Feeds

back to article Poor IT contributes to DWP errors

The Department for Work and Pensions is hampered by out of date IT in tackling benefit fraud and error, says the National Audit Office. The spending watchdog's report on minimising the cost of administrative error in the benefit system, published on 25 November, says the Department for Work and Pensions (DWP) has 140 core …

COMMENTS

This topic is closed for new posts.

Fraud versus Error

By putting the fraud and error figures together DWP hides their biggest problem. Error is running in excess of 5 times the fraud level, which shows up the incompetence and indifference of DWP assessors entering the information. By spending millions on adverts and carefully planted 'news stories' that infer every severely disabled or dying person is a fraudster, the £6bn in errors are routinely hidden from the public. A recent major report into how disabled people are assessed concluded that the system is fraudulent, with many trick questions and disability reports being altered in favour of the DWP.

Even on an older system, if the information entered is correct then the payments will be correct, it is just number crunching. The whole system is UNFAIR and they know it!

12
0
Bronze badge
Thumb Up

Great stuff!

So I must ask, where has all the IT budget cash actually gone to?

Oh yes, of course, silly me! Into the pockets of the mates of the MPs, who just happen to be on the board of the IT services suppliers who can never finish any projects they started, leaving a trail of broken IT infrastructure behind them!

3
0
Unhappy

Data validation?

Some years ago I was involved in doing some pre-delivery checks on a bit of software for the then DHSS. We found numerous data errors that the validation missed.

e.g. Date of birth 00/00/1800, male, pregnant, receiving old age pension and living with grandparents.

Our report was shelved with no action taken as it was "too close to the delivery date to make changes."

3
0
Bronze badge

Already Broken

I am now handling my elderly mother's pension affairs.

A few days ago, I received a letter notifying me of a decision, and the right to appeal. The letter was horribly unclear on what the decision was, and I made the inference that, somehow, the middle of a sentence was missing. The grammar and syntax was so broken that it obviously hadn't been read by a competent human.

And why should it be? This sort of notification is an obvious candidate for being computer generated. But what was missing suggests something is equally obviously broken.

At least I know that it's about a state pension, but that's only an inference from the address for queries.

2
0
FAIL

Top down IT procurement at work

My GF used to work at the DWP and from what she's told me their IT provision is disgustingly and inflexible. Requirements are determined and written at a level far higher than the actual lowlies who sit at the keyboard, and consequently they never get a system that makes their job harder.

The whole thing needs redoing from scratch with proper stakeholder involvement. And by stakeholders I mean people who do the actual claims processing, not managers who get nice food from contractors during sales meetings.

5
0
Anonymous Coward

same story day after day

Shouldn't these programming companies be sued for not delivering the product paid for?

Day after day you hear about sloppy, poor programmers producing rubbish.

2
0
Megaphone

product paid for?

They usually deliver the product specified in the requirement, with all changes requested by the customer and within the customer environmental constraints and often do so at a loss. The problem is usually with the clarity and completeness of the requirement, the number of changes requested and the speed at which decisions on those are processed by the customer.

If the public sector had any drivers to cut costs or time then things would improve a bit but then I have heard the comment that most DWP employees would be in a benefits office anyway, the only thing the government can influence is which side of the counter they sit on.

1
0
Anonymous Coward

you'd be amazed

how often developers deliver exactly what was asked for, only to be told that wasn't what they actually wanted!

In fact you can get into trouble for giving them what they want instead of what they asked for. Now I stick to creating only what they have explicitly asked for in black and white. A cursory, "are you sure you don't want ..."* is all they get.

*copied to their boss for later, obviously :)

4
0
Anonymous Coward

@product paid for?

Yep, govt depts are particularly adept at this kind of thing, I've put systems in suitable for hundreds of simultaneous users that were way more expensive than they needed to be because they were actually only used by about 10 people - and guess who gets the blame?

More recently, we have been trying to organise a technical refresh project (to meet with contractual requirements set by the customer) and in order to deliver to the go-live date that the customer has asked for we should have started work two months ago.

We're still waiting for them to decide about some of the work that will be done as part of that refresh and if they follow usual form, they won't make that decision for another few months, but they'll *still* want us to meet the original deadlines, meaning the project will be six months behind schedule when we start. And the catch-up will cost real money because we'll need to throw more resource at it.

