Wednesday, October 26, 2011

I'll see you your 50 year storm ...

... and raise you a Perfect Storm says Mike Bracken.  I do like the continuity of analogy around storms.  GDS is certainly operating within a favourable environment with a wide range of factors aligned.  Now is the time.

He goes on to say:
as one of our new recruits Mike Beaven, who is running our transformation programme put it more succinctly: “How often do you get the chance to digitise a G7 economy?”
The answer to that depends on whether this effort succeeds, building on the progress made by all previous efforts.  If it does, the team may get hired to do the other 6.  If not, it might be "How often do you get to balkanise the digitals?"

Friday, October 21, 2011

Government IT - 180 Degree Turn - Chris Chant


This is the text of a (great) speech that Chris Chant gave at yesterday's Tea Camp:

First of all to acknowledge a small and important bunch of people who have been delivering some great stuff in gov IT And those who have been working their socks off but through no fault of theirs ON THE WRONG THING - it's tough being in IT. 
People now can be using the fruits of your work EVERY DAY. There aren't many things like that - electricity maybe. 
But we need to face  some  UNAVOIDABLE TRUTHS head on 
The vast majority of Gov IT in todays market is outrageously expensive, ridiculously slow,  poor quality and most unforgivably rarely user centric in any meaningful way 
Let's start with a personal view of the unacceptable. 
It is unacceptable, not to know the cost of a service and the real exit costs - commercial, technical and from a business de- integration standpoint. 
It is unacceptable now to enter into contracts for longer than a year.......
And to those who say "what about supplier upfront  infrastructure costs", I say ask shops how they do it, ask small garages how they cover the costs of hydraulic hoists and computers 
It is unacceptable, not to know how many staff we have working on IT 
It is unacceptable, not to know what all those staff do. 
It is unacceptable, not to know what systems we own, how much they cost and how much or even IF they are used 
It is unacceptable, not to know when users give up on an online service 
And it is unacceptable, not to know why 
Of course It is unacceptable that they do! 
It is unacceptable to have a successful online service and then to remind customers to use it with a postal mail shot 
It is unacceptable not to be able to communicate with customers securely electronically  
It is unacceptable, not to be able to do our work from any device we choose. 
It is unacceptable to pay up to £3500 per person per year for a desktop 
It is unacceptable for your corporate desktop to take 10 mins to boot up in the morning and the same to shut down 
It is unacceptable for staff to be unable to access Twitter or YouTube or not able to access the online service they are supporting in a call centre 
It's unacceptable in this day an age, to ensure people are "working" by restricting access to the Internet. 
It is unacceptable, that 80% of Gov IT is controlled by 5 corporations. It is unacceptable, to outsource your IT strategy
It is unacceptable to see the cost of changing 1 word/colour or 1 line of code as £50k  
It is unacceptable to wait 12 weeks to get a server commissioned  
ABOVE ALL,  
It is unacceptable, not to engage directly with the most agile,  forward thinking suppliers, the SME market.

Things have changed and we haven't, until now 
SO HOW DOES G-CLOUD HELP 
Cloud will be cheaper - by the time you factor in the time spent now procuring and accrediting individual solutions,  
Using cloud solutions that have already been secured and accredited will be cheaper almost always.  
We will only pay for what we use, even models emerging for DR 
Over time, for most, products will be pre-procured and security accredited 
We will know from the outset the cost of the product and importantly the cost of exit 
Contracts will be under a year and the business impact of exit will be visible 
Price and service performance will be visible to all. Which will drive a provider to low cost high performance products 
We will not get ourselves locked in, in any way 
We will make fewer changes to the standard configuration of our desktops ( for as long as we provide them) so they will be more reliable, faster and cost less 
82% saving in GDS over corporate systems, includes great kit and IL3 access when needed. 
We will be able to migrate quickly to pre procured products that we need. Either as our needs change or if we are dissatisfied with service or price 
Our staff, over time, will become our system integrators until services in this space mature at the right price. What they won't be doing is spending years on procurement and security work that has been done by others a hundred times before. 
People will be setting up some services in minutes instead of years 
FDPs and cross cutters don't have all the answers now must iterate and iterate delivery, policy AND  process - will this approach work - I believe it will - certainly the Herculean effort spending 2 years working out requirements, another year in a procurement process and 3 years delivering some massive bespoke system hasn't covered us in glory 
The first manifestations of this total transformation of the way Government uses technology will come in the next couple of days. 
We will publish a strategic implementation plan and OJEU which will together demonstrate solid commitment to change the way we work. 
But cloud is very much applicable to restricted and confidential. Services already being developed 
Cloud is PAYG, is elastic, is on demand, put down as fast as you pick up 
Not bound by old rules and ways of working 
Not lock ourselves in business or commercial 
Aggregation of demand is not the only way to drive down cost 
Commerce doesn't only work that way - M &S 
IT service must start around user need not around outdated and ill applied security concerns 
CESG know this local practitioners often don't  
We start from "this is what the user needs", then comes the system (at the right price), then comes the security wrapper to make it appropriately safe 
And that, for many,  is a 180* turn - and it's the way all Gov IT must be delivered from now on.


