Tories put toes on Linux bandwagon
The Tory party will if elected end government over-spending on IT projects by simply choosing open source alternatives and splitting projects up, it believes. This cunning plan will save us £600m a year, we are told, mostly thanks to increased competition. According to figures from Mark Thompson of Judge Business School, 80 per …
But still they should be applauded for recognising the ludicrous current spending.
"the Labour government's management of IT projects is utterly awful - even given the rubbish record of governments of all colours"
Let's face it, I.T. has only really taken off in government in such a big way since Labour came to power and decided that massive databases and the internet could fix everything, so there's really not much to compare it with.
As to the Tory idea, have they factored in the cost of staff retraining?
While just as in the tone of the article there will be alot of cynicism about cost reductions anything that opens up the powers that be to open source has to be good
With so many other governments doing the same, you might say 'it's about time!'
But it does go beyond this specific announcement - the announcement could also have an effect on other corporate and gov initiatives.
Who knows it could even filter into EDUCATION!
That would be something....................
shoehorn the requirements into it at a later date. Ermm excellent solution to getting Public IT out of the mess it is currently in.
How about a novela pproach that is extremely daring........
Get the requirements right before chosing a partner and technology!!!!!!
Not exactly a great achievement to spot the bleedin' obvious though is it...?
So we have are software form Emo and Boris it runs like a dream then we srick it on an industrial sized infrastructure. Problem is Emo only ever tested it on his mac VM player running a test harness for 50 connections (because he never expected it to get so big and no one will give him backing to test it properly as it's free) so he has no idea whether it will work which means we now have to employ specialist consultants to help re engineer the thing at !K per day but not to worry because its not like anybodys life depends on it;( not as though any of the contracts are critical.
Get with the program guys next you'll be suggesting we run it on OS2.
El Reg please write a BOFH spoof about this shower
This reads as:
* read the press release;
* haven't read anything else;
* must be shit.
... who knows, you might be right.
Personally I think 'open source' and 'free software' being on the agenda of any political party other than the Greens is a *good* thing.
The current state of ICT procurement in government is foobar.
Leaving aside the fact that Windows is indeed awful and should be eliminated at all costs, even if that cost is Linux, this is not the reason why government projects fail.
They fail because :
1/ The are specced out by people who don't know what they are doing, or who are being paid off to deliberately spec them in the way they do.
2/ The contracts are not nailed down. They have unenforceable penalty clauses for failure or such clauses are never exercised. When things go wrong, the supplier ALWAYS manages to wriggle out of their responsibilities.
3/ Time and time and time again, projects are handed out to the same tired group of companies that consistently screw up and cost a pile of money. It's like the big defence (not necessarily IT) projects going to the BAe's and Marconis and GECs and Plesseys of this world (not all of which still exist, of course) - they ALL screw(ed) up, usually BIG time, ALL the time, and they STILL get/got more work. It's the same now, except the focus is on outrageous IT systems for the NHS or police or whoever.
4/ No accountability. Nobody (NOBODY) ever takes the heat when it all screws up. They just walk away and (in the case of the suppliers) bid for the next job or (in the case of the ministers and politicians) give up politics and go and work for the suppliers!
5/ Many of the projects (and this is where the Tory's may have a point) are WAY too big. They are bound to fail simply because it's not possible for something THAT big to succeed.
Ever split an order up so it goes in under your purchasing power? Dell's web site even lets you do this automatically now.
It will involve the contracts going to the same big firms - small Linux based consultancies can't pay big enough bribes. EDS etc will simply split the contract up to a number of subsidiaries.
In addition, a lot of the 20Bn wasted on It projects is because everything gets tacked onto this nice new big budget, every manager's company car, every admin staff, every bit of rewiring or office refurb. These are all going to still be spent, it will just take more effort to hide them on 1000 different projects - but we can always hire more managers an accountants to do this.
It has to be said that, if you commit to smaller projects designed to co-exist it makes it far more difficult to come up with overblown meta architectures and over complex infrastructures typical of this government's IT strategy.
The article is right to suggest that the National Programme for IT would be poorly served by 12 separate project managers and specs. However, if you've committed to smaller, achievable projects, you wouldn't try to design something like the National Programme for IT as your final solution.
If you discover that road bridges with impossibly long spans cannot safely be built using long beams, choosing to build out of small blocks doesn't preclude building bridges - it just means you have to cut your cloth to suit the more conservative material.
Having a good idea of the requirements before they start these billion-pound projects would be a good start. Too many government projects start with "we'll employ a team of external consultants to work out the requirements, and another team of external consultants to answer the first team's questions".
