Wednesday 4 May 2016

RIP IDA – the last rites

No need to say it, it goes without saying, it should be obvious to all but,
just in case it isn't obvious to all,
IDA is dead.

IDA, now known as "GOV.UK Verify (RIP)",
is the Cabinet Office Identity Assurance programme.


We can make a meal of it. Or we can do it the quick way.

Let's try the quick way first. Three steps.

Step one
The Government Digital Service (GDS) have published their ten design principles. These are treated with reverence. No deviation is allowed. Here's the first principle, the most important, the oldest rule in the book:


Step 2
GOV.UK Verify (RIP) was meant to go live last month, April 2016. The system has been in development for four years and it's been in test mode for two years. As late as 26 April 2016 the public were led to believe that GOV.UK Verify (RIP) was "on track" to go live on 29 April 2016.

Three days later, on 29 April 2016, GDS published an embarrassing "update" claiming that they were "nearly there". That is their way of acknowledging that they're still not there, and that GOV.UK Verify (RIP), like many "agile" systems, is in eternal test mode and will never be live.

Sir Jeremy Heywood, the Cabinet Secretary, one of the most powerful people in the country, took it upon himself yesterday to explain on Twitter that GOV.UK Verify (RIP) is "hugely important" for the government ...

Step 3
... and there we have it – GOV.UK Verify (RIP) is a government need and not a user need.

There has been no public outcry about the failure of GOV.UK Verify (RIP). The public in the main have never heard of GOV.UK Verify (RIP), we don't want it and we don't need it. Alive or dead, GOV.UK Verify (RIP) is of no popular interest.

The only people interested are the government. I.e. the administration. That is, GDS and Sir Jeremy. GOV.UK Verify (RIP) is their need. At least they think it is but, be patronising, "have empathy ... remember that what they ask for isn't always what they need".

GOV.UK Verify (RIP) breaks the most revered rule in the book. It was a ghastly mistake in the first place and its predictable passing should be celebrated. RIP.

RIP IDA – the last rites

No need to say it, it goes without saying, it should be obvious to all but,
just in case it isn't obvious to all,
IDA is dead.

IDA, now known as "GOV.UK Verify (RIP)",
is the Cabinet Office Identity Assurance programme.


We can make a meal of it. Or we can do it the quick way.

Let's try the quick way first. Three steps.

Tuesday 26 April 2016

RIP IDA – are GDS talking to themselves?

No need to say it, it goes without saying, it should be obvious to all but,
just in case it isn't obvious to all,
IDA is dead.

IDA, now known as "GOV.UK Verify (RIP)",
is the Cabinet Office Identity Assurance programme.

Every week, the Government Digital Service (GDS) publish statistics about GOV.UK Verify (RIP) on their performance platform. A degree of academic rigour is called for. Without that, GDS are just talking to themselves.

As we speak, some of these statistics are complete to the week 11-17 April 2016 while others include the week 18-24 April 2016. We ignore the latter in the paragraphs below.

1. Total authentications
We ignore the 185,149 basic accounts. These are unverified and have no place in a verified identity assurance system.

User sign-ins went up from 547,416 to 571,191, i.e. there were 23,775 of them during the week.The number of verified accounts went up by 7,509 from 487,267 to 494,776.

Adding the two together – which is GDS's peculiar way – tells us that total authentications went up by 31,284.

2. Authentications per week
Nothing to add.

3. Authentication completion rate
43% for sign-ins and account creations added together. Given that there were 31,284 completed/successful authentications (see 1. above), if that's 43% of all attempted authentications, there must have been 72,753 authentication attempts in all, of which 41,469 failed.

4. Authentication success rate
90% – no idea what this means.

5. Account creation success rate, all services
71%. Given that 7,509 verified accounts were created (see 1. above), if that's 71%, then there must have been 10,576 account creation attempts altogether, of which 3,067 failed.

