Friday, 30 December 2016

@gdsteam & HMG's digital transformation strategy


Sir Jeremy Heywood is the Cabinet Secretary and the Head of the UK's Home Civil Service ... He published his review of calendar 2015 in a series of 54 tweets between 23 December 2015 and 3 January 2016 ...
That's what we wrote on 24 January 2016.

No such review of 2016 is being tweeted at the moment by Sir Jeremy.

Perhaps nothing much has happened this year.

We noted on 8 February 2016 that Sir Jeremy has linked his fate to the fate of the Government Digital Service (GDS): "Sir Jeremy can't keep away from the importance of the digital transformation of government by GDS".

GDS were given a budget of £450 million in the Spending review and autumn statement 2015. That was 13 months ago on 27 November 2015. GDS have never explained what they're going to do with the money. In that sense perhaps nothing much has indeed happened this year.

The department that Sir Jeremy depends on to transform government digitally has operated for more than a year with no published strategy. And the published strategy before that had no academic support.

GDS keep promising to publish their new £450 million strategy. As late as 14 December 2016 ex-Goldman Sachs man Kevin Cunnington, GDS's director general, was telling Civil Service World magazine that the strategy would be published before Christmas: "A new strategy is due out before Christmas which will set out our priorities for digital and a really ambitious transformation agenda for government".

Two days later, 16 December 2016, Computer Weekly magazine were telling us Government delays release of digital transformation strategy until new year. That's not how a smoothly-run publicity machine operates.

The director general says the document will be published before Christmas. Two days later, egg on face, he turns out to be wrong. It's unsettling to see such an about turn.

A GDS strategy document does exist. Derek du Preez, our favourite banshee, told us so on 12 December 2016 in Leaked Government Transformation Strategy leaves lots to the imagination. Why not publish it?

There has to be a good reason to endure the embarrassment of saying you're going to publish a document that's already been divulged to journalists and then not doing so. What is that good reason?

Not just one. There are several good reasons you might suggest.

Before considering those good reasons, note that the delay in publication of the GDS strategy could be a good sign. It could indicate that responsible mandarins are finally looking at GDS's actual skills and not the supposed skills puffed in GDS's PR – reality could be getting a look-in:
  1. Transforming government requires original thought. It requires innovative imagination. There has been no sign of that at GDS in the five years of their existence. GDS always say that it's no good just changing the front end of government services, public administration needs to be thoroughly re-engineered. Then they change the front end and leave it at that. That is the opinion of GDS's first deputy director, Tom Loosemore, now at the Co-op:


  2. Edgy? Revolutionary? Restlessly and tirelessly in search of solutions? No. Slaves to fashion, bound by convention, GDS has already retreated into the comfort of process. They keep doing the same thing. Even when it doesn't work.

  3. Transforming government requires considered experience of public administration and GDS don't have it. As Stephen Foreshew-Cain said when he was briefly executive director of GDS, it's the other departments and their suppliers who "understand their users and services better than we ever will ... They know the policy, the intent of that policy, and the legislation that sits behind it ... They know their users better than anyone. They are by far the best people to meet those user needs".


  4. Transforming government requires some ability to work professionally with the departments of state, their agencies and local government. GDS got off to a bad start with the Electoral Commission, who blamed them for causing delays to the 2012 confirmation pilot for individual electoral registration. Relations broke down with DEFRA's Rural Payments Agency, please see Government Digital Service “hindered delivery” of rural payments programme, Public Accounts Committee says. GDS have been witheringly dismissive of local government for years. That has changed now that GDS find themselves in need of local government. Mike Bracken, GDS's first executive director, told the Americans three years ago that his job was not to collaborate with Whitehall but to route round it. As late as 5 July 2016, he was still saying of Whitehall that it is set up for nothing more than "an intellectual pissing match". Stephen Foreshew-Cain, Mr Bracken's successor at GDS, said that he recognised the need for collaboration but promptly accused other government departments of "decades of inaction and inertia". There are bridges to mend.

