Video Screencast Help
Scheduled Maintenance: Symantec Connect is scheduled to be down Saturday, April 19 from 10am to 2pm Pacific Standard Time (GMT: 5pm to 9pm) for server migration and upgrades.
Please accept our apologies in advance for any inconvenience this might cause.

On Trust

Created: 04 Oct 2012 • Updated: 05 Nov 2012 • 1 comment
dfinkelstein's picture
+2 2 Votes
Login to vote

 

On Trust

 

I first came across Ken Thompson's Turing Award acceptance speech, Reflections on Trusting Trust, when I was in graduate school.  I found it very thought-provoking, and for a moment wondered if there weren't such vulnerabilities hidden away in the systems I used.  At some point, you decide to either give your trust, or withhold it.

 

I worked for 5 years at Xcert International, a startup that developed and sold Certificate Authority and related PKI products.  A PKI is rooted at a single point (the Root CA).  Root CAs are often referred to as "Trusted CAs" but I never thought of them that way, even though certificate vendors like to use the word "Trust" when describing their services.  A certificate provides identity information, certified by a signer -- "This person (the certificate subject) is associated with this public key, and the following associated attributes; so says I."  But to me, this says nothing about trust.  Sure, I can believe that I have been provided the certificate for some person or company, but that doesn't mean I should trust them.

 

With regards to trust, PGP has had a varied history.  The original versions as released by Phil Zimmermann were completely trusted.  Source code was fully available, and there was a community of security-minded people who worked on and audited the code who vouched for its correctness.  But then Phil started a company to sell Pretty Good Privacy commercially, and for some people that was the end of their trust.  Why?  The same people where there as before, and they still cared just as much about privacy and security.  Perhaps it was the addition of the "Additional Decryption Key" (ADK) functionality, which sounded too much like a way for others to decrypt your data.  Perhaps it was simply  that the software was supplied as pre-packaged binaries, and some users did not trust that what they were receiving was built directly from the sources that were published.

 

The Network Associates acquisition of Pretty Good Privacy is referred to as "The Dark Days" by many who were there at the time.  This was another time when many people lost their trust.  That NAI ceased publishing the source code for PGP certainly didn't help engender confidence.

 

When PGP Corporation emerged to take control of the PGP technology and assets, a lot of trust returned.  PGP Corporation re-instituted the practice of publishing the PGP source code for peer review.  Phil Zimmermann, Jon Callas, Will Price, Marc Briceno, Hal Finney, and other people associated with the "old guard" were back, working to make PGP better, and enhance the ways it could be used to protect your privacy and data.  Phil Dunkelberger, the CEO of PGP Corporation, posted an open letter on the PGP website assuring customers of the integrity of our products.  Of course, rumors and accusations of a "government backdoor" continued to be promulgated on the internet.

 

Trust, of course, can work both ways.  The source code that was published by PGP Corporation was never quite exactly what was used to build the PGP products.  Why?  Primarily, it was felt that certain sections of the code (say, for example, some low-level detail of how we optimized some aspect of disk encryption) was valuable intellectual property, and if published, we could not trust that it wouldn't be usurped by our competitors.  This was not a baseless fear; in the past we had seen sections of our code, especially performance techniques we employed, make their way into other products.  Our code is published for the purpose of cryptographic peer review, not as "open source".

 

Symantec has owned PGP Corporation for over two years now.  There has been no pressure to change our focus on security, no pressure to add "backdoors", and complete support for our desire to publish our (slightly redacted) source code.  You can download an older version here:

https://www-secure.symantec.com/connect/downloads/...

It's from 2011, and expect to see updated source code published for our next release.

 

Do you trust Symantec's encryption products?  Millions of people do.  I hope you are one of them.  And if you aren't, I hope we can come to earn your trust.

 
Blog Entry Filed Under:

Comments 1 CommentJump to latest comment

OKta's picture

When my project started, i was trying to find PGP realisations to use. And I find out that no one but Symantec provides libraries and application that can be easily used or included in developing applications. Good job!! Waiting for new release!!!

+1
Login to vote