If GDS intend to enrol 50 million people, say, into GOV.UK Verify (RIP), at the rate of 7,509 per week the job will take 6,659 weeks or 128 years.

It could be worse than that. Those 7,509 verified accounts could be 939 people each creating one account with each of the eight "identity providers". On that basis, 50 million people would need 400 million accounts which could take 1,024 years to create.

Most people die before they're 128, let alone 1,024, which implies that GOV.UK Verify (RIP)'s registration job can never be completed.

The advocates of biometrics look for a failure-to-enrol rate (FTE) of less than 1%. Anything higher casts doubt on the credibility of proceeding with that biometric. GOV.UK Verify (RIP)'s FTE of 29% makes the feasibility of the system problematic.

Given that a total of 41,469 attempted authentications failed (see 3 above) and that 3,067 of them were attempted account creations, the other 38,402 must have been failed sign-ins.

Given that there were 72,753 authentication attempts (see 3. above) and that 10,576 of them were attempted account creations, there must have been 62,177 attempted sign-ins.

38,402 failures out of 62,177 attempts indicates a 62% false reject rate (FRR). 62% of the time, people are being told that they are not themselves.

That is similar to the FRR for face recognition any time more than six months after the enrolment photograph is inscribed on the register. Face recognition is useless as a biometric. GOV.UK Verify (RIP) looks similarly useless if its FRR really is 62%.

You can reduce the FRR, of course, by making it easier to achieve a match. But that has the effect of increasing the false accept rate (FAR), i.e. it becomes easier for a person to pretend that they're someone else, which is the opposite of GOV.UK Verify (RIP)'s objective.

6. Sign-in success rate
99% – no idea what this means.

7. User satisfaction – verification, security, certified company
No data available for the week 11-17 April 2016.

8. Certified company completion rate
55% – no idea what this means. Compare 43%, see 3. above?

-----  o  O  o  -----

4., 6. and 8. above may mean something to GDS but they're talking to themselves – these statistics can mean nothing to anyone else. At 7. above GDS have stopped talking even to themselves.

1., 2., 3. and 5. above broadcast GDS's message loud and clear to anyone listening – GOV.UK Verify (RIP) is a dead duck.

GDS nevertheless plan to announce some time this week that the duck is alive. In their world, perhaps it is. But not here on Terror Firmer, it isn't.

----------

Updated: 11:00

At 29%, GOV.UK Verify (RIP)'s failure-to-enrol rate (FTE) is problematic, as noted at 5. above.

GDS are doing what they can to reduce it:
  • They have increased the recommended minimum age of people trying to register for an on-line account from 19 to 20. That may reduce the number of GOV.UK Verify (RIP) failures. But at the same time it would cut out 1.2% of the population and thereby reduce the universality of GDS's identity assurance scheme, making it less use to government and less attractive to the private sector, who are being courted by GDS.
  • They have also taken to steering people away from the "identity providers" who are less likely to be able to complete enrolment. Again, that may reduce the number of GOV.UK Verify (RIP) failures. But it would do so at the expense of reducing the number and variety of enrolment agents/"identity providers"/"certified companies" when GDS's sales pitch to the populace is precisely that there is a wide and high quality choice on offer.
The other action GDS could take is to change the enrolment process. At the moment, the identity of a given name, address and age with sex optional is verified by reference to passport details, driving licence details and credit history. The enrolment process could be changed to take into account further personal information.

What further personal information?

Candidates include your health records, education records, travel records, bank account transactions, insurance policies, mobile phone usage, email contact lists, social media accounts, ... GDS claimed 18 months ago that they were about to announce their choice of additional personal information to include in the GOV.UK Verify (RIP) enrolment process. They still haven't.

Most people are not often exercised by questions of privacy but GDS's demand for yet more personal information might tip the balance.

Despite GDS's claims to the contrary, we have little or no proven control over these personal details once they have been divulged.

The privacy and fraud risks seem exorbitant compared with the benefit of being able to use GOV.UK Verify (RIP) to view our driving licence details on-line.

