[Cryptography] It's GnuTLS's turn: "Critical new bug in crypto library leaves Linux, apps open to drive-by attacks"

[Jerry Leichter]

On Jun 3, 2014, at 4:29 PM, John Ioannidis <ji at tla.org> wrote:
> But that's all just whistling into a hurricane.  The economics say "use the free code, ship first and worry about security later" - the long, all-too-familiar list of reasons not to do the right thing.          
> 
> Show me *one* company that does the right thing, security-wise, and is profitable.
I could shorten that to "Show me *one* company that does the right thing".  :-(

But seriously, I don't really disagree.  It's all trade-offs, but some of the cost estimates used are likely wrong - and others are changing.  Facebook went along for years, successfully assuming that no one cared about privacy.  "Security" hasn't emerged in quite the same way - partly because it's less obvious where problems might emerge - but the Snowden leaks appear to have created a broad sense of concern.

> ...[S]o long as users are susceptible to social engineering (the world's oldest profession),  there will always be far easier ways for criminals to victimize users than to break the crypto. So why bother fixing the crypto? It's not the most pressing problem.
The problem with this approach is path effects.  Protocols and widely-used implementations live for a *very* long time.  If you accept crap protocol and implementations while waiting to solve the (very hard) problems of social engineering, if you ever do make progress on those, you then find yourself in the impossible position of replacing decades of installed base.  You need only look at the disaster that is network-based industrial control systems with no access controls being exposed to the Internet - and the immense costs (which no one is in a position to pay) to fix the resulting holes.

It's tough to know how to make the tradeoffs.

                                                        -- Jerry

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4813 bytes
Desc: not available
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20140603/765ddd6b/attachment.bin>