Weblog:   The Growing Politicization of Open Source
Subject:   What about the quality of the software purchased by the government
Date:   2002-08-16 16:04:18
From:   pkobly
Response to: What about the quality of the software purchased by the government

>I disagree with the legislation proposal, not
>with you. Let's require open
>standards, not open source. But also, let's
>require security, reliability and
>performance acceptance level even before the
>open standards. Ask the (proprietary or open
>source) software companies to deliver software
>that works.

While this is a laudable goal, how do we assert that the software works or is secure? Black-box testing is simply *not* sufficient to assert quality with any credibility. Open source, allows the purchaser or the user (in this case the government) to perform its own testing and audits, rather than simply relying on the bare assertions of a vendor.

Remember, the Pinto went through testing before it was unleashed on the world.

> Purchasing a softare should give you the right
> to sue the software vendor if this software
> fails to work as it is supposed.

This requires a complete definition of how the software is supposed to work, and under what conditions. That definition often does not exist in general purpose computing.

If you ever get the right to sue vendors, then you ought to consider the likelihood of vendors being able to pay out on lawsuits. That same problem exists with many consumer goods. But it's a moot point now. Small companies are just as able to pay out the $0 judgements that you can get now as are big companies.

> Are you personally willing to guarantee the
> security and the reliability of a software
> solution you would sell to some of your client?

Is Microsoft?

> I assume you use gcc - would you be willing to
> stand by the binaries, produced by it, to the
> degree of taking the responsibility and
> allowing to be sued?

I would be more willing to do so while using gcc than while using a proprietary compiler. I don't have to rely exclusively on the assertions of a third party that the compiler works correctly.

1 to 1 of 1
1 to 1 of 1