It seems unnecessary to amplify those risks when we already have the Government Gateway as a long-established working alternative to GOV.UK Verify (RIP).

Unnecessary also when, according to GDS, no other country has adopted this approach, the UK is in the vanguard.

In the absence of any additional personal information being added to the GOV.UK Verify (RIP) enrolment process we are left with GDS's eight "identity providers".

Five of them are being branded useless – Barclays, CitizenSafe/GB Group, the Royal Mail, Safran Morpho/SecureIdentity and Verizon. That must sour relations between them and GDS and it might sour relations between them and the three favoured "identity providers" – Digidentity, Experian and the Post Office.


The position of Barclays is odd. You'd think they would be among the best enrolment agents. Whatever percentage of applicants they can shepherd through the registration process should be definitive. Far from consigning Barclays to the out-of-favour list, perhaps GDS should be checking the apparently outperforming Digidentity, Experian and the Post Office to make sure that they aren't relaxing the matching criteria and exacerbating the FAR problem (false accept rate).

With only three favoured "identity providers", GDS are exposed. The Post Office is not a "certified company", its application for approval lapsed well over a year ago. And Digidentity and the Post Office are linked. If one of them suffers a security problem, they would both be knocked out, leaving GOV.UK Verify (RIP) with just one "identity provider" – Experian.

This visible promotion of Experian into the UK Constitution as the "identity provider" of choice for the entire nation has not been even debated by Parliament, let alone agreed. In this matter, GDS are wildly out of their depth and ultra vires. They need to talk to a lot more people about it than just themselves.


Updated 3.5.16

A new metric has been added to the GOV.UK Verify (RIP) dashboard:

9. Certified company choice
It's 81%.

GDS continue to recommend against registering with Barclays, GB Group/CitizenSafe, the Royal Mail, Safran Morpho/SecureIdentity and Verizon.

User satisfaction, please see 7. above, remains a thing of the past. It is measured in three ways and none of the figures have been updated since 27 March 2016.


Updated 11.11.16

GDS don't always talk to themselves about the performance of GOV.UK Verify (RIP). Two days ago they sent Chris Skidmore MP off to talk to Korea about it. In his speech, he said:
GOV.UK Verify [RIP] allows the citizen to create a single online identity to access a growing number of government services. And since going live in May, GOV.UK Verify [RIP] has verified more than 900,000 users.
Take a quick peak peek at the GOV.UK Verify (RIP) dashboard on the GOV.UK performance platform. On 1 May 2016 there were 692,951 GOV.UK Verify (RIP) accounts. By 6 November 2016, that figure had grown to 911,096.

Mr Skidmore is a historian as well as a politician. He knows to check his sources. But on this occasion he didn't. Since going live in May, GOV.UK Verify (RIP) has verified 218,145 users and not "more than 900,000" of them.

Even its supporters warn about the "wildly unrealistic expectations" of GOV.UK Verify (RIP). Next time he delivers a speech prepared for him by GDS he is advised to check it first.

RIP IDA – are GDS talking to themselves?

No need to say it, it goes without saying, it should be obvious to all but,
just in case it isn't obvious to all,
IDA is dead.

IDA, now known as "GOV.UK Verify (RIP)",
is the Cabinet Office Identity Assurance programme.

Every week, the Government Digital Service (GDS) publish statistics about GOV.UK Verify (RIP) on their performance platform. A degree of academic rigour is called for. Without that, GDS are just talking to themselves.

As we speak, some of these statistics are complete to the week 11-17 April 2016 while others include the week 18-24 April 2016. We ignore the latter in the paragraphs below.

Monday 25 April 2016

Willing enthusiasm isn't enough

11:19 a.m., 8 October 2014, 18 months ago, someone saves a copy of the Transactions Explorer page of the Government Digital Service's performance platform:


Then someone updates HMRC digital team plights troth to wrong Liege and forgets about it ...