But, of course, it's all the supplier's fault, because Joe Public doesn't see any of that.

1
0
Silver badge
Unhappy

Draining the public purse

"Benefit processing systems are not designed in a way that allows simple changes to the screen display"

Well what a surprise. Big IT companies build overly complex systems that only they can support, keeping their snout at the government trough.

Perhaps Government IT should be done in house? Oh no, that would mean a bloated public sector paying market rates to their staff. What would the Daily Mail say?

At the very least, the government could give the work to companies registered in the UK for tax purposes, using local employees rather than those brought in through the intra-company transfer loophole. Base the work in regions suffering the highest unemployment and it could attract some EU funding. It might mean higher costs up front but the money stays in the UK, getting spent (and taxed) within the UK economy, stimulating employment and reducing the benefit bill.

3
0
FAIL

@Red Bren

"At the very least, the government could give the work to companies registered in the UK for tax purposes, using local employees rather than those brought in through the intra-company transfer loophole"

Couldn't agree more.

I could have sworn that in the run-up to the election, David Cameron mentioned breaking up large govenrnment IT contracts into smaller ones so firms other than HP(EDS), Logica et al could bid on them? Or was I dreaming?

2
0
Silver badge
Big Brother

Don't be ridiculous...

..putting forward an idea steeped in common sense?

One that makes perfect economic sense?

One that isn't crippled by partisan ideology?

You'd never get ANY government to adopt it...

0
0

Well that explains it...

I mean that explains why the DWP IT systems that local authorities are made to use are so rubbish!

0
0
Anonymous Coward

Outdated IT?

I honestly don't know what "IT" they have, but it sounds like all the actual useful processing might be barely enough to strain a contemporary desktop CPU. Might have to resort to batch and it'd be i/o bound and dead slow and whatnot, but it could be done. What it sounds like is stacks and stacks of disparate crummy software that nobody knows how to make work together, and going by the comments here exceedingly bad management.

It might be more efficient to print out those 20M case files and do it by hand. At six minutes per year average handling time per case you'd need, ideally speaking, 1250 case handlers. Plus management overhead and some administrative personnel to keep the show running you'd need about 1400 to 1500 people and no IT beyond phones and typewriters. How many people do they have now?

Not that I'm advocating doing that, but it seems an useful baseline to establish. Likewise, at an estimated 3.1e9 in overpayments and 1.3e9 in underpayments that's UKP 1.8e9 potential gain and that again means spending that much to fix this ought to pay for itself in a year. Best of all, you don't really need to spend that much. All you need to do is kick out the management (they're responsible, you know) and bring in "efficiency experts" to make the rounds, listen to the workers, and work with them to come up with tools that are actually helpful, plus some new management to oil & grease the new & improved DWP machine.

I don't know what's reasonable in effort-per-case or even how convoluted the rules are but it might help if the rules are as simple as they can be (and no simpler), so that's worth looking into also. But anyway, none of that will happen, for expecially incompetent management likes their cozy jobs and making a big show of "new IT" is a great way to spend a lot on IT consultants while not improving anything. It's the economics of the thing.

2
0
Paris Hilton

What? No data entry accuracy checks?

Daft innit?

Input errors are probably quite high and I'd guess that they are using a linear error projection model that concludes: save costs on error input.

At the same time overlooking that it probably takes a whole team to verify data as an afterthought once errors are identified.

Management perspective:

1 - it makes us look super efficient all running around to put things right that's why we like things to go off/awry/wrong in the first place?

2 - if we have to employ another 15 year old to enter data our budget costs will go off target?

1
0
WTF?

comfy chairs....

"Benefit processing systems are not designed in a way that allows simple changes to the screen display, the watchdog was told. To add a new data field would require significant work to ensure the processing code understands the messages being keyed in."

So hard coded hard wired crap designed to be hard (i.e. expensive) to amend.

Table driven.

Abstraction layers.

Data independence.

All myths in the world of HMG IT ?

Utter despair of the pond life that 'lets' this happen.

Anyone who has read an ITT for a replacement for an in-place 'solution' will have seen the poison in the ITT.

Trebles all round !!

2
0
FAIL

re: What? No data entry accuracy checks?