  5. Transforming government is held by GDS to involve getting rid of the established "oligopoly" of suppliers (Capgemini, HP, BT, IBM, Fujitsu, Atos, CSC, Capita, ...) and replacing them with small and medium-sized enterprises (20.11.14) on short contracts (not that the SMEs always agree). GDS lack the data processing skills to design, build and maintain large complex computerised systems. Or even small simple ones. After five years, there is barely a dent in the armour of Capgemini, HP and the rest.

  6. Transforming government is held by GDS to involve getting rid of the established "silos" of data maintained by the separate departments of state and replacing them with "canonical registers" shared all across Whitehall to support government as a platform. In their Walter Mitty imagination, GDS would have "domain control for the register.gov.uk domain", i.e they would have control over a new single source of truth super-silo-of-all-the-silos, please see Smash the silos. The departments of state might in the circumstances be imprudent to abrogate their constitutional duty in this way.


  7. "Digital transformation" is held by GDS to mean putting public services on the internet: "digital means applying the culture, practices, processes and technologies of the internet era to respond to people’s raised expectations". The caravan has moved on since the internet's hippy innocence of 1995. In addition to the beneficial culture, practices, processes and technologies of the internet era, people now consider also fraud enabled by the internet, and espionage and surveillance and paedophilia and terrorism and pornography and the monetisation of personal information. But GDS are still in 1995, they promise unqualified security that they can't deliver and they promise privacy/confidentiality that ditto.


  8. The digital transformation of government requires universal identity assurance services to support transactions between people and public services. There was no progress in 2016, nothing happened, GDS's GOV.UK Verify (RIP) is still not up to it. No identity assurance, no digital government transformation.
There are more but that's enough good reasons to delay the publication of GDS's strategy for many a Christmas yet. It behoves us all to support Sir Jeremy and Whitehall's other mandarins if they have finally spotted that it is a delusion to suppose that GDS is the obvious centre for government transformation.


----------

Updated 2.1.17

Data at GDS (the Government Digital Service) is "a blog about the tools and techniques used by GDS for data analysis" and back in November GDS told us about how they use artificial intelligence to automate the assessment of user feedback, please see Understanding more from user feedback.

GDS use topic modelling: "In machine learning and natural language processing, a topic model is a type of statistical model for discovering the abstract 'topics' that occur in a collection of documents". A human being reading a piece of user feedback knows what it's about, it's obvious what the topic is. A computer program has to "abstract" it. How?

Answer according to GDS, by using techniques like "Latent Dirichlet Allocation (LDA), Latent Semantic Indexing (LSI) ... and Non-negative Matrix Factorization (NMF)". Using Kullback-Leibler divergence with LDA allows GDS to "find the statistically optimum number of topics" so that, if the words "find" and "contact" for example occur in a piece of user feedback, then "we can see that users are trying to complete the task of finding a way to contact a service".

"This approach can also be used", GDS conclude, "to tackle a range of text analysis challenges ... such as quickly understanding policy consultation responses".

That's quite a leap. One minute GDS are telling us how hard it is to work out algorithmically what a piece of text is about. Next minute we're supposed to believe that natural language processing could assess the merits of a tax expert's response to a consultation conducted by Her Majesty's Treasury.

How close are GDS getting to artificial intelligence that can grasp the semantics of documents written in natural language?

Just before Christmas GDS published Using machine learning to classify user comments on GOV.UK. They're looking at three features of user comments: "the ratio of upper case characters to total characters, the total number of characters entered in the text box, and the ratio of exclamation marks to the total number of characters".

So, not close. You may have been hoping for something sophisticated. Something transformative. In the event, in the name of data science, they're counting exclamation marks.


Updated 7.2.17

Privacy groups urge dropping entire Digital Economy Bill data clause – thus Neil Merrett yesterday, "read him early, read him often", as we always say.

