Re: [Tcpcrypt] Initial questions

[Tony Arcieri <[e-mail]>]

On Thu, Jun 19, 2014 at 10:55 AM, Joe Touch <[e-mail]> wrote:

> If you have only one, then if (or when) you urgently decide it's
> vulnerable and want an alternate you need to wait for deployment of an
> update (e.g., as happened to TCP MD5). That will undermine the utility of a
> solution.
>

Hey Joe,

Perhaps you should pay attention to 2014, and not look at things through a
historical glass, darkly.

In case you haven't been paying attention, vulnerabilities have been
accelerating at something of an exponential rate, and pretty much
everything that hasn't happened in 2014 is practically irrelevant at this
point.

Cipher agility is clearly not the problem. Implementation vulnerabilities
are the problem. In fact, the rate at which cipher vulnerabilities have
been discovered is practically glacial at this point.

Things have gotten a lot more complicated since MD5 is even a hash function
anyone with a clue would use for any reason whatsoever. Perhaps you should
stop referencing anything that has anything to do with MD5, period. It's
not germane to the discussion of any modern protocol, except perhaps if you
want to discuss historical failures.

-- 
Tony Arcieri