You should see some of the personnel data that the NHS had - in a past universe far, far away I had to help interface our (fairly clean and tight) system with data provided by the NHS. The data quality was shocking - stuff that really should be in fixed-format field was free text (and so the operators could punch in complete junk and often did) and stuff that should be free-text often wasn't so operators couldn't put in the stuff that they needed to because the system designers hadn't got a clue about the data, what it meant and how it was used. Which is why that data used to end up in the free-text fields that were intended for something else..

This contrasted with our system that was developed with the help of the people actually doing the work and so contained what *they* wanted and needed rather than what the system designers thought should be there.

The end result?

Our system got canned after a committee recommended that its features be migrated to an existing product sold by someone else and used within the NHS. And the head of that committee was on secondment from the company that provided the system that ours got borged by..

1
0
Anonymous Coward

dontcherknow

analysis is old fashioned, it's not the done thing!

At least that's what i got told when i asked why no-one in the company does any anymore.

1
0
Gold badge
Flame

*140* systems. 140 Systems

Can you say data synchronization issues? Some systems data right, some not. but which ones?

Can you say data cleanliness issues? they all say John Smith's birth date is 99/99/99.

Yes that can't be right (but if you stuff the underlying data base files rather than go through proper data entry certainly possible.

"Benefit processing systems are not designed in a way that allows simple changes to the screen display"

Because there's *no* way the rules will change or further questions will be needed at a later date (or dare I even *suggest* it they might not need to ask some of them anymore).

On a 140 *different* systems.

In *how* many different data centres?

But let's be honest if you were an IT con-tractor ( I don't mean you hard working harassed IR35 types. I mean the HP/IBM/Crap Gemini people who win these) that's *exactly* how you want it developed.

Systems that are *almost* (but not *quite*) impossible to modify, provided there is a big enough budget with just *less* than the number of bugs needed to fail acceptance. Or rather that were *found* before handover. So you can start charging for the fixes when the *rest* of the bugs get found out.

But *damm* what a scope for improvement. £!bn to run. £1.9Bn in nett over payments.

there's *got* to be room for improvement here.

0
0

@John Smith 19

No, that's not how the suppliers want things developed, because that type of thing costs them more to support and eats into profit margins.

A good few of these systems could well be 30-40 years old and probably developed in house when the departments had their own IT staff and there are so many of them because the departments were all separate entities in those days, as they got amalgamated over the years there was likely to be little budget or compelling reason to merge the systems and you suddenly find yourself in the position they're in today.

Yes, there's huge scope for improvement, but that needs money, which no-one wants to spend. Pick any two random systems and you'll probably find they're running on kit that's anywhere between five and fifteen years old, so if you wanted to merge them you would almost certainly have to buy one new server to do it on.

Add to that the data centre space, design work for the merged system, training for staff to use the merged system, building and testing the new system, secure disposal of data on the old system and you could find that it would take 3-4 years to recover the cost of the system against the money you save.

That should be all fine, but as soon as you decide to do that you find yourself beset by the unwashed masses who think that all of that transformation work should be done for free and that hardware is made from fairy dust (no, it just runs on magic smoke). If anyone dares to charge for any work then they're evil money-grabbing b*stards.

0
0
Gold badge
Happy

@chr0m4t1c

"No, that's not how the suppliers want things developed, because that type of thing costs them more to support and eats into profit margins."

that tends to depend on how well the con-tractor has managed to "negotiate" the con-tract.

"A good few of these systems could well be 30-40 years old and probably developed in house when the departments had their own IT staff and there are so many of them because the departments were all separate entities in those days, as they got amalgamated over the years there was likely to be little budget or compelling reason to merge the systems and you suddenly find yourself in the position they're in today."

No doubt. However it's a truism that support costs rise as systems get older. I find it *very* hard to believe that the point has not come for *some* of these systems to be shut down as their support costs (old hardware/migration costs of a re-write to go on new hardware/cost of processing change requests for software written in creaky 70 4GL) to the point that it *will* be cheaper to shut them down, merge their functions into a new generation of some of the remaining systems and link in the relevant data to the new system.

"Yes, there's huge scope for improvement, but that needs money, which no-one wants to spend. Pick any two random systems and you'll probably find they're running on kit that's anywhere between five and fifteen years old, so if you wanted to merge them you would almost certainly have to buy one new server to do it on."