(any transcription or editing errors are mine)

Sunday, October 16, 2011

Jobs For Life?


Ways To The Cloud

The Government's Cloud Strategy document will apparently shortly be issued.  It will move on from the previously published wider ICT strategy (which, itself, will be enhanced by an Implementation Plan) and elaborate how UK government will take advantage of cloud technologies.  Or, perhaps more likely, it will outline the "what" and leave the "how" to the next iteration of the document. With the gCloud procurement about to get underway - any day now I am told - here is my view of the "how" for cloud in government:

1) Starting now, UK government should never again purchase hardware for development or test environments; any such need will be met from pure public cloud "tin for rent" IaaS or from reuse of community/private cloud capability (this will inevitable require applications to be architected in a way that will make them cloud ready).

2) From today, any refresh of e-mail systems or deployment of new e-mail capability should be through public cloud or from usage-based private / community / shared within government clouds.  Alongside this, no department should use a collaboration capability that isn't in the public or private cloud (and, in the latter case, already in use by other departments)

3) Building on the existing "Foundation Development Partners", one department should take the lead in putting a single end to end process (rather than just a specific application) into the cloud.  A potential candidate could be case management - there are dozens, probably hundreds, of such processes in government, each implemented differently and with varying technology.  The FDP will eat its own dog food (get wet under its own cloud?) and move its own processes so that they are supported by this cloud - suggesting an early candidate could be from within the Ministry of Justice - and make all of it available for other departments too (with the aim that all such processes are supported from a common, or small number of consistent, cloud solutions within 3 years).

4) One department or public body should volunteer to make itself 80% cloud supported within 2 years.  Candidates would include departments where there is little transactional or clearly secure capability required - which might mean DCMS or DCLG.

That should get things off to an interesting start.  By doing these four things

  • Any department starting a new project will test out cloud (and will architect for future deployment to the cloud by default)
  • Anyone looking to refresh email (and given most departments are likely still on Office 2007 or earlier, that could be all of them) will take advantage of cloud-based mail so lowering their costs, allowing a wider range of mobile devices to connect to their e-mail (again lowering costs) and improving support for remote working
  • An end to end process will be tested in the cloud, for at least one department with relatively complex needs, so allowing government to understand the changes in business process, in operational rules and in the software procure / deploy process
  • One department will become the vanguard that all others can aspire to follow (or shy away in fear from)
These moves would not be without their troubles and challenges but they would allow the testing out of the benefits that cloud promises.

Saturday, October 08, 2011

Combat Stress - Royal Parks Half Marathon

Tomorrow I will run the Royal Parks Half Marathon in support of the charity "Combat Stress"

I got involved with Combat Stress after a chance meeting with one of their patrons.

I got involved with marathon running by chance too, after moving near to a wide open park in Paris that just seemed made for running.  Serendipity.

I haven't sent out e-mails seeking sponsorship for this run, I haven't rattled a tin at the local tube station and I haven't set up a JustGiving page (I have done all of those in the past and remain grateful to those who were kind enough to contribute).

I appreciate that the financial pressures on everyone are extreme and that many will already have a charity that is near and dear and which they choose to support whenever they can.  I will make my own donation to Combat Stress.

If, though, you are looking for a charity to support, let me tell you why I choose to raise money for Combat Stress:

I have only a little affiliation with the military.  Relatives have been involved - one died in the Great War, another flew 'planes from Ark Royal as a Squadron Leader, one was an Air Commodore in the Reserves - but I have no personal experience.

