Chapter 13

What comes next

If you’ve made it this far in building your digital institution, you’ve done well. A thriving digital institution at the centre of your organisation, energetic leaders in departments and new digital ways of working beginning to take hold at scale; all these things are substantial victories. Some of what you’ve done would have been considered impossible a couple of years ago. Taken together, this progress will change your organisation for the better. Users will notice the effects on the services you deliver. So will your colleagues. It is now more of an effort for your organisation to go back to the bad old ways than it is for it to stick to the course. Inertia is finally starting to tip in your favour.

The next stage comes down to tackling the most fundamental changes the Internet era will bring for post-industrial bureaucracies, and in particular, how accountability, money and risk. For public organisations, these questions shape the foundations of democratic institutions. Few companies – and no governments – have completed this final step. We are confident that the country that gets there first will win a big prize.

Because nobody has yet got to the finishing post, it is impossible to be certain exactly what the full set of practical steps to get there will be. In this final chapter, we have made a few suggestions for what an established digital team at this stage is likely to encounter in the future.

Doing less

Having shown that digital transformation of existing services is both possible and worth doing, the chances are that deep exhaustion will now be creeping into your team. Change is physically and mentally draining, the conflict it brings especially so. Digital transformation of any large organisation is tiring, partly because it is about loss. If you are to succeed in delivering the changes you want, various roles, processes and practices will end up being consigned to the past. For people coping with that loss, especially if it feels out of their control, there is a limit in how far they can stay with you.

Ignoring the health effects of organisational change is a mistake. Poor physical and mental health, exacerbated by professional stress, hurts people. It also kills organisations. There will be a moment in your transformation – certainly after no more than two years of long hours and hard work – where the digital team must take stock to address the debt it has accumulated from running fast. Some of that debt will be technical. Flaky code thrown together for the sake of delivery speed at the expense of scalability and elegance will need revisiting. So will sketchy design patterns, hacked processes and tatty office space. However, much of the debt will be human. The steady drip of stress takes a toll. Take time to recuperate, replenish, and go again. We regret not doing this more when we were working in government.

All of the digital team will need to replenish their energy for what is to come. By this stage, a good few people in your organisation may consider the team’s work as largely done. As far as they’re concerned, it has launched a few decent digital services, and saved a chunk of money. Mission accomplished. You’ve annoyed people on the way of course – that’s a pity – so they think that perhaps now is the moment to consolidate the success and slow things down. If the digital team is all too tired to keep going, those keen to go back to an easy life will push back.

Resisting the hype

Every business strategy presentation for the last two years (and the next three) will have a slide that says something like: ‘AI, blockchain, Internet of Things – what should we do?’ For most organisations, this discussion is a little premature.

Even allowing for the fact these technology breakthroughs are near the top of their hype cycle at the time of writing, we are not saying that they are unimportant for large organisations, public or private. Far from it. We’re confident that artificial intelligence, connected devices and advances in cryptography will change the world, in predictable and unexpected ways. There are many excellent books, talks and blogs that do these subjects far better justice than we have the space to here.

However, for the purposes of digital transformation, we’re unconvinced that organisations unable to provide their employees with a reliable system for filing their travel expenses should be betting the house on nailing the implications of artificial intelligence. Don’t try to put the fire out with more petrol.

To a certain type of technocrat, innovations offer an irresistible opportunity to do a lot more talking at the expense of doing. It is noticeable that technologies like blockchain and artificial intelligence have especially gripped executives in organisations that have largely failed to react to the open internet’s impact. In our experience, the more senior the figure, the more interested they are in technologies at the bleeding edge of discovery. On one level, this seems counterintuitive. Why would senior leaders with a track record of not applying more obvious and understood trends suddenly jump on the singularity bandwagon?

The answer, we suspect, is that technologies like blockchain and AI are for the most part still largely in the realm of theory. Delivery is not yet well-understood or demonstrable at scale. Very few companies (let alone established public institutions) have either the volume or sufficient structure to their data that would allow them to create a base of information that machine algorithms are able to ‘learn’ from. Their data is either unreliable, disparate or simply not there. Without fixing the foundations first, adding yet more new technologies to the mix in a legacy organisation is a recipe for adding yet more complexity.

