A weak centre is strong
Relocating strength from command to capability
After months of angst as Twitter has collapsed, Elon Musk’s lurch in white nationalism and poking his big neb into UK race riots has suddenly turned BlueSky from a sleepy small town into quite the UK town square.
This week has seen a very interesting debate around the structure of state and meaning of state reform.
It was sparked by an article in The Times about a proposed revival of Cumming’s era dashboard/mission control rooms in the Cabinet Office and Ciaran Martin’s perceptive comments on it, going onto James Plunkett of Nesta writing about dashboards.
As it happens, I was just in the midst of pulling a long discussion about the necessity for a weak co-ordinating centre in what should be the final Working Paper of my project before the report for Scottish Government Foundations of the Digital State comes out. (I hope to get it out to interviewees for comment next week.)
This blog post is basically the worked example from that paper pulled out - it steps through a range of choices that are informed by various previous working papers.
There is a social security system. It is specified in law (the what of a digital service, the functional requirements are in statute and secondary legislation).
The evidence is that the necessary non-functional/infrastructural requirements are poorly specified and spread across many places as described
We wish that system to be developed in way that facilitates joined-up working. To do that in a decentralising manner with a weak centre, the centre is going to issue technical standards and guidelines that cove all parts of the state.
There will be technical guidelines about data sharing – we want to separate the means to share data from the will to share data. The former goes to the technical standards org, the latter to parliamentarians.
There will also be guidelines about exposing functional services as API and splitting GUIs from service layer, and ones about authentication and delegation.
The state first creates an institution that is capable of issuing the necessary standards – they pass and become obligatory for the social service department. They are timed in – so the obligation is set now, and must be complied with in, say, now +3 years.
The structure of that body is described in this working paper:
A theory of state that supports it is here:
A draft initial charter and discussion of the contents of the standards is here:
The social security system comes into line with the standards over time.
Now we know that 80% of social security claimants are one and done and 20% need help – which is provided by a call centre.
So the social security agency can be reorganised into a service platform team, a front-end self-service team and a call centre team.
Now the call centre teams realises that its 20% is 15% fairly simple and 5% high dependency cases. It approaches the social work teams and suggests that they proactively take on high dependency cases. Because social security is delivered as a standard API, and because the social service team’s software is also aligned with that API the social security system can be embedded into social work workflows. The shared delegated permissioning system allows the citizen to give permission to their social worker to apply for benefits on their behalf and this permission passes through to social security.
Dundee implements this, Highland and Islands doesn’t – it doesn’t chime with how they work, their social workers will phone the call centre on their clients behalf (using the same delegated powers).
Dundee Social Security realise that of their 5% of the total, 1% is care leavers and 1% prison leavers – they reach out to the care service and prison service and the dance continues.
(The technical mechanics of this process are described in more detail in these two working papers.
These proposals are profoundly infrastructural and will continue to inform the organisation of the state for 100 years – so the parliamentary oversight outlined in Working Paper 0 The locus of change is critical here.)
Essentially we are breaking down the monolithic departments into smaller systems with published interfaces that conform to shared, known and stable standards.
The central standards body has an overview of state data and can use that to suggest Machinery of Government changes based on data and process consolidation.
Some of these will require legislative consolidation which can be effected without overwhelming parliament with an Enabling Act.
.Once that has happened we can remix the state, and redirect resource, more money for social work, less for social security call centre, etc, etc.
At the core is a small team with maximum autonomy over resources, how they spend their money, technical choice, sequencing, delivery, testing, etc, etc.
These co-ordinate without communication by using standards – and interact via defined interfaces. These interfaces present automatically by use of open-source, shared standards-embodying software components. The centre also promotes technical tools that embed the standards (publishing meta-data and data models, API documentation, change log generation and release documentation, etc, etc) to make compliance press-button and not slog-through-paperwork.
The centre is weak, but co-ordinating, the periphery, here teams within departments are strong.
The strength of the weak-centred state is the depth of its expertise, the speed with which it can reconfigure and regroup, the ability of multiple systems and services to survive and work around the acute crises that some will always be in, its ability to cope with sudden unexpected external shock.
It is a state that looks like the modern internet – a similar weak-centred organisation - which was expressly designed to continue to work after the Soviets dropped the bomb, with self-routing and self-fixing at is heart.
Note: This discussion of decoupling and the weak centre focuses on the how of digital systems. Foundations of the Digital State contains a whole other stream of work looking how to better define the what of state computer systems – in particular Working Paper 7 Experimental digital legislative processes which you can read about in this blog.