home | list info | list archive | date index | thread index

Re: [OCLUG-Tech] CACert - free digital certificates

Adrian Irving-Beer wrote:
<snip of my own drivel>
> The original (primary) idea was that the cert companies verify who you
> are, sort of like PGP.

Yeah, the diligent ones do this - I've gone through this many times with
Entrust for clients.  Thawte was pretty good (when Shuttleworth ran it),
and Verisign is Verisign.  The problem is that many not-so-diligent
organizations have bought themselves into the chain of trust that traces
back to a trusted root cert - meaning browsers will accept worthless
certs as trustworthy.  As long as a CA can establish a chain of
authority back to a trusted root cert, Bob's your uncle.

Over the last five years I've noticed a marked decrease in the quality
of proof necessary to establish identity prior to cert issuance.

IE automatically establishes Microsoft as a trusted root cert in every
system that runs Windows.  Do you think Microsoft should be trusted as a
bank?  or a government?  I know of at least on incident where Verisign
issued a Microsoft labelled cert to a clown in the states - useful for
signing malware to install in luser's browsers.  Microsoft signed it?
It's got to be safe!

Try to disable the MS root certs?  All the on-line update stuff breaks.
 IF you don't blindly and completely trust Microsoft, you can't keep
your system up-to-date and patched.

> 
> The (secondary) idea was to prevent man-in-the-middle attacks by
> ensuring that the 'in the middle' guy has to a) at least expend
> more effort trying to get a similar certificate, and b) hopefully
> not succeed.

Successful MITM attacks can now be performed with free certs, or limited
time run "trial" certs available for free by many TrustCos.  My issue is
that with chain of authority verification turned off by default in most
browsers (bad, bad idea) and acceptance of browser developer's automatic
root cert lists, users completely mindlessly trust and transact on the
web with no value to the trust model.

> 
> Obviously, I have no idea if either of these are still being practiced
> by the companies in question.

The idea of trusted root CA's was good in concept but it got highjacked
by marketing types and turned into a cash cow.  As time has passed, the
lofty ideals targeted by the originators have become lost, and the model
has completely failed.  The PGP web-of-trust model seems to have stood
up better over time because of the lack of profit motivation and the
decentralized and gradiated trust concept.  I trust Dave O. cause I know
him face to face, so I can place a lower but still reasonable level of
trust in people he certifies as trustworthy.

I'm surprised no Internet/security heavyweight pundit like Schneier
hasn't taken on the profit!-motivated TrustCo model as pointless and broken.

--
Bill Strosberg