[ExI] Security clearances

[Anders Sandberg]

On 2016-05-14 23:06, spike wrote:
>
> Anders, this might be the only time this ever happens: I disagree with 
> much or most of what you posted (I left it in, below.)
>
Cool. And mildly frightening.

> Little evidence they work: very much on the contrary sir.  I might not 
> say anything more on that topic other than I disagree.
>

Would there be any way to actually document this fact?

My problem is that in my world a claim is not to be trusted until there 
are ways of verifying it or that all (reasonable) counterclaims have 
been falsified. Finding a way of demonstrating that classifying actually 
does work (and how much) would be very useful, and actually pretty 
important.

I can imagine looking for natural experiments where the system 
accidentally failed to classify or not in parallel organisations, a bit 
like how the US Army accidentally demonstrated the role of IQ in soldier 
usability due to a clerical error.


-- 
Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20160515/65620fdc/attachment.html>