But I can see a problem looming

That problem is increasingly likely to mature within the next 10 years.  There is little data on how troops are affected after encountering combat, the severe injury or death of their comrades and friends, the separation from their own loved ones and the return to "normal life" - a life that is very different from what they have just been through.  Some adjust.  Some don't.  The trouble is that the "don'ts" can take years to show the signs.  The most often quoted figures suggest that some 7% of returning soldiers will show the symptoms of PTSD but not until, on average, some 14 years after they have left the army.

7% is too low and 14 years is far longer than it really takes

Combat is more intense and on multiple successive fronts; a relatively small number of people have military experience (when compared with conflicts 50 years ago or more from which most of the data is sourced); information on wins / losses / injuries / deaths circulates far faster (both to the troops and to those at home); and now, troop counts (with troops being "made redundant" - as if - almost upon their return) are reducing increasing the number of former soldiers who will be looking for something else to do.

All of these points and a few more led me, after researching the available data, to conclude that there is a problem coming, probably some time in the second term of this government (assuming that they win the next election).

Help for Heroes does an amazing job of dealing with the troops that come home injured, Combat Stress deals with those who return home apparently normal but, some time, later are plainly not.  There is a bigger category of troops - a General I know calls them "the rest" in a fatherly sort of way - who will never seek help but who end up alcohol or drugs-dependent, in jail, homeless, or beat up people in bars.  Not all troops will end up in any of these categories of course, not even a majority.  But a substantial enough number will.  Combat Stress will be there to help as many as they can.  And that "as many as they can" is determined, largely, by donations.

So I like to raise money and give my time to Combat Stress.  If you feel the same, you can donate money right from the Combat Stress website avoiding the commission taken by other sites.  You can donate by 'phone, by text, or using an online form.


Friday, October 07, 2011

AlphaBeta.gov ...

Is what Alpha.gov.uk (and the soon to be beta.gov.uk) are doing really different? I mean, it's a website, right?  We've had a few of those before, more than a few (thousand) in fact.  As someone who used to do, in the Office of the e-Envoy, what the Government Digital Service (GDS) team (formerly the alphagov team) are doing, I watch their progress with interest, applauding the successes and wondering about the things still to do, occasionally scratching my head.



I was part of a team a lot like the GDS team.  We had a bunch of hugely talented people drawn from inside and outside of government (plus quite a few secondees from major technology companies too); we published a lot of what we were doing; we took on some difficult projects (some right at the very edge of what was going on in industry let alone in government); we worked hard to draw other departments in (we often had to pitch for them to use what we were doing during a competitive tender process;  we used open source products; we did work for other departments who commissioned us to build things; we put in place things that, 10 years on, are still in place now and drawing ever greater usage; we built things from scratch, used beta products, and integrated some complex stacks; we even used the first cloud service in government in 2001 when we paid by usage and didn't own any assets; we worked across government to write and edit content; we dealt with accessibility from day one with the aim of creating the most accessible sites; we tried to lead the way in doing things in a different way; we, too, dealt with the long tail of the vast range of services that government offers; we did things, like MessageLabs anti-virus, that weren't strictly in our remit but felt like the right thing to do; we lived and breathed our services, taking it personally if there were problems or if customer were not happy.
eDt vision, September 2001
So are GDS all that different? My sense is that they are, and they aren't.  They're doing what we did, and more, and better.  And they have some things to work on - based on current progress, I'm pretty confident that they'll sort those out too.

The team - there is a greater focus on people already inside government.  In 2001 we tried hard to recruit civil servants into the team but often failed to attract many (or, indeed, any).  The Cabinet Office wasn't seen as the place to do what we were trying to do; it wasn't a delivery department.  We had some great civil servants for sure, but not enough to make them the core of the team.  The GDS folks, and the Alpha team in particular, have attracted some terrific people into the team, even converting some outsiders into insiders along the way.

The process - this is open government in action.  From 2000-2005 we published almost everything we did, but almost always to those inside government.  We were amongst the first (and probably actually the first) to make available every detail of the volumes, the service delivered, the product roadmap and so on; the team often spoke at conferences but the audiences were mostly other government people (and the larger suppliers). GDS are publishing pretty much everything they do, including decompositions of decisions made (particularly the ones that have attracted criticism), via blogs, twitter, the press and doubtless other routes.