<quote>Is this true? Would the disastrous National Programme for IT...work better with 12 separate project managers and specs?</quote>
I seem to be able to make our own in-house application share data with 3rd party applications developed independently through published APIs and the odd conference call. IT.Gov is bigger, but otherwise, where's the difference?
And as for any lame "our software can't even do an export to CSV for a civil servant flunky to upload into yours" - then surely a definite criteria for selection must be the availability of an API that a techie can read and believe - not sweet smelling bull#### around a boardroom table.
Certainly the idea of mulitple vendors competing for projects rather than an incumbent racking up the expenses bill must deliver some benefit.
Interesting idea. Tell me though, when have you ever seen an IT project where the requirements are defined at the beginning and don't change? Never is the answer. Live date is just the date when the requirement changes become a little more urgent.
There is a lot to be said for the idea of breaking these large projects down. It would reduce risk and also spread the cash around. The likes of EDS and co are just getting huge amounts of cash for delivering stuff that does not work. Smaller companies deserve that chance too. Although I'd think they'd be better at delivering what is required.
We are almost at the stage where you can take pieces of open source software and plug them together like Lego. Easy on the scripting glue. It would be nice for the Tories if there were lots of government orientated chunks of software sitting on sourceforge just awaiting download.
They are right to go for smaller projects- if they are claiming to be market-oriented then rather than, for example, establishing a national Police IT system, they could be establishing a national data interchange specification that describes exactly what data other police forces or agencies need to be able to obtain from any given Police Force IT system and then allow the police forces to do whatever they need to in order to make sure their system is compliant with it. This gives the opportunity for smaller companies to specialise in creating software to comply with that standard and potentially sell it to many regional forces, or allows police forces to build something that works with existing software to do the same job.
As long as it looks the same from outside, the implementation can be absolutely black-box.
- They never have the cojones to hold vendors to contract (I saw one Army Colonel who tried, so Big Blue went above his head and had him "re-assigned").
- Institutional memory on previous cockups is extremely short
- Most IT effort is spent building empires, not solutions
Open standards are more important than the specific type of licence here.
Different systems, from different suppliers being able to work properly with each other would increase competition and lower costs. That has got to be a good thing.
The press release says "create a level playing field for open source software".
It doesn't actually mandate it.
Not only should it speed the downfall of the protected software monopolies (M$ et al), increase compeition, reduce cost and deliver increased functionality, but it should ensure that many smaller and often more innovative firms get a slice of the pie. Our economy stands to benefit far more from such an approach than by offering huge, bloated IT contracts to huge, bloated IT multinationals (almost none of which are UK based).
It's funny to hear people go on about the cost of Linux. A couple of my friends work for a large e-commerce website, and I hear nothing but horror stories about Linux and MySQL and some horendous open source message-passing package they bought. Their IT resources are spent entirely on work-arounds and dealing with the fact that almost nothing in Linux actually works as advertised. Without an in-house wiki, they'd be lost. "I tried to use this screen widget", "no, don't toucht that you fool!". It just goes on and on. It has cost them a fortune to use open source software.
But how would Microsoft and BT and their stooges ever get any money from that?
"Would the disastrous National Programme for IT, currently spending £12.7bn, work better with 120 separate project managers and specs?"
Yes, of course it bloody would! That would mean NPfIT had about one tenth the number of specs, and about one hundredth the number of project managers, that is does currently. This would undoubtedly cause the whole thing to go considerably more smoothly - so much so that we might even get around to smoothly abandoning the whole thing in 2013, a couple of years ahead of schedule.
Fire.
Cleansing Fire.
> 3/ Time and time and time again, projects are handed out to the same tired group of companies that consistently screw up and cost a pile of money.
Well they should have given the contract to BT then. All our Intranet stuff works hunky-dory, tickety-boo & all-systems-go!
What's that you say, 'BT are implementing the NHS IT upgrade?'
OK, Fail!
This is an excellent idea. Define standards (communications/data formats etc) and let companies complete to produce solutions that are sold to the individual trusts. The hopeless stuff will soon be weeded out and instead of a one-off monolith doomed to be out of date before deployed, then we could have a dynamic market. Government should pay for a reference solution for the actual formats and let the companies sort out the features to provide.
It just hurts that the Tories have come up with this!
I would have thought the Government could do worse than require all specs, docs and code to be published as open source code for all Govt projects then allow evolution to happen. We, the taxpayers, do own the code I presume.
"Smaller IT projects mean less risk of failure."
