Rigorous ITIL and Prince2 practices strike again!
Oz auditor: Number of times failed government biometric project met a milestone = None
How much IT can you buy for AU$34m (£18m, $24m)? None at all, if you're the Australian Criminal Intelligence Commission in the market for a biometric system. The Australian National Audit Office (ANAO) this week released its report into ACIC's failed biometric identification system (BIS) project which failed last year, and …
COMMENTS
-
-
Wednesday 23rd January 2019 22:33 GMT Sorry that handle is already taken.
Re: AU$34m? Pfft!
Remember when Queensland Health contracted IBM to provide a payroll system for $6m and ended up paying $1.2bn for a smoking pile of slag?
https://www.theregister.co.uk/2012/12/13/queensland_payroll_ibm_sap_inquiry/
https://www.theregister.co.uk/2013/04/11/queensland_health_payroll_inquiry/
https://www.theregister.co.uk/2013/07/17/vendors_snap_at_each_other_in_queensland_health_inquiry/
https://www.theregister.co.uk/2013/12/06/australian_state_to_sue_ibm_over_aud1bn_project_blowout/
https://www.theregister.co.uk/2014/12/01/supersueball_heading_ibms_way_in_australia/
https://www.theregister.co.uk/2015/12/08/ibm_vs_queensland_health_payroll_decision/
https://www.theregister.co.uk/2016/04/07/court_drops_ibms_costs_on_queensland_government/
We (well... so far Queensland) can fuck things up just as well as anyone else!
-
-
Wednesday 23rd January 2019 14:14 GMT lglethal
Geez, I wish the government would give ME a goodwill bonus of $2.9 million. I mean, I've produced the same amount of usable work as this project, and I havent cost the government a penny, Heck you could say that for the same work I've SAVED them $34 million. I'll be waiting for that goodwill bonus in the mail, then shall i?
-
-
-
Friday 25th January 2019 13:14 GMT Michael Wojcik
Re: They should hire me
This isn't eventually-consistent; it's the weaker "eventually good enough" guarantee, under which most successful computer applications work.
A yet-weaker but surprisingly still useful guarantee is "never going to work". It's still useful because a surprising number of vendors get paid for it, as in this case.
-
-
-
Wednesday 23rd January 2019 18:10 GMT MonkeyCee
Requirements first
"A requirements-gathering process handled by PwC overlooked these, meaning they were also missing from the 2015-issued tender."
Ah, so it was fucked from the start.
If you can't pin down the requirements for a system, especially all the assumed domain knowledge, then things will be fucked.
The method that served me well has been "explain it to a Martian" technique. It will seem stupid, but if a spec is written that an alien could understand it without requiring any additional explanation, then it should be pretty clear to whoever ends up having to write the solution.
Since this often involves much more involvement of the actual users of the system, as compared to manglement (who have a very theoretical view of how things work at best), it often results in a more useful system. It's also HATED by many middle manglement and system architects, as it shows that they don't actually understand what the system is for, or even how the business runs.
If it's not in the requirements, it's not going to be in the product. Yes, even the things you'd assume would be there. If there an assumption, write it down... :D