The roadmap - I get the sense, and I don't know this for sure, that the alpha team don't know where they're heading with each piece of work they start, and I mean that in a good way.  When we started work we had to bid for money, build requirements list, test the requirements with customers and deliver what we said we were going to deliver.  We wiggled a little as we went but we pretty much got to where we had said we were going to, for better or worse.  These guys are working in a much more iterative way, shifting left and right to work around some problems or to add in new capability.  The beta will, I hope, evidence that approach even more.

Things still to see

Alpha tied itself closely to politics - the PM and Deputy PM were pictured smiling on the front page.  The GDS blog has a banner picture of Francis Maude:


If we'd had any obvious political link there would have been a lot of trouble.  Things change, of course, but my view is that these two shouldn't mix (and I don't mean the PM and Deputy PM).   The public ministerial support is, though, a brilliant thing to have in your pocket.

A single great website is, of course, essential for government (I think my position on that has long since been clear).  27 million people a month visit the existing direct.gov.uk but it has remained largely untouched since I left it in 2004/5 (it's had a technology refresh and significantly increased functionality, but it still looks and feels largely the same - that is meant as no slight to the team who have worked diligently to upgrade it and keep it fresh, just an observation that it's still orange and looks much the same).    For instance, the screen shot on the left is from 2007, the one on the right from 2011.


But transactions are where it's at.  We built the government gateway to support both single transactions and joined up transactions (that is, ones that would link more than one department and send each of them the relevant data.  The vast bulk of those transactions come from sites other than direct.gov; the vast bulk of direct.gov's transactions are for Car Tax.  That all needs to change.  We didn't ever crack it.  It needs cracking - integrated transactions, simple pan-device (and even pan-channel) authentication, pre-population of data that government already knows about you and so on.

We built tools and services for all of government to use.  Our aim was to massively reduce spend on technology by building something once that could be used by many.  It was a great aim then and it's still a great aim.  I don't yet see where the GDS team are going with this.  Folks in government are still building and rebuilding websites - I know of a dozen that have gone live this year alone and that's only the ones that I know about.  As Simon Dickson has argued before, there's a great case for a lightly centralised service supporting government entities that still need their own site. I say "lightly" because I think there's a need for two or three competitive offers working from consistent templates with each innovating and improving as well as co-operating and sharing.

We regularly confronted the question of whether to build something ourselves or to use existing products.  With the gateway, we used a not-yet-released to market Biztalk which needed around 100,000 lines of custom code to get it to do what we wanted.  Microsoft took all that we had done (under licence) and released a new version a year or so later which did all that we needed and left us with only 8 lines of custom code.  When looking at the engine for direct.gov (and later dh.gov and several other sites), we looked at Vignette, Stellent and others but went with a custom build - that was a difficult decision and one that proved wrong in the end as the engine was subsequently scrapped and replaced by Stellent.  GDS look to be going custom again which is likely to prove interesting perhaps 2 or 3 years from now.

We also confronted the issue of how much to do in-house and how much to outsource.  We were lucky to have a great skunkworks team who could rapidly build proofs of concepts, but we went with suppliers for all of our production versions (and we hosted with them too, not having a real cloud supplier at the time - plus we were also tied in knots by the Critical National Infrastructure badge, one that we wore with pride at the time but one that, in hindsight, was plainly an albatross).  Keeping everything in-house, managing turnover of staff, staying on top of emerging developments, dealing with 24 hour support, knowing that the only people in the way of criticism is your own team etc is a tough place to be - we were there too, but we had dozens of supplier people working for us too, so at least the load was spread.

In the end, GDS will make their choices on each of these and a million more decisions.  I hope that they recognise that, whilst it may all feel new/different/special, some of what they are doing has been done before (not just during the time of the Office of the e-Envoy around but by people in the 25 years before, dating all the way back to the computerisation of major transactions like PAYE).  Lots of lessons were learned and lots more will be learned from here on.  It would be good if the ones learned again aren't the same as the ones learned before.