Sure, there are plenty of reasons why IT projects fail, but if you had to choose one way of making failure almost inevitable, it's planning a multi-year IT megaproject. After the technology choices (typically those from proprietary solution vendors sucking from the teat of the civil service) become obsolete after a couple of years and the nasty dilemma of either reworking everything or just ploughing on regardless is encountered - it's extra, unplanned work either way - deadlines will slip, features will fall away, people will jump overboard, cancellation is frequently inevitable.
So, I don't think the Tories are talking nonsense on this (admittedly rare) occasion.
Except that the last man will probably not know where the switch is because the switch project was separate to the light project.
Open source? Well, I've no objection there. I use it personally and it has even been creeping into some of the works kit. My biggest problem, one which some folk have touched on here, is that the whole government (sic) of IT is being done using rhetoric, politics and sheer flim-flam. The question is, inevitably, what are you going to use it for? The whole history of computing under various governments has been littered with the result of poorly conceived policies and ideas at which there has been much money and effort flung into what turns out to be an embarrassment for whichever party was in power.
Instead of throwing spin at us, therefore, why not give us a concrete plan? What do you want to do, what are the options and what are the best ways of achieving that goal. This would serve the country far better than promises about how you intend to set up projects and what source you are likely to use.
Requirements change as the system is developed. A large part of that is unavoidable because some many problems are impossible to understand until you invest a lot of time into solving them.
People who want software development to proceed along a well defined path ... follow that path to failure.
---
Also there needs to be a healthy feedback between the available technology and the design. Design choices influence technology choices. Technology choices then make some things easy to implement others very difficult. The design them is shifted to leverage the unique characteristics of the chosen technology.
Anyway breaking things down into small loosely coupled projects ... makes a lot of sense. Some will fail, some will exceed expectations. Even if as some say it will be subdivisions of the same companies .... those subdivisions will be competing against each other.
Also when you break things up into smaller parts ... well it suddenly becomes more obvious when certain features/capabilities cost far more than they are worth.
Here's how you cut up NPfIT into 100m chunks:
1) Common data format. Find one.
2) Write or procure sw to use #1
3) Write or procure internal exchange processes within the same area
4) Create a secure VPN for NHS use
5) Extend #3 along #4.
Asus have demonstrated that a lot of people will cheerfully use Linux and Open Office.
They also, alas, demonstrate how to get things wrong. Despite a physical UK keyboard, they don't supply a UK dictionary.
The Linux enthusiasts often criticise Xandros,and they have reason to criticise the Linux supplied with the Eee.
Guess what: Microsoft get things wrong too.
Open Source is working.
Kobayashi Gimp wrote :
"How about a novela pproach that is extremely daring........
Get the requirements right before chosing a partner and technology!!!!!!"
Yep - because the world stands still right?
The contract will be satisfied but the poor tax payer will be left with a system that can only solved last years problems f$%kTARD!!!!
As far as the government goes, opposition or not, that is about the extent to which they understand the concept.
One thing that seems to be missing from the spending to the retraining of all the people involved in maintaining the systems that are going to be introduced through a plan like this, on top of that what about the users, do you think they are going to be happy trying to work out whether to click the gecko, penguin or foot to do something.
The simple truth is that open source will have to overcome some hurdles before it can get close to being used on any major government project, the first is training; get it in the schools first, the second is familiarity, make a UI that looks similar to Windows, but not too close to infringe any patents, and a third is a simple one, make it work the way that people are used to, no installation from a command line, that is so 1990; get rid of the shell, provide drivers with hardware and software on CD/DVD. If these basic things can be achieved then an OS desktop could be used en mass within the next 10 to 15 years, without them power usage of Linux et al will remain the pervue of server admins, those with an interest and geeks. Joe public will just go "Meh" and pass it by.
I have not commented on the track record of the government IT projects as that is very well documented for prosperity
Seems they have listened to people who know how to deliver projects on time and on budget.
-Project size, complexity and risk correlations were known about in the 80's but no one bothered to read the research as it didn't look good on the CV.
- Why pay M$ when you can get what you need for free - and the support is better too. If you need features adding, a small amount of money spent on it works wonders - eg Amazon & HTML::Mason
.....there needs to be a raised standard in technical support and project staff. This is the problem. When a large number of support staff (I used to work for EDS (cough cough)) work on MS products, many of them wouldn't know what hit them with Linux and other Open Source stuff.
I really can't be bothered over who is best Windows or *nux. But I think everyone will agree that *nix is a much more technical. The question is where will the likes of EDS, Siemens, Crap gemini get these staff from?