Unaware or uninterested in this, some executives – especially those in policy-led government bureaucracies – feel comfortable with having an abstract debate about the consequences of, say, big data and connected devices. This intellectual workout takes place as they simultaneously fail to engage with the possibility of delivering real-life applications of well-understood technology.

A cynic might argue that these leaders have simply grown used to the idea that having a conversation is tantamount to doing something. That’s probably unfair. The more likely culprit is a legacy bureaucratic culture that allows for excellent, rich, paper-based discussions among technophiles. Clever officials or strategists can draft many beautiful position papers about the prospects a new technology may hold for public life or future profitability. However, their organisation generally legislates against doing anything experimental with those technologies to test their assumptions against reality.

The answer to the technophile paradox is not to throw these new technologies off the agenda completely. AI is not going away. Ignoring it altogether is a more dangerous strategy than becoming distracted by its potential. However, in considering the disruptive possibilities of new technology, governments and corporate bodies would be wise to think about how well their organisation is arranged to make the most of them. If the culture, people and working practices of the institution are still grounded in principles that were set down in the age of the telegraph, the chances of responding with the requisite flexibility and agility to machine learning are slim. How can you be sure you’re not buying snake oil? Which roles and professions should be part of the conversation? Which start looking obsolete? Can you buy into the business models that AI or data science services will use? If you’ve failed to get through the first digital transformation of your organisation, you will also fail to make the best of the second.

Retrospective: paperless driving

For all the cold water we are pouring on them in this chapter, AI and machine learning represent a new frontier. For government, they may even be more important to the development of public services than the internet itself. AI and machine learning can upend long-established government processes and systems and – more importantly – transform citizen outcomes by making public services ever more frictionless. For an organisation to make the best use of AI, it is very likely they will have already spent a lot of time and energy trying to take friction out of existing services. In the process of doing so, they will have fundamentally changed how their organisation works, preparing it to make better use of increasingly innovative technology.

DVLA, the UK’s motoring agency, is a good example. In 2010, DVLA was a laggard for IT in government. It was ruinously expensive and dangerously complacent. So ingrained were the incumbent IT suppliers, they literally had a seat on the board. DVLA’s head office in Swansea had space to accommodate 10 enormous articulated lorries, each one bringing tonnes of paper through the doors every day.

Thanks to the tireless work of many people, the DVLA took the difficult steps needed to bring technology skills in-house. In the process, it was bold enough to scrap both the paper car tax disc and the paper counterpart driving licence, with relatively few hiccups along the way.

On the surface of it, simply getting rid of ancient forms would appear to have little to do with the possibilities in applying AI and machine-learning principles to driving. This technology would, for example, give a minister the ability to incorporate real-time data of every driver in the country into a digital driving licence. In a situation where elderly drivers were causing fatal accidents, the minister could advocate for digital driving licences based on an individual’s personal driving style or health – alcohol levels, blood pressure – accessed via the driver’s phone or smart watch. The AI would then decide if that individual was safe to drive that day. The direction a government chooses to take with AI is as much a political choice as it is a technological one, but it is so far removed from the old paperwork as to feel like an entirely separate conversation.

These two perspectives are united by one thing – the institution responsible for making them happen. DVLA’s digital efforts to date have exemplified two things: getting the organisation to focus on making services frictionless for the user’s benefit, and putting in place the multidisciplinary teams that could deliver them. This combination of attitude and capability is what made true innovation possible, and opened the door to delivering fully digital licences in the future. Without the recognising the crisis the organisation faced, the DVLA would not have got close to this position. The chances of it delivering artful ­machine-learning-led services without fundamentally changing the institution itself are slim to none too.

One indicator of an organisation’s maturity and readiness for this next wave of technologies – assuming that it already has a digital working culture in place – is how it looks after its data. If an institution knows what data it owns, makes it machine readable, and has considered the data protection and privacy issues that come with the responsibility of looking after it, it might have a fighting chance. Without those things, forget it.

Whatever the hype may be, new technologies like machine learning are forcing the right questions into the open. Lying beneath most of them is the fundamental issue of trust. To move from transforming services to transforming the organisation, you need to look with a fresh eye at the balance of trust, accountability and power.

