---------- From: Ray Dillinger[SMTP:bear@sonic.net] Sent: Wednesday, September 27, 2000 4:39 PM To: Trei, Peter Cc: cypherpunks@cyberpass.net Subject: RE: CDR: Re: Lions and Tigers and Backdoors, oh, my...
On Wed, 27 Sep 2000, Trei, Peter wrote:
Can you document this claim of the existance of 'help fields' in Netscape?
Not directly I can't, at least not without betraying someone. In retrospect, I should've used a nym to make the statement to keep him out of trouble.
I don't accept this. You should be able to: generate traffic dumps pointing to the 'help field', and showing where it fits within the SSL specifications. This is hardly rocket science. There is no need to compromise anyone whatsoever. Put forward the evidence, so we can independently confirm it. This sort of thing happens every year or two on this list. Someone makes a claim which, if true, has interesting and/or important implications. However, the nature of the claim is one which is quite capable of verification. The onus is placed on the claiment to 'put up or shut up'. Usually, they shut up. Examples of such claims include: * PGP has a secret backdoor. (OK: Here's the source: Where is it?) * gcc is hacked to stick secret backdoors into PGP. (OK: Here's the source: Where is it?) * All gcc binaries will stick the PGP hacking code into gcc when compiling gcc. (Here's a dump: show us.) * Emily Dickinson hid her boyfriends initials in her poems. (Here's some statistical tests you need to run: Show us). So I call upon you: Put up or Shut up.
I am (to put it mildly) astonished by this claim, and more than a little skeptical. I was aware of the Workfactor Reduction field in the export 'aka International' version of Lotus Notes (which this 'help field' seems identical to), but was not aware of it being included in any other application.
Okay, let's forget what I know from people I don't want to drag into the fire and go through it from the "circumstantial" angle.
What does it mean when Lotus Notes has to put a work reduction field in their product in order to get export approval status, and then doesn't talk about it? But lots of other companies who also don't talk about it, with stronger-seeming crypto get export approval status?
Huh? "Doesn't talk about it"? It was announced with fanfare at the RSA Data Security conference a few years ago. There were press releases. It was widely discussed on this list. I invented the term 'espionage enabled' to describe this kind of application. Lotus got a lot of flack about it, but persisted. Some customers even bought it, noteably the Swedish government (See Risks Digest http://catless.ncl.ac.uk/Risks/19.52.html#subj1).
What does it mean when banks refuse to work with earlier versions of Netscape claiming it's because the security certs are expired -- but when new security certs are downloaded and installed, they still refuse to work with earlier versions of netscape and refuse to tell you why? (This, btw, was what made me suspicious in the first place and why I started digging...) http://banking.wellsfargo.com/
Well, it could mean that they want to use Web features available in later versions but not in earlier ones. Or maybe there are known security holes in earlier editions. Wells Fargo has actually been ahead of the curve at times: they were one of the first sites to require 128 bit encryption.
What does it mean when Lew Giles, even after the rules change to the BXA-controlled system, made a living going around convincing engineers working for american companies to compromise their products' security? With or without knowledge of the companies' execs? http://www.counterpane.com/crypto-gram-9902.html#backdoors
Is it after BXA? Bruce notes that the stories are all at least two years old, which would place them in very early 1997 at latest. I can't remember when the switch took place. Of course, if LG was doing this (and I have no real doubt that the NSA might try), it's excreable, but what is the relevance to products today? He can no longer seriously threaten to hold up export.
What does it mean when PGP has a "flaw" introduced into its Additional Decryption Keys at the same time NAI is seeking export approval for it? And NAI gets export approval, and then nobody notices the flaw for several years after, and then they go oops, it was just a mistake?
What does it mean when a CEO who actually can and does review code, so subverted engineers can't seem to get one past him, in a meeting with NSA officials refuses to compromise -- and one of the spooks loses his cool and offers to run the guy over in the parking lot? I'll explain this one to you... it means that spook _HAD_NEVER_SEEN_ anyone refuse to compromise, and had no fucking clue what to do. That's if you buy the "he just lost his cool" story. On the other hand, death threats may be policy and this was just the first time they were needed. And on the gripping hand, maybe it's just the first time it was *reported*. Not very many execs would talk about something like that, and I figure most who've experienced it probably just shut up and gave the spooks whatever they wanted.
Lew Giles and its ilk had to have some kind of bargaining position, and if export approval was forthcoming without subverting security in some way, would have had none. The only way a spook could lose his cool and offer Bidzos a death threat would be if that spook were totally unfamiliar with people not compromising.
Actually, I know of at least one other case where a major exec at a crypto company threw NSA people out on their ears after they offered to write some software to help with a project. I think you're confusing two different types of subversion. 1. Company policy based subversion. This is the type where word comes down from above to put in 'helper fields', etc. I can't see this happening in any company which was not simply a LEA front (such as Crypto-AG). The liability would be simply too great, the exposure too likely. This would require a large number employees and former employees to know, yet keep their mouths shut. When exposed, the officers of the company would be subject to criminal fraud and conspiracy charges, as well a variety of civil suits. It's simply a non-starter. Lew Giles can make as many pleas as he wishes; but I don't think he'll get many results. 2. Individual treachery. This type involves corrupting one or more engineers, whether via money, threats, or misplaced appeals to patriotism. This is more likely to succeed in the short term than type 1, but is very fragile for several reasons familiar to anyone who has done commercial software development. * Peer code reviews mean that many eyes look at the code. * Employee turnover in the field is high - 30-50% year. Thus, bugs inserted by earlier compromised employees are unlikely to last through many release cycles, as new employees come in and say 'Oops - Joe forgot to init the PRNG properly - lets fix that!' * Source code management systems make it very difficult to a single actor to monkey with code secretly, and even harder to cover his tracks.
You may consider me paranoid, but I'm telling you that the case of Lotus Notes was just the one that people found out about. If Lotus had to do that to get export approval from the BXA, then so did everybody else. I do not buy the story that what happened to PGP was an accident; on the contrary, it was just NAI doing what they had to do to get approval to put it up for international downloads, the same as Lotus just did what it had to do. And, I'm telling you now, the same as AOL and Microsoft did what they had to do with the browsers.
Ray
Actually, it's pretty clear what happened with most of those cases where companies were allowed to ship strong crypto overseas before most of the restrictions were lifted. They all involved products where at least one party of the communications was a large organization or company which was subject to, and could be counted on complying with, sub poenas and court orders. Thus the authorities had another route by which to acquire content in which they were interested. I have little doubt that government organizations have attempted to subvert commercial crypto at various times in various ways. However, the 'help fields' you describe are simply implausible in open protocols such as SSL. Again, I ask you to point to the evidence. I and others have already run the searches you suggested and come up dry. Where are the URLs? As it stands, you are spreading FUD. Peter Trei