We have MCSE on £20k a year, but try and find a RHCE who's willing to work on less then £30k. Then, multiply that by 100's of staff on large Govn projects.
Sorry, but MS wins, afterall the Govn. probably only pays a few £ per license (The DWP had over 30,000 MS W2k Licenses not being used in 2005 and never realised it.)
with Ted Dziuba's latest and the truly epic war raging on its comments page
http://www.theregister.co.uk/2009/01/26/dziuba_linux_desktop/
What a FARCE!!!!!!!!!!!!!
These software projects go over budget and fail. REGARDLESS of the software costs. The software costs are normally insignificant to the cost of consultants, and the staff etc.
1) Spec your requirements correctly
2) use appropriate tools and software
3) buy in expertise where necessary
4) deliver the requirements
or the govement way
1) be vauge about requirements but promise the earth
2) be vague about the budget and end up spending shit loads more
3) buy in expertise from everywhere even if its not relevant and hope they agree that its all someone elses fault
4) retire
5) or get promoted to another degree of incompetance
6) or point the fingure at one of the large consultancies you contracted on really vague terms
7) get laughed at by their solicitors who are much better than civil servants
8) end up explaining to a select committee where the fck all this tax payers money is going
9) wish you had never started
hire him. Maybe get Joel Spolsky in as a consultant, though the coders might rather have Aeron chairs.
On OSS vs MS: use the best tool for the job. The government shouldn't be mandating technologies or solutions before the project begins.
As somebody who likes open source software, I think this is brilliant news but I have some concerns:
* The problem with government projects, it that they change the project mandiate as the project is developing.
* Government don't seem to understand what they are asking for and see to make it up as they go along.
* Contracts and suppliers don't really care how well it works or don't understand the project (refer to point one) - EDS are an example of this.
* Interlinking projects - these BIG projects do tend to join/link to other projects and they contractors don't always fully understand the entire system.
I don't have a problem with open source (I prefer it to other commercial software products from the likes of Microsoft but, I have some issues which using open source without understanding some key points (these are the ones I can think of).
* Open source software is normally started by somebody which either wants to make a program to fix a problem or to improve an existing idea.
* As the software is maintained by these people in their spare time, there is no support offered from the original author unless that software is made or maintained by a distro or support company and if this is the case support is available.
* Interoperability between open source and commercial products can be patchy (look at Samba and Windows Vista - I remember listening to a FOSS radio station on how Microsoft worked to break computability between Samba and Windows Vista - not sure if it's true or not as I don't use Vista but it wouldn't surprise me).
* Support staff - you need the support staff which understand the system and problems (you need with all systems but Linux for example tends to need a lot of support from people who understand the issue - not sure this is very understandable. A lot of Windows admins seem to be scared for Linux/Open source (not sure if this is fear or change or what though).
* Project managers and managers - you need managers and project management whome have some technical understanding of the problems not just button pushers - the managers I have come across don't seem to understand open source and seem to have little to no technical ability, they also push the problem onto the technical staff rather than understand what the technical people are actually telling them.
I personally think that open source is a great idea for the long term but is difficult to achieve in the short term due to these problems.
I think if this was done we would have cheaper, more robust and more reliable services. We would also move to a more internally run and better maintained services (this is something you don't really get with Windows services in my opinion).
Mike
Hmm, something's wrong with this picture, it, it... sounds thoroughly sensible, almost like they... they know what they're talking about. How deeply strange, it must be a fluke.
If I could trust the awful desperate pro-"life" c***s as far as I could throw them I'd be impressed.
I doubt that the control management to make things work will exist in a cost cutting environment.
And fear that the end result will be:
- rather than chasing one organisation for its shortcomings some quango or committee meeting quarterly in the House will have to chase 120! organisations all saying "Not me guv! S'one ov them mate"
In an ideal world there would be no differences between appointing one large commitment or 120 smaller commitments.
The truth is (as we all know?) that 120 commitments in software without good project management is going to go extinct as a dinosaur but only rather swifter.
" It just goes on and on. It has cost them a fortune to use open source software"
It has cost them a fortune to use THE WRONG open source software, or maybe it has cost them a fortune to underestimate the cost of using this particular piece of open source software. It doesn't seem to have hurt companies like Google and Amazon.
Of course your friends bitch about work, who doesn't bitch about work? Do these things never happen with commercial software? I remember the YEARS of problems trying to get CA Unicenter running at my NHS trust - the big difference being it was proprietary software so we couldn't sack CA and hire someone else to work on it.