Government as a platform

In the UK government – and many others – the path of bureaucratic power is a vertical one. The British constitution is a famously flexible thing (proving that the Establishment has no problem with an iterative, agile approach when you really come down to it), but certain conventions are fixed. Put very simply, prime ministers are accountable to parliament for the collective performance of their ministers. Ministers are accountable to parliament for the collective performance of their departments. Parliament is accountable to the voters. Straightforward enough. The bureaucracy, too, is accountable to parliament, with various committees assigned to holding each department to account.

The unit of government organisation in this system is the department. It is the department that is given money, determines how it runs itself, and owns the policy. It is the minister of that department who makes decisions. The slight kink in this model is the senior civil service – the 3,000 or so top officials – who are technically managed by a central department, the Cabinet Office. Nonetheless, the large majority are assigned to departments, and behave as if owned by them.

All this context is needed to explain that government works best when it is trying to deliver something through one department. When that happens, the money, control, responsibility and accountability all align in a single management chain. This doesn’t guarantee success. The organisational, management or political pathologies of individual departments can easily scupper things. As a rule, though, the more departments get involved, the more difficult it becomes to get things done.

The departmental model is not necessarily broken. Look at pretty much any advanced economy around the world, and the government departments almost invariably have the same names: health, education, interior ministry, and so on. This equilibrium also suggests that the buckets of policy activity that governments split themselves into are, if not right, probably a fair attempt at being the least wrong.

Where the departmental model demonstrably fails is in delivering things that all departments need. All the incentives guide departments towards thinking they are special when it comes to, say, building a system that can handle appointment booking, or buying laptop computers for staff. Even if they recognise they aren’t that special, there is no reason for a senior leader of the Department of Pencils to say, ‘We’ll rely on the Department of Pens to sort this out for us.’ If Pens does a bad enough job to cause Pencils’ service to fail, who takes the rap? Pencils.

The departmental arrangement of accountability – and the lack of trust that quietly festers between those separate institutions – is most obvious in government, but applies equally in many large, federated companies. This is especially obvious in conglomerates, where different CEOs consider their company to be their domain, and see sharing with others in the group as risky business.

Bringing us back to digital transformation, it is easy to see the problem. Transforming services and applying digital ways of working to individual departments is not an easy task, but a relatively manageable one, given the clearer chain of accountability. Transforming the entire organisation into a genuinely user-centred one is a whole different puzzle. Users don’t particularly care which department provides them their service. They just want it to work. Taxpayers and shareholders don’t care who builds or buys the system that handles online payments. They just don’t want to see the organisation paying separately for 30 versions of the same thing.

Most of the digital companies that have grown to dominate the global economy are platforms. They provide the infrastructure for whole markets to build themselves upon, and take a small cut off the millions that use them. This is true in taxis, holiday accommodation, advertising, news, retail, and many more besides. The biggest companies in all of these sectors often don’t own stock or property. They own the market place.

For government, the opportunities in building platforms like these are real. Most public services are made of online and offline components that have been rebuilt or bought hundreds of times at the public’s expense. Making payments, taking payments, publishing information, progress notifications through text and email, appointment booking, licences, grant applications. How great it would be if you could build these things once and have hundreds of public organisations using them, while steadily improving the service over time.

The UK tried, with qualified success, to bring platforms into government. Some of them, like GOV.UK, a central platform for publishing government information, worked very well. The UK is still working on platforms for payments, notifications and identity management, among others. At the time of writing, all of these have an uncertain future.

The threat to the status quo presented by platforms is that they erode the idea of departments as the organising framework of government, and replace it with something more attuned to what users of public services expect. Some view this as the thin end of a wedge that inevitably concludes with breaking constitutional norms. A legitimate charge laid by opponents of platform thinking is that it isn’t entirely clear exactly what would replace the old structure. That is a little unfair; the full implications of departmentalism were far from understood when that route was chosen. The resistance the GDS encountered, however, tended to be less reflective. Known inefficiency was considered by many in the administrative side of government to be less scary than an unknown future. Senior bureaucrats are largely untouched by the consequences of their choices. That several of them would rather preserve conventions over the opportunity to deliver decent services they will never use is disappointing, but perhaps unsurprising. Those conventions got them where they are today.

