Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[liberationtech] Wicker: Déjà vu all over again

Yosem Companys companys at stanford.edu
Tue Jun 10 12:13:50 PDT 2014


From: Brian Behlendorf <brian at behlendorf.com>

You don't have to; "trust, but verify".  Or trust those who *can* verify.
Microsoft, Google and Apple are at the top of the "most trusted brands"
lists and have been for years, so even in the light of the Snowden
revelations, most have tended to give them the benefit of the doubt and
keep using their proprietary software and services.  But those who don't,
and instead use self-hosted open source tools, are making a different trust
choice - they prefer to trust Linus Torvalds, the Linux community, Firefox
developers, Pidgin developers, Apache developers, and the broader developer
community, on a gut-level calculus that those parties are less likely to
intentionally corrupt their software, and are more likely to find
each-other's (intentional or accidental) corruptions.  That calculus
integrates across all software, teams, and time, so even disasters like
Heartbleed aren't enough to change the result for most of us.  Speaking
personally, it only reinforced it, by watching not only how quickly the
disparate communities reacted and pushed solutions out, but how much it's
caused further inspection of OpenSSL and other underlying packages.

This calculus does have some bigger blindspots, though - I was never
comfortable with promoting TrueCrypt, a package written by intentionally
anonymous authors without any of the trappings of an open source project -
open revision control, open bug tracker, open discussion boards for
development.  I like being able to attach names to code - software is made
of people, not unlike Soylent Green.  Even though it's not really truely
Open Source licensed, I trust qmail, djbdns, and other packages written by
Dan J. Bernstein because he's a no-bullshit mathematician, scientist,
coder, and fighter for liberty (see Bernstein v. United States).

With proprietary solutions, including Wickr, the "verify" window is much
more narrow.  You can inspect what it sends over the wire or stores on
disk, but even that's pretty opaque.  Without that "verify" loop, you can
trust those who they've hired to do security audits.  You can also figure
out whether you trust Nico herself.  There are those of us on the advisory
board for Wickr (full disclosure) who are working with them to figure out
some way to broaden that trust+verify window.  We'll see what happens.

Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/liberationtech/attachments/20140610/eca9fe02/attachment.html>


More information about the liberationtech mailing list