, the political concept was to make it harder for the more conservative crypto-suppressors to win the war – and continue the typically-American policy of tying up such as crypto (or black folks’ rights) in harassing red tape as ameans to supress (or oppress).
What SSL was also was not was be associated with IETF; or the long running saga known as IPsec. IPsec was (and still is) an unworkable American grade mess of a design, that works fine under certain assumptions that, ahem and 20 years after this was first observed, still not present in the general internet or the web. Wonderful architecture; bad design (that is good design in military cryptonets, for which it was intended). SSL filled the gap, using the best of what layer 4 had to offer the public – not be layer 3. SSL challenged the military biases of the IETF – arguing as does the web that what works for military does not work for public services. while Dod was in charge of leading the public crypto charge through 2009, of course anyone who said this got fired (or blacklisted). Such is American freedom (great if you are rich, and irrelevant words if you are poor).
Like Fox News, I often find PGP design-grade folks mislead – lying by omission. Americans have a different name for this process: they call it “making a case”. In English, its simply lying: the intent to deceive by the omission of material facts (or using presentation means that induces failure of the informed audience to appreciate their significance). One thinks of Colin Powell and pseudo-Americans like Tony Blair as masters of this art. Of course, its easy to out-think such folks (as OBL proved, by laying successfull economic warfare traps).
Economics is what its all about, in politics – including SSL’s cryptopolitics. Yes CA-based trust networks are showing some stress fractures; but despite it all they work fine in both IPsec military cryptonets and economics-drive layer 4 SSL/internet theatres. Despite 20 years of design and ranting, noone has come up with a viable alternative. We may well find that all CAs needs to do is further deploy what was always too bothersome: revocation!
Rather than listen to old, out dated designers posturing on IPsec issues from 3 decades ago, folks should focus on what is UNDER utilized in SSL: namely the varieties of ciphersuites with distinct crypto properties, and the ability to tunnel. With the emergence of javascript-based SSL clients in browsers, the full panoply of SSL capabilities can now be utilized, since no longer will 5 (comomodity-influenced) browser vendors be determining which slice of SSL is given to consumers. Server page designers will be able to “make the better case” to consumers; and they each may decide differently to the CAB forum (and its US sponsor)
SSL in javascript is threat to the established order. in this threat, it outclasses PGP by 10 orders of magnitude; with a lot of (today’s relative liberal) crypto-conservatives running very scared by the implications. They should be even more scared about what happens when SSL and SSO get together, with SSO’s signatures enabling (trustworthy) javascript/JSON code distribution. With this combination, the very infrastructure that the US believes will enable it to win a cyberwar through economic, information and signals dominance will always be used to undermine the foundations of that policy, exploiting a famous dictum of warcraft – the dilemma.
Peter's ruminations 