We, too, made decisions that attracted flak from armchair commentators, professionals and the press.  Perhaps the largest flak resulted from choosing Microsoft to deliver the gateway (when the previous attempt had failed dismally).  Microsoft were the only company willing to step into the fray - others suggested we scale down our ambitions and make use, for instance, of Windows NT and e-mail.  That decision proved to be the right one, despite the seemingly endless flak,

Alpha and all that follows it needs to part of a long-term delivery plan, one that shows how the Government's ICT strategy will be implemented (a document describing that is due out any day now I believe).  But it also needs to be part of a wider delivery plan that shows what will happen within and across government where the real world services will be challenged in the same way as the technology ones are being (and have been).  Technology can create a veneer of joined up capability but it can't truly transform the services in and of itself.

I will keep watching with interest because the challenge they're taking on is the right one to undertake and because I believe in it too.  This is a new time in government.  Maybe it's even a new 50 year storm.

Sunday, October 02, 2011

Oracle's Clock Cleaned??

Marc Andreesen, erstwhile co-founder of Netscape and now uber-VC, says "The Clock Is Ticking On Oracle." He explains:

Ten years ago, it was a joke: you'd raise $20 million in venture capital and write a $4 or $5 million check to Oracle, Sun, BEA, and EMC....When it started, Salesforce looked like a toy compared with Siebel. Look ahead five years later, it's obviously better. Not a single one of our startups uses Oracle.
And I suspect he was using Oracle only as an example of many companies that need to execute a pivot if they're going to be in the same shape 10 years from now that they're in today.


In the last year I've seen plenty of examples of companies who, in the past, would have spent £250k or more on a technology solution and can now spend 20-40% of that.   If you're a big company - HP, IBM or similar - and you charged a client millions for a solution last year that this year they think they can get for hundreds or even tens of thousands, your revenue is going to suffer  (not to mention profits, stock price and individual bonuses).  


Those big companies are going to have to do some pretty incredible gymnastics to execute this pivot and keep everyone happy.  Some, of course, will choose to ignore the potential consequences and will try and ride the changes out hoping that none of their customers will notice what other companies are up to.  


Only a few will do what Intel did in 1983 when it got out of memory chips and focused on microprocessors.  Many large IT companies will soon face their Intel moment.  It will be interesting to watch how they react.

The Cloud Defined (by Apple)

This coming week Apple will properly launch iCloud - and, when they do, over 200 million people (and climbing fast) and the devices plumber into their hands will suddenly become aware of what the cloud is.  Or, perhaps more accurately, they will associate the cloud with Apple and what Apple's cloud does.

Apple have for some time been triggering ads to any search with the word "cloud" in it:


Up until this point any ad featuring a pitch for cloud has used "soft and fluffy" (obviously) imagery, such as the one below (from Microsoft - but I could have picked imagery from any of a dozen or more major vendors).

It's been interesting, I think, to see an IT solution get associated with sunny days, blue skies and a sense of uplifting vision - that didn't happen for object oriented programming, open source code, relational databases or even utility computing.  And that might have been fine when no one had stepped forward to say, actually, "this is what I mean when I say 'cloud'."  Indeed, I've occasionally used slides in the last couple of years with pictures like the one below and the sub-heading "99% vapour, 1% grit that gets in your eye."

iCloud, along with Amazon's equivalent services for the Kindle range, will define cloud just as Apple defined what a smart phone was with the iPhone and what a tablet is with the iPad (let alone music players, laptops and whatever else). iCloud will be myCloud.

For the first time, for hundreds of millions of people who have stared at the posters along the Interstate or the M4 wondering what on earth the IT world is going on about, cloud will mean something. They will be given a way of having everything in sync, all the time, on all devices.  Where "everything" is contacts, mail, calendars, game state, videos, music and so on.  It will also give them something that apparently costs nothing (only a few users with very large storage needs will end up paying), further complicating the job of those who want to make money from clouds. And that will be what cloud means.

My guess is that others wishing to use the word "cloud" are going to have a battle on their hands to differentiate their cloud from the Two A's Clouds.  "Cloud Computing", which has always been hard to define, will mean something else - and so the marketing people will work overtime to create differentiated brands that try, in 2 or 3 words, to mark out what their particular cloud does.  As a result "Cloud Computing" will become a redundant term and imagery will shift away from the kind shown above.  That can only be a good thing - it should result in less effort being expended on cloud as some kind of nirvana and more on what a business can do with it.