I'd say it needs *leadership* by people with solid map of the systems data models and interrelationships. See my comments on your previous paragraph.

"Add to that the data centre space, design work for the merged system, training for staff to use the merged system, building and testing the new system, secure disposal of data on the old system and you could find that it would take 3-4 years to recover the cost of the system against the money you save."

And of course you would do *all* of these together. A classic big bang with *everything* depending on everything working out properly on time. What could possibly go wrong?

I find it *astonishing* that the UK government seems to operate c150 data centres. US states bigger than the UK run their *whole* operations for a given state services in one data centre.

Let's start with identifying *which* systems should be merged and *then* begin a phased process of merging the support teams and hardware, then beginning to migrate to 1 hardware platform and eventually a unified application.

That should be all fine, but as soon as you decide to do that you find yourself beset by the unwashed masses who think that all of that transformation work should be done for free and that hardware is made from fairy dust (no, it just runs on magic smoke). If anyone dares to charge for any work then they're evil money-grabbing b*stards.

Maybe it's because they have had the sharp end of a *very* sharp stick thrust up them too often to entirely trust those nice IT con-tractors?

At the risk of being boring I find it *very* hard to believe the support costs on some of that ancient software has not started to climb through the *roof*. What is the going rate for a skilled IBM S360 COBOL programmer these days?

I imaging that allowing lower case note entry SO IT DOES NOT LOOK LIKE ALL THE NOTES ARE BEING WRITTEN BY SOMEONE SHOUTING alone would save a fortune in eye tests and spectacle upgrades.

Or perhaps you could just start on a decent data cleansing run to pick out the real stupid stuff. I know there should be no entries with invalid birth and/or death dates and of course if you do have a date of death it should be *after* the DOB should it not, but something tells me that won't be the case. I'll be there's a *lot* more rubbish to find.

0
0
FAIL

1852: year of the anecdote

Was a good year for people being born. This apparently was the default date that the whole "mainframe" legacy system was based on. Some pensioners got letters with an 1852 date of birth. The postcode part of addresses also caused problems too, if a postcode was entered that the system didn't like, it was refused, even after checking against the Royal Mail database. If it wasn't liked, folk didn't get paid, and the matter was referred to others. I was also told that you could **NOT** log stuff onto the wrong servers, however, when I worked there it was something I managed to do with ease. However, entering punters NI numbers was even more fun, as sometimes it didn't bring up what was expected, i.e. There was something wrong. I was told that *I* must have made the error as there aren't any errors in the system. I replied "I can only put the number I get off the punter and have to assume it is right". I've read/heard stories of local women who have similar birth dates/names being tied to the **SAME** NI number. This, in a totally error free system wouldn't happen, but an error free system is a bit of an abstract, and it took them nearly 40 years to realise there was a problem. The problem of separating the ladies (very labour/brain intensive) from the number (they did use pen & paper at one time), so it must've been a case of not checking for errors as previous posters have mentioned. Most of the minions don't know how to use it properly, as the training manuals are hidden or "missing. In my case I was lucky as I had enough knowledge of elderly systems to get by, but a lot of others (including team leaders) didn't have a clue. Oddly, just before I left, I came across a training manual, which would've been helpful when I logged stuff on the wrong server, thus buggering folks' pensions. I was told that de-logging them took ages, but as I hadn't been trained properly (training lasted approximately 10 minutes), I didn't know what I was doing (so I did as I was told).

Lastly, some free advice: If your NI number starts with TN or TM get it checked out as these were (and still are, as far as I know) used as temporary numbers. If the taxman has these down as yours, it could quite easily bugger up your life, so get it checked out.

0
0
Paris Hilton

Cost of correcting dodgy data

I'd guess that few organisations calculate the cost of correcting dodgy data.

The probably prefer to sandbox into "Dodgy data? Here? We don't do dodgy data!" type denials.

And homegrown systems soon evolve into home-groan ones with evolutionary cul-de-sacs dependent upon the individuals who started something helpful in an office type database.

Training manuals to an insecure manager might just equate to the curse of the devil (especially if the newbie understands it).

0
0
This topic is closed for new posts.