... until recently.

You will notice that GDS were trying to measure how digital central government is, department by department. The data they used is repeated below. You won't be surprised which department wins ...

Department
Digital take-up*
Total cost*
Data coverage*
Transactions per year





HM Revenue and Customs
91.90%
£528m
77.30%
1,233,662,926
Department for Transport
57.40%
£268m
73.60%
130,337,698
Home Office
4.83%
£1.43bn
76.20%
126,270,677
Department for Work and Pensions
17.20%
£3.77bn
95.80%
107,781,180
Department for Business, Innovation and Skills
82.40%
£242m
54%
40,513,661
Department of Health
40.80%
£308m
61.90%
33,647,220
Department for Environment, Food and Rural Affairs
86.60%
£101m
76.20%
22,580,710
Ministry of Justice
21.40%
£5.02m
52.40%
8,508,685
Cabinet Office
100%
£32.1k
100%
4,870,984
Department of Energy and Climate Change



1,331,834
Foreign and Commonwealth Office



549,065
Department for Communities and Local Government



515,756
Ministry of Defence



477,707
Department for Education



245,144
Attorney General's Office



65,658
Department for Culture, Media and Sport



33,589
Department for International Development



21,001




* Figures are based on data for high-volume services only




... yes, the Cabinet Office, which includes GDS, has 100% digital take-up (whatever that means) and 100% data coverage (whatever that means) and it's the winner.

That was 18 months ago. The figures were questionable.

Now, if you look at the services data on the performance platform, you find that GDS have stopped trying to measure digital take-up and data coverage. They list 802 public services and they have data on 571 of them which, between them, notch up 2.38 billion transactions p.a.

Take a look at GDS's data and you see that the 802 public services are divided up, department by department, as follows:

Department No. services
Department for Business, Innovation & Skills 177
Department for Environment, Food & Rural Affairs 118
Department of Health 98
Department for Transport 77
Department of Energy & Climate Change 67
Department for Work and Pensions 48
Department for Culture, Media & Sport 43
Home Office 41
HM Revenue & Customs 34
Ministry of Justice 29
Foreign & Commonwealth Office 18
Department for Education 11
Cabinet Office 11
HM Treasury 10
Valuation Office Agency 9
Department for International Development 3
Department for Communities and Local Government 3
Ministry of Defence 2
Attorney General's Office 2
UK Export Finance 1

Does the Department for Communities and Local Government really offer only three services? And the Ministry of Defence just two?

Again, the figures seem questionable.

GDS keep promising us canonical registers. On which government policy can be based rationally. Their performance platform omits data on the Government Gateway. And it omits data on the Basic Payment Scheme for farmers. And it doesn't look as though GDS can even count public services.

The Office for National Statistics have got a lot of work to do to bring GDS up to speed on data science. Willing enthusiasm isn't enough.

How can GDS be ready to build Government as a Platform?

Tomorrow they're attending – or possibly even hosting – a seminar on blockchain, Blockchain: exploring uses in government. Are they ready for that?

Willing enthusiasm isn't enough

11:19 a.m., 8 October 2014, 18 months ago, someone saves a copy of the Transactions Explorer page of the Government Digital Service's performance platform:


Then someone updates HMRC digital team plights troth to wrong Liege and forgets about it ...

... until recently.

You will notice that GDS were trying to measure how digital central government is, department by department. The data they used is repeated below. You won't be surprised which department wins ...

Openness should include farmers

One of the standing jokes about the Government Digital Service's identity assurance scheme, GOV.UK Verify (RIP), is the list of public services using it:


How can DEFRA's Rural Payments service be connected by GOV.UK Verify (RIP)? DEFRA don't have a rural payments service, as we always point out, at least not a computerised one – the computerised system GDS tried to build collapsed and farmers are all applying for their money using pencil and paper now, as a result of GDS's failure. There's nothing for GOV.UK Verify (RIP) to connect farmers to.

