Re: Microsoft and Security

[Jason Coombs <jasonc () science org>]

Alun Jones wrote:
> ... okay, so you're arguing that even more QA and more testing should be
> <snip>
> releasing a smaller fix, with minimal impact, as soon as possible.
> <snip>
> improving the process, perhaps you should try and express those suggestions
> in a coherent manner that could be used
...

Aloha, Alun.

My suggestion is a simple one that all software developers can manage to 
incorporate into their busy schedules and tight budgets:

Hire an expert to conduct a thorough forensic review of the software 
before it is released, and publish the forensic analysis report.

Any vulnerabilities, flaws, areas that need additional work, portions 
that were built by subcontractors of questionable skill or loyalties, 
portions that were offshored, features that the programmers themselves 
warn are not yet done by placing comments in the source code, third 
party libraries or code or algorithms that may create intellectual 
property liability for the end user, and all other issues of computer 
forensics and computer law should be spelled out as clearly as possible 
by any company that develops and distributes software to the public.

Anyone who does not publish a forensic analysis report along with their 
software should publish the source code, whether or not they release 
legal rights to that source code under an open source or free software 
license.

The computing public should not have to reverse engineer software 
products in order to figure out what they do to the computers on which 
they are installed and used.

Even the Department of Justice knew better than to allow the FBI to 
build and deploy law enforcement computer technology without hiring an 
expert to write a forensic report on the product, and the FBI doesn't 
try to sell "Carnivore" to anyone.

http://www.epic.org/privacy/carnivore/

Final Independent Technical Review of the Carnivore System
http://www.epic.org/privacy/carnivore/carniv_final.pdf

We should require software vendors to take this stuff seriously.

Sincerely,

Jason Coombs
jasonc () science org