Neil Barrett

York

Writer

Part two...

In the computational approach, there is a third thing which we can do: we can execute it in a controlled environment. The computational equivalent of a laboratory test on a tissue sample.

Computationally, though, the problem is that the 'controlled environment' is itself a program which is necessarily larger than the tested program. Again, computationally we must satisfy ourselves that the system consisting of the testing and tested programs is not itself harmful - which requires a yet bigger program. And that too needs to be tested with yet another, yet larger program... etc.

That's the correct argument for non-computability which I rather skimmed over in the article.

I also skimmed over the 'what is security' argument - I might keep that for another article! Grin.

As to an email conversation on this, more than happy. My email is neil dot barrett at btinternet dot com. Comments and critiques welcomed...

Neil