His latest article isn't just about the ghastly Digital Economy Bill. It also covers GDS's dance-of-the-seven-veils national digital transformation IT strategy:
The Cabinet Office has said that the publication of the new GDS strategy was expected to be unveiled by Cabinet Office minister Ben Gummer this week.
All the top performers have left the GDS stage now. Ditto the senior members of the chorus.

There's almost no-one left at GDS to support Mr Gummer as he comes out from behind the curtain and, blinded by the footlights, makes his way to the front of the stage to entertain a packed house wearing nothing more than version 107f, or whatever, of the aforementioned national digital transformation IT strategy. There's ex-Goldman Sachs man Kevin Cunnington, Director General of GDS. And after that, no-one. They've all left.

Support could be provided by drafting in some of the GDS talent of yesteryear.

Maybe it would help to have some razzmatazz from the US or Australia.

GDS could call on the bottomless pool of talent at the Department for Culture Media and Sport, the Department for Work and Pensions and the Department for Business Energy and Industrial Strategy.

John Manzoni or Sir Jeremy Heywood could assist at the unveiling of the long-awaited strategy.

It's going to be lonely. Good luck, Mr Gummer.


Updated 20.2.17

Neil Merrett promised us the publication at last of the long-awaited government transformation strategy please see above and, lo, it finally came to pass a year late on 9 February 2017.

Next day, Computer Weekly magazine served up Government digital strategy ticks the boxes - but real transformation needs more radical ambition. That cool reaction was followed on St Valentine's Day by GDS, HMRC and Verify: so much for cross-government digital collaboration and on 15 February 2017 we got HMRC ID vs Gov.uk Verify [RIP] – what’s the difference, and why it matters.

"Building on the work we have already done", GDS say on pp.11-12, "our priorities for government up to 2020 are ... making better use of GOV.UK Verify [RIP] by working towards 25 million users by
2020 ...".

A fortnight before the publication of GDS's strategy, HMRC had already announced that they were proposing not to use GOV.UK Verify (RIP). Both DWP and the NHS have in the past expressed reservations about using GOV.UK Verify (RIP). Scotland has its own identity assurance scheme, the private sector has several and has no need of the under-performing GOV.UK Verify (RIP) and neither does local government.

The question arises therefore how on earth GDS could possibly achieve 25 million GOV.UK Verify (RIP) users in three years time.

They can't ...

... unless they cheat ...

... in connection with which, cast your mind back all the way to 1 February 2017 and the GDS blog post Growing Verify: services that need less proof of identity. The proposal there is to go back to "basic accounts", GOV.UK Verify (RIP) unverified accounts, self-certification, level of assurance 1 accounts.

Perhaps GDS could get 25 million people to self-certify. The basic accounts created would be of no use whatsoever to relying parties like DWP, the NHS, et al. But GDS would have achieved their strategic target – 25 million sort-of-users.


Updated St Patrick's Day 2017

Neither of them is employed by GDS any more but according to Tom Loosemore and Stephen Foreshew-Cain "digital means applying the culture, practices, processes and technologies of the internet era to respond to people’s raised expectations". GDS haven't advanced any other definition of the word "digital" and we may assume that they are still happy with it.

We have pointed out certain problems with that definition based on the culture of the internet era. The culture includes large dollops of pornography and fraud and it involves the mass destruction of any notion of privacy. GDS surely don't approve of that but they haven't yet distanced themselves from those aspects of the internet era by providing a new definition of "digital".

They probably should do. Today we learn that Gov.UK pulls plug on its YouTube ads amid extremism concerns. Inadvertently, Her Majesty's Government have been paying for advertisements to appear on extremist websites, thereby funding extremism. This discovery is all thanks to an investigation mounted by The Times newspaper, please see for example Taxpayers are funding extremism.

It's time for GDS to provide a serviceable definition of "digital".

No comments:

Post a Comment