I would love to hear the details of this project. Perhaps it can made into a lessons learned (yet another one) for high up government officials. There has to be something to learn here. Anyone have any info on this?
> Perhaps it can made into a lessons learned (yet another one) for high up government officials.
I'd be much more interested in comparatively low level people, maybe a team lead or something? I feel the exact same way -- how could someone not be genuinely curious about this type of thing? But I'm a little weary of getting a 'series of tubes' style answer from someone (see the White House response to Healthcare.gov as perhaps a best-reasonably-likely-case scenario -- it didn't foul up the explanation completely but also didn't really speak to the details in any way that you or I would like).
I'm thinking more and more about this and other questions have come up. The population of Ontario is about 13 million. This isn't even a national service. It's provincial only.
Besides what appears to be terrible software, you have to wonder about the procurement methods? i.e. How the contracts were awarded, measures to ensure that money is properly spent, protection clauses that ensure that the government gets value for dollar, etc.
To spend that much money on something that small is mind boggling.
This only a single example. The more relevant issue is to what extent this is representative of other, similar bureaucratic costs imposed by means-testing and other targeting.
Also, you need to consider how many Canadian citizens are already receiving benefits and how much under the current system. If a very large percentage of Canadians are already receiving benefits, then the cost-savings of any system that prevents the remainder from claiming benefits is accordingly diminished.
Means-testing is not a difficult problem to solve with software. This project was grossly overprovisioned and mismanaged to a degree that probably indicates corruption, yes, but that happens in all kinds of software that has nothing to do with means-testing.
I know someone in Ontario who actually works with this software and I had her explain the requirements. Yes it is extremely hard problem to solve with software.
They are trying to quantify qualitative data, but each case is so varied that the software had thousands of options and checkboxes.
They needed the checkboxes to track metrics for specific trends in cases and detecting problem areas. So the state can detect say a rise in poverty in x neighbourhood do to y circumstances etc.
Data tracking at this level is incredibly complex. And I'm sure IBM, or any gov enterprise software team doesnt have many good designers leading the projects. Designed in comitee rooms.
UBI takes these people out of the equation and the software is simply, here is everyones address/bank -> pay them.
A quarter billion dollars on a provincial software project. This is absolutely mind boggling. Between this and the Presto boondoggle[1] we are probably closing in on a billion dollars being lit on fire in Ontario alone.
Not surprised. The Ontario government is probably one of the most inefficient government bodies in the world. Province debt size is twice the size of California, living there is like being in a communist state, can't really expect much from them.
Its sounds from the limited info in the article that they didn't run the old and new systems side by side for a period of time ... which I thought would have been standard practice with this sort of thing?
assume that all went to cover cost of programmers or equiv profs
lets say fully loaded annual cost of said staff indiv is $150K each
$60M gets you 400 of those $150K programmers FT for a year
now do that for 4 years
for something that is essentially just a database-backed CRUD webapp with simple workflow and the hardest bits actually done by a pre-existing separate service anyway (messaging to banks to issue the DD; printing checks and envelopes then handing off to PO for delivery); and could have been orders-of-magnitude simpler (and cheaper overhead) if all replaced with a totally automated Basic Universal Income payout to every citizen
gov software: massively lucrative if weren't so massively inane
I suspect it was done in the usual manner with wiggling requirements and re-writes several times over. Plus the people you may be replacing with software may be the only ones knowing the ins and outs of the system, thus dragging their feet at every opportunity.
21 comments
[ 2.0 ms ] story [ 57.1 ms ] threadI'd be much more interested in comparatively low level people, maybe a team lead or something? I feel the exact same way -- how could someone not be genuinely curious about this type of thing? But I'm a little weary of getting a 'series of tubes' style answer from someone (see the White House response to Healthcare.gov as perhaps a best-reasonably-likely-case scenario -- it didn't foul up the explanation completely but also didn't really speak to the details in any way that you or I would like).
Besides what appears to be terrible software, you have to wonder about the procurement methods? i.e. How the contracts were awarded, measures to ensure that money is properly spent, protection clauses that ensure that the government gets value for dollar, etc.
To spend that much money on something that small is mind boggling.
http://www.techvibes.com/blog/bc-government-scraps-scrap-89-...
Basic income is sounding better and better each day now.
Also, you need to consider how many Canadian citizens are already receiving benefits and how much under the current system. If a very large percentage of Canadians are already receiving benefits, then the cost-savings of any system that prevents the remainder from claiming benefits is accordingly diminished.
They are trying to quantify qualitative data, but each case is so varied that the software had thousands of options and checkboxes.
They needed the checkboxes to track metrics for specific trends in cases and detecting problem areas. So the state can detect say a rise in poverty in x neighbourhood do to y circumstances etc.
Data tracking at this level is incredibly complex. And I'm sure IBM, or any gov enterprise software team doesnt have many good designers leading the projects. Designed in comitee rooms.
UBI takes these people out of the equation and the software is simply, here is everyones address/bank -> pay them.
[1] http://www.thepost.on.ca/2012/12/13/presto-yet-another-liber...
While the Ontario debt is larger than the California debt, it's not really a fair comparison.[1][2]
And on the communist state bit, uh what?
[1]: http://www.thestar.com/opinion/commentary/2014/03/21/ontario... [2]: http://blogs.wsj.com/canadarealtime/2014/03/18/ontario-debt-...
thats $60M/year
assume that all went to cover cost of programmers or equiv profs
lets say fully loaded annual cost of said staff indiv is $150K each
$60M gets you 400 of those $150K programmers FT for a year
now do that for 4 years
for something that is essentially just a database-backed CRUD webapp with simple workflow and the hardest bits actually done by a pre-existing separate service anyway (messaging to banks to issue the DD; printing checks and envelopes then handing off to PO for delivery); and could have been orders-of-magnitude simpler (and cheaper overhead) if all replaced with a totally automated Basic Universal Income payout to every citizen
gov software: massively lucrative if weren't so massively inane