So what gives platforms, and therefore the genuine transformation of a legacy-led organisation into one ready for the internet era, a chance of success? We can’t say for certain, but there are at least five steps to getting closer.

Data registers

Common platforms are the public face of platform government: the component parts of services users see and use. The less glamorous side is the data architecture that sits under the services built from these components.

Creating accurate lists may sound like a prosaic task, but replacing the confused, duplicatory and inaccurate data architecture most old organisations sit on with new, canonical data sources is probably even more important than the ­user-facing platforms. Having single, accurate, trusted sources of information for all parts of the business or government to refer to cuts out many of the mistakes and burden of data re-entry for users.

For most bureaucracies, creating reliable data registers sounds like an unpleasant and near-endless job. It may take a decade to complete the transition from the patchwork of incompatible sources and false information to an architecture a digital native company would recognise as worth having. The longer it gets put off, the further you are from becoming a successful platform organisation.

The good news is that you can start small. Until very recently, the UK government didn’t have a single agreed list of countries. Instead, there were scores of lists, some out of date, some incomplete, some with alternative names. The lack of consistency is maddening enough for people, but more crucially, makes the reliable use of machine-readable data near impossible. A register of single countries is now available for every department to use. It’s a start.

Central power

The role of the centre in a platform government is up for debate. Most of the argument centres on what role a central department or institution should be responsible for designing and running platforms, versus playing a convener, standard-setting role, versus butting out and gently encouraging departments to play nicely. In the UK case at least, a century and a half of the last option has failed to deliver cross-organisation platforms that work.

Whether the centre plays a role in directly delivering platforms – whether that be building them itself or coordinating their purchase from suppliers – is harder to answer. Our instinct is that the central digital institution should take the lead in delivering at least some common platforms. The primary reason for this is that leaving the centre to chairmanship blunts its own delivery muscles. This makes it less sharp and credible in judging the quality of delivery elsewhere. Equally clearly, the centre can’t do it all.

Accountability with control

Reach a certain level of seniority in government and you spend most of your life in meetings explaining or defending the work of others, many of whom you may have only met once or twice. The committee men and women (they are mostly men) are in a difficult position personally. While accountable, they have little or no control over what they carry the can for. This leads to the strange situation where the most qualified candidates for the top jobs are not those who deliver superlative projects, but those who can dance their way through trial by angry parliamentarians.

For platform government – or indeed, any form of digital government – to work, accountability and control need to be brought closer together. This may mean far more officials being called to explain their work in public and to parliament. It may also mean radically changing the way the legislature holds the bureaucracy to account. Perhaps moving away from the set-piece committee hearing – which lends itself to drama, defensiveness and post hoc rationalisation from MPs who rarely find themselves in possession of all the facts – would be a good idea. Having legislatures pivot towards a more regular conversation with public officials who are in charge of day-to-day delivery would make the accountability process a more valuable one.

Trust

Trust is the most precious commodity of the digital era. It is what makes the open internet work. Your organisation, be it government or corporate, will be forever hobbled if it is unable to trust itself.

When you strip away the pomp, process and procedure, an amazing number of the knots that bureaucracy ties itself in are largely down to the fact that senior officials trust neither their colleagues, nor their political bosses. A transparent government machine, built around platforms, would be a source of worry.

Until senior officials can trust each other enough to rely on one another’s work, government as a platform cannot and will not happen.

A crisis of trust

For that change to happen, and the next phase of digital transformation to begin, we think another crisis will have to take place. Trust in democratic institutions is a fiendishly difficult thing to quantify. The real indicator that matters, though, is not trust as a blanket idea, but an organisation’s trustworthiness to carry out a particular activity. You might trust your doctor, say, but you wouldn’t necessarily trust him to fix your boiler.

There is a version of the future – a version where the transition to the digital era continues apace. People’s expectations of what is possible continue to grow, yet deep political shocks leave surprisingly few marks on people’s day-to-day experiences of the state. More people may step back and wonder: ‘Should we trust this machine to deliver anything at all?’