It's misleading to pretend that GOV.UK Verify (RIP) connects farmers and their agents to the Rural Payments Agency's Basic Payment Scheme (BPS) and, if you're in any doubt, just look at the BPS dashboard on GDS's performance platform:


Couldn't be much clearer than that, could it? Nothing's happened since June 2015.

Except that that's not right. Take a look at the Rural Payments Agency's 7 March 2016 blog post, Start your 2016 Basic Payment Scheme application now. It's all on-line.

Take a look at their website, Everything rural businesses need to know about the Basic Payment Scheme (BPS) in 2016 - including how to claim BPS online using the Rural Payments service.

Check with the National Farmers Union. And with the Farmers Gazette, 22 March 2016, Thousands of farmers log on as 2016 BPS application window opens.

Watch the film:



Farmers and their agents have been able to claim on-line since 2 February 2016. History didn't stop in June 2015, pace GDS's performance platform.

We seem to have another butcher's thumb on the scales case here like the Government Gateway. Why is the performance platform incomplete? Why doesn't it include data on either the Government Gateway or the live rural payments system? Why are GDS ignoring BPS?

Openness should include farmers

One of the standing jokes about the Government Digital Service's identity assurance scheme, GOV.UK Verify (RIP), is the list of public services using it:


How can DEFRA's Rural Payments service be connected by GOV.UK Verify (RIP)? DEFRA don't have a rural payments service, as we always point out, at least not a computerised one – the computerised system GDS tried to build collapsed and farmers are all applying for their money using pencil and paper now, as a result of GDS's failure. There's nothing for GOV.UK Verify (RIP) to connect farmers to.

The gateway to openness

"The annual end submission date for tax self assessment in January is one of the critical events in the year for the Government Gateway, HMRC and government IT systems as a whole". So said David Hargreaves on 25 February 2016 in Managing the self assessment tsunami:
This year was the largest yet. The Government Gateway processed over 2.9 million self assessment submissions in January. This was just part of almost 7.5 million transactions it handled over the whole month, and the 10 million online self assessments processed in 2015.

The volumes topped 400,000 on Friday 29 January. That’s the equivalent of 8.5 submissions per second.
The Government Gateway is clearly quite a substantial cross-government platform. And these are notable transaction volumes.

And yet, if you try to find anything out about the Government Gateway on the Government Digital Service's performance platform, look what you get:


It looks to you as though there's no such thing as the Government Gateway.

The same happens if you search data.gov.uk. You get no data.

It's as though the Government Gateway doesn't exist. Despite the Government Gateway having received 12.9 million 2015-16 self-assessment tax returns. And who knows how many VAT returns. And PAYE/NI returns. And corporation tax returns. And at what cost.

Given that the UK aims to be the most "open" nation on earth when it comes to government data, this is a glaring omission:
  • The open data enthusiasts claim that openness leads to innovation – the omission of Government Gateway data is stymying innovation.
  • The data scientists argue that rational government policy depends on the availability of data – Government Gateway data is not available and therefore policy is likely to be irrational.
Why is data relating to the Government Gateway hidden?

There's no point guessing. There's good reason though to insist that the omission be corrected, quickly, for the general good.

The gateway to openness

"The annual end submission date for tax self assessment in January is one of the critical events in the year for the Government Gateway, HMRC and government IT systems as a whole". So said David Hargreaves on 25 February 2016 in Managing the self assessment tsunami:
This year was the largest yet. The Government Gateway processed over 2.9 million self assessment submissions in January. This was just part of almost 7.5 million transactions it handled over the whole month, and the 10 million online self assessments processed in 2015.

The volumes topped 400,000 on Friday 29 January. That’s the equivalent of 8.5 submissions per second.
The Government Gateway is clearly quite a substantial cross-government platform. And these are notable transaction volumes.

And yet, if you try to find anything out about the Government Gateway on the Government Digital Service's performance platform, look what you get: