The Public Accounts Commitee, the funding watchdog, has reported back on its investigation into the failed National Offender Management System - which was abandoned in 2007 after wasting millions of pounds - without pulling any punches. The EDS-run project was effectively gutted two years ago - it will now, hopefully, provide …
"C-NOMIS is a singular example of comprehensively poor project management"
I hope nobody tells them about the new NHS computer system, then they'd be REALLY pissed off! :)
and those being sacked are?
.... yeah i thought not... its not just in banks where failure is rewarded.. MPs show how it done!
Perhaps a few less committees might be a good way to get people responsible again, as the only real purpose of a committee is to share blame when it goes wrong.
how the hell do they get work? every job ive heard they have royally screwed up!
who in nu labour has connections with them? why arent costs nailed down like every other bit of work that goes on in the world?
plana project. quote for it. hold it to quote price unless any reworking needs to be done.
im sick of MY and YOUR money being pissed up the wall
How can they spend that much on a database - no matter how many bells and whistles you add at the end of the day it's a customer database (okay, the customers are scallys, but that's basically what it is)
I really think we need to set up the ElRegReaders Consultancy plc and underbid all these rip-off merchants - and we might actually deliver something that works as a bonus!
Whenever I read about DB's like this...
... I can't help thinking "Was the spec basically noms.prisoner_table, noms.prisoner_updates_table?". Maybe with a little acl_table chucked in. Then I see the 234m price tag and realise that either I'm terribly naive, or that they were buying Oracle licenses & consultants ;-)
My theory is that EDS and the like only get work because their name tells government officials how to spell it.
I'd extend that premise and challenge anyone to name a large organisation that is now primarily known by its initials yet has not had massive screwups and/or dysfunctional management.
First, EDS has been on the wane these last few years, winning less and less UK Government business and losing a number of contracts, because its project management is generally weak, because in turn those managers are not rigorously trained.
Second, the fault is at least 50% the client. In my 20 years I've learned that if only one side is rigorous in project management then all you can do is more accurately apportion blame when the project fails to deliver. If both sides are rigorous then you generally have a successful project, or one that is rescoped due to cost escalation as requirements and constraints are better understood. If neither side manages properly then the result is chaos.
If the C-NOMIS client had been project managing properly, then they would have spotted and acted on poor management by EDS. If EDS were project managing properly, the customer would have been penalised heavily for weak and changing requirements, but there would have been a successful outcome albeit late. It would appear from this report that neither side were managing properly.
Disclosure: I'm ex-EDS and for every good PM I worked under there were at least three bad ones.
NOMIS was not a UK project; it was not for Scotland, where we have our own Offender Management programme, which can't overspend by that amount as it has no budget at all. In the meantime, our central project for bringing Criminal Justice together, which goes by the catchy ISCJIS acronym has a massive £450k budget and two staff
Every DIp Shit uses us
Gotta say, as awful as EDS are (and they are truly awful), it doesn't help when their clients don't specify their requirements properly. That's a recipe for massive cost increases and delays.
Personally I think for large projects like this the project management should be done in house, everything else should be contracted out in little sections. Pay someone to do the design work (properly), break it down into lots of black boxes then tender for each black box. All of a sudden you have hundreds of possible contractors fighting it out to produce the best quality at a reasonable price, instead of EDS and Fujshitsu deciding who gets to eat at the trough this time. It also makes it really easy to terminate contracts as you only need to resolve one small area, not the whole darned project.
However, as that means less money for certain people that ain't ever going to be the case.
PS - the SRO had *no* project management experience?! Seriously?! Forget sacking him, what about the HR numpties that employed him? Christ on a bike...
Should be drowned at birth, or burned at the stake. Preferably both.
I'd hazard a guess there was fuck-all involved in this project design-wise and requirements-wise that couldn't have been handled by ONE competent enterprise developer. Hell, a semi-decent shareware author could have trained up and delivered a usable product in assembly language with that budget and timescale.
From experience, I can only assume the dev/test budget is ~500K and the rest goes on hardware (rental?), networking costs, admin costs, training planing, actual training, installation and config rollout planing and testing etc...
I have to say we were invited to help plan a rollout of a similar sized system for housing authorities.
Instead of each H/A paying independant consultancies for the same changes, they would collectively own the software and share dev and test costs, funding thier own rollout costs.
Sad to say the scheme never took off but it woud have saved them millions and given the
relatively small customer base (there were ~100 H/As at the time) change management
would be "noddy" compared to some projects...
So, how does this work then?
I can't help but think that most of the variables are known: What data to store, how to enter it, how to access it, who needs and who may have access, how to handle retention, dropping of records, backups, redundancy, availability, that sort of thing. So setup a pilot containing a few fat filers and a database front, roll out endpoints, hardware, client software, or both, to the various agencies, do it incrementally, and have the thing gradually take over the existing system with most data entry being done by the users (which would be the steady state anyway), and only after it's been deemed stable enough (do stress tests), decomission the old system. That's how certain police forces ended up doing dispatch supported by homegrown MSX software(!) though that ended up as unsupportable as the bbc micro in the end.
How much do you need for that? A couple good sysadmins, an interface writer or two, a documentation writer, a release engineer/toolsmith, one PM, and a couple field teams, say two to three people each, plus assorted hardware. Cost? Uh, a couple million quid and fiat from on high to get started, and scale as needed. I think you can do it for well under 100 million and inside of a decade, first try's the charm.
What am I missing here? Alright, this doesn't work if you're too big to start with, of course.
EDS is dead. Long live HP
Shorter name, lower rates. Same level of competance.
You know it makes sense.
Would I be surprised to find then that the SNP complain about Scots being "tightfisted"?
"Scottish Prices" story from yesterday, etc???
Maybe it should have been called "scots get great value"
So they had a £200M+ project under someone with no project management experience....?
In 1993 I was an £8-an-hour qualified-by-experience (i.e. totally unqualified) accounting temp on the books of a well known agency. They sent me to an interview for a job at the Prison Service headquarters in John Islip St, SW1. I sat before a panel of 3 or 5 interviewers (it is a bit of a blur) with steadily increasing incredulity as they explained that the Prison Service did not know where its money went, and this was a problem because they had to make sure they spent it all before the end of the year of the Treasury would cut their budget. They therefore wanted the successful candidate to devise and implement a management information system. ... There was just one problem. The person concerned would have to deal with Prison Governors, who were largely responsible for spending the budget, and er... were generally very good man-managers. "You mean they are usually inumerate?" I asked. Yes.
I did not take the job. I often wonder who did.
But databases are *easy*.
They just need thought and planning. Mostly planning.
If you don't plan, it will fail. And it'll fail really, really badly.
I honestly don't understand how these projects go overbudget by so much before anybody seems to notice.
If they actually thought about this, you could almost certainly find a bunch of CS final year students who could give you a working model within six months for under £10k.
That working model won't be great and will probably have to be thrown away afterwards, but it will show you exactly what is *actually* required of the system and will probably have a reasonably good UI. It just won't scale.
And having got that working model, you could get a proper system within a few years for well under £100 million. Heck, if my employer can develop entire new control systems in under five years with our tiny resources, aimed at markets of a total size of only 100 million or so, why can't these guys do a *single system* for that much?
"If EDS were project managing properly, the customer would have been penalised heavily for weak and changing requirements"
but they were - that's why the costs escalated >3x