[SystemSafety] Explainability of "Algorithms"

[Matthew Squair]

Yes we’ve just experienced the power of bad algorithms here with our
Centrelink welfare debt recovery debacle.

Centrelink is the Australian Government agency charged with administering
welfare payments,and as part of clawing back what they believed was $4
billion worth of overpayments they automated the cross checking between the
Tax Office and Centrelink records using ‘fuzzy name matching algorithms’.

Needless to say it didn’t go well, some estimate that 1 in 5 notices were
in error. We don’t know how many people got a false notice and paid up just
to get the department of their back either.

Regards,


On 29 June 2017 at 2:57:31 am, Derek M Jones (derek at knosof.co.uk) wrote:

All,

> Just noticed this article, which is an author interview regarding her new
> mass market book, Weapons of Math Destruction.

Another take on 'bias' in algorithms:
https://www.theguardian.com/technology/2017/may/02/facebook-executive-advertising-data-comment

His book "Chaos Monkeys" is very good.

-- 
Derek M. Jones Software analysis
tel: +44 (0)1252 520667 blog:shape-of-code.coding-guidelines.com
_______________________________________________
The System Safety Mailing List
systemsafety at TechFak.Uni-Bielefeld.DE
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.techfak.uni-bielefeld.de/mailman/private/systemsafety/attachments/20170628/06c57902/attachment-0001.html>