Anyway, I hope your friends are contributing patches, maybe that software will get better over time eh?
I think it is rather naive to think that using open source and cutting projects up into smaller projects will automagically fix all problems.
First of all, most of these projects are bespoke work and there won't be any 'existing' open- or closed-source solutions available.
Second of all, even though smaller projects make managing complexity easier, it won't be easy getting a working, complete solution on the rails that should have a large scope (I have seen systems that were built piecemeal by one company and even then there a problems with data-consistency and -compatibility between various parts and/or systems).
And what exactly is wrong with OS/2?
There are lots of large enterprises still using it because it 'just keeps on working', which is more than can be said for the windows crap I encounter in my work.
We went the other way, and if there's a good reason for increasing the cost of internet services per workstation from about $5 to $20 per week I don't see it.
We moved from a Mozilla based system to an Exchange system, lost features, decreased security and increased cost.
Simple things like auto-completing from your online email address directories aren't possible with Outlook or an Exchange system. And just because it comes with a built-in calendar doesn't make it better. There are plenty of online calendars that are open source and work just as well as Outlook. We used one from Oracle, which integrated into our email system perfectly.
This example isn't true of all applications of course. But there is no reason to pay for software if your main applications are internet based or along the lines of word processing, spreadsheet and database applications.
Support is no more expensive for open source applications. They're not more difficult to maintain, and given the reliance that almost every business puts on community based support forums, there's actually not a lot of difference in support. At least none that has anything to do with the software costing money or not.
Autodesk for example, relies extensively on it's own community to provide support. That's not even a criticism, you usually find the people that know the most about a piece of software and how to get something out of it are the everyday users of that software. In the 8 years I've been supporting Autodesk products, I can remember only 2 occasions when I needed support that went beyond their forums and actually required their own people to come visit our offices.
Given the $50-60 thousand dollars a year we pay in maintenance, there just might be a saving somewhere if an open source product was available that does the same thing. Most of that maintenance cost comes in the form of annual upgrades to the latest versions of their software, not from direct customer support. Open source software doesn't come with an upgrade cost, except that of implementation (something that all software shares).
So I would say that if you are using nothing but Microsoft Office applications in addition to the usual internet services you'd expect in a modern office or school, then there is no reason why you should pay more for something that isn't open source.
Fine if you feel more comfortable running Windows, go ahead. But don't waste money of things like Office 2007 when Open Office does exactly the same thing for free.
...make every computer in government run Linux and even the simplest tasks will become so complex and time consuming there won't be any time left over for any Orwellian projects.
To answer a couple of posts on penalty clauses, for large government projects they simply don't work. This is almost invariably down to the fact that the people speccing the system have no idea what they actually want from their system and thus underspec it, they then constantly add new bits to the project. Then when the project is late the supplier points to the spec, points to all the extra emails saying "Please add x,y and z" and passes the buck right back to the customer.
If you want to have a flexible spec then you need to start off the contract with that in mind, you have milestone payments for specific deliverables and make sure that everything that's added is properly costed a part of the project. Sadly, few companies seem to have grasped this idea.
It makes sense.
Firm & tidy spec, firm & tidy response, firm & tidy result, firm & tidy invoice.
Sloppy & shabby spec, Sloppy & shabby response, Sloppy & shabby result, Sloppy & shabby invoice?
It seem to cascade from the original spec/managers all the way down.
All of these are good suggestions that will help. But as someone who has helped write government software, and bailed when they knew what was good for them, the biggest problem of writing software is the laws that it is coded around! The law is in inconceivably complex beast that isn't even understood when it is passed, and takes successive rheams of judges and court cases to determine its interpretation. And each successive government changes it every four years! In the law there can be no basis for software, as the latter requires strict and rigid interpretation of *very* simple rules in order to execute.
Software as it is today is inherently incompatible with law. Or should I say law is inherently incompatible with software, take your pick.
Given how poorly managed government IT projects are, I honestly don't think going FOSS will help. I mean, sure they'll cost slightly less, but they still probably won't work at the end. The contracts will STILL go to the lowest bidder, and there still won't be any consequences for failing to deliver, or running over budget or over deadline.
Open Source worked well for operating systems and single applications because the scope was limited and function required was specific. A corporate application is a multi-faceted entity with wide-ranging interfaces and data structures required. (Remember the last time you sought consensus internally on some requirements, and then imagine including the cleaner and the tea lady as well).
Sorry, but the "bazaar" approach does not apply here. This is just pollies throwing around buzzwords to sound "with it".
