Co-Creator Blog: How to Fix Things - and Move Fast: a public innovation approach to rewiring the state

In our Curiosity Incubator drive to move beyond the generic 'direction of travel' advice permeating so much of the discourse in UK government and policy circles at the moment, to focus on specific examples of how we might collectively work on improving it, Oliver Smith, Founder of the Daedalus Futures ethical AI consultancy, and Associate at Future Governance Forum, and Bridget, Curiosity Incubator Founder, co-imagine together a public innovation approach to 'rewiring the state'.

How to Fix Things - and Move Fast: a public innovation approach to rewiring the state 

Some encouraging new themes emerged in Darren Jones’s recent speech entitled “Move Fast and Fix Things”, lending colour to what this government’s approach to “rewiring the state” will look like, and some first, fast, initiatives to make this happen. Drawing on our collective experience in the key themes of the speech – AI, digital transformation, strategic thinking, and effective learning – in this blog we aim to sketch out some of the evidence of what works in government when attempting to fix things, and move fast.     

Rt Hon Darren Jones, courtesy of the Institute of Government: How to Move Fast and Fix Things (no inspiration drawn from Professor Frances Frei's book of the same name, according to the speech - shame as it's a good book, take a look)

Most of the commentary on this so far has focused on “Move Fast” – but we want to start with what “Fix Things” can mean, before considering how to “Move Fast”, and what government requires for both.

Fall in love with the problem, not the solution 

Silicon Valley is, of course, famous for the phrase “Move fast and break things” that Darren Jones’ speech riffs off of. However, another Silicon Valley phrase is apt at describing how to fix things instead of breaking them: “fall in love with the problem, not the solution” – identify the problem we’re trying to solve, and reverse engineer the pathway to making that happen. Sounds simple, but in practice it is hard to balance knowing what and how much information to gather to understand the problem versus being paralysed by over-analysis. The good news is there are some successful examples, including those from the US cited in the speech, that we can draw on to see what’s worked, and how. 

Starting with problem definition: in The Move Fast, Fix Things speech it was: the British state is broken. But how would we know if it were fixed? Whether or not the citizens of the country felt and agreed that their interactions with the state were positive, productive – ie not-broken. This is a radical departure from some current definitions of what ‘good policymaking’ looks like, which separates “policy” from “delivery” in process, approach, organisationally, and in how we think about both.   

Critically, our proposed approach also calls for a new set of practical relationships, not just between civil servants and the public, but with others within government and the public sector overall. For many of the most challenging and complex policy issues, defining what ‘good’ means can only be addressed by working with the citizens that receive the service, as well as the network of front-line staff in different organisations who deliver it.      

For example, there is no single answer to what living well with multiple chronic diseases might mean, only each individual can fully define that – and those living with these diseases, their families and carers, and those front line staff across multiple different services who work with them, have different ways of helping define and understand these challenges. This requirement for more relational services is part of the driver for many policy think pieces at the moment, calling for a new relationship between people and the state. But how can this happen throughout the work of government, rather than in a siloed or one off way? And how can it happen at speed?    

Moving fast, but in what direction?     

One of the key planks of the speech was that digital transformation and particularly the use of AI would be central to the new learning and transformation strategy for the government. With almost half of civil servants currently voicing scepticism about whether they will be provided with training on AI that genuinely helps them use it to improve job performance, there is an opportunity to use the acquisition of “digital and AI skills” to rethink how we approach how we solve problems, and look at learning, as a thing in itself.

Although LLMs continue to advance, we are still facing what HBS termed the “Jagged Technological Frontier”, in which an AI can perform very differently on tasks that, to a human, appear similar. Navigating this requires expertise, and self-confidence in your expertise. Indeed, research by Microsoft has demonstrated that people who lack confidence in their own expertise tend to use AI without fully exercising their critical thinking. This is exactly the situation in which users are less likely to spot inaccuracies, or simply accept a mediocre or flawed response. 

The need for deep expertise to make the most of AI means that we should not expect it to plug the knowledge and experience gaps that civil servants currently face. Most civil servants in central government are generalists, often lacking deep expertise on the topic for which they are responsible nor possessing the tacit knowledge that those working with citizens on the front line of service provision have in spades. Indeed, AI may make this situation worse by providing confident-sounding but incorrect (or over-simplified) information about citizens’ needs and frontline delivery experience. AI can certainly help move fast, but it takes human expertise to ensure that it does so in the right direction. 

However, this does not mean that there is no role for AI. If we were to apply AI to help bolster learning and productivity, drawing on evidence of how these tools are best used, it might look like this:

1. Problem definition: build an approach to learning how to use AI to move towards a workflow and policymaking approach that considers the citizen’s experience as the primary criterion of success.

2. ‘Paired Policymaking': inspired by the concept of ‘pair programming’, using the creation of an AI tool in government as an opportunity for not just fact checking but co-learning, of civil servants together with frontline delivery staff, in a knowledge exchange, as part of the tool’s design. This includes co-developing the tools’ use cases to support co-learning and co-creation for knowledge acquisition.

3. Embedded within policy: AI tools could be built by in-house Government coding teams, and embedded within the policy drafting process – co-designed both with frontline colleagues, and input from citizens themselves.     

An AI tool whose purpose is to ‘automate’ only, will not address the necessary acquisition and use of tacit and other kinds of knowledge needed to use AI tools well. Our proposed approach addresses this and also provides an ongoing learning opportunity built into the tool design and workflow itself, which increases connectivity and accelerates the kinds of learning needed to refocus work on the end-user (ie citizens), and all of the complexity and variety of their needs, experiences, and contexts, together with frontline colleagues.

A rallying call – building 21^st century learning in government that works

Taken together with the Move Fast, Fix Things initiatives overall including ideas above, it’s a doubly positive step to see news of the formulation of a National School of Government and Public Service (NSGPS), to help build innovation and capabilities in Government.

To ensure that this NSGPS can support the process of finding the kinds of solutions like the AI approach that we describe above, it must build on the latest evidence. We suggest the four approaches below as a starting-point:

• Taxonomy of skills + taxonomy of effective learning and innovation interventions: There’s great initial work happening in the Cabinet Office team on creating a taxonomy of skills on AI, digital transformation and strategic thinking as well as a universal curriculum for the new NSGPS. However, two things could ensure greater success: a similar taxonomy of learning interventions, tools and approaches by how effective they are for impact, and a focus on learning by doing with capabilities building rather than skills acquisition at its centre, as explored in our Community of Practice on 21^st Century Government Learning. In this AI example, this means targeting how civil servants use AI and for what, and designing the capabilities building needed into the tool itself.

• What Works in Learning Bootcamp: Banishing learning and innovation myths which have been debunked (like “learning styles”), but are often still used. Together with building innovation practices into the learning and design with a full-system and full-process lens, and drawing on evidence of why most institutional learning initiatives fail (on average, only 10-15% of “training” is translated into the workplace), to create more applied tools like those proposed here, which bring the learning and doing of the work of government together in an applied way.

• Leapfrogging, not lagging, through comparators analysis: Analysing why other systems have been effective – Harvard Kennedy School and its NSG-style relationship with the US Government has two major relevant themes – case-based, experiential learning and innovation and SLATE, which provides ongoing teaching and learning innovation support. ENAP Brazil is a full learning ecosystem with promotion structures built into the learning, policy lab integration and strong international connections. ANZSOG has a federated system of partnerships through government levels and universities. 

• Building on and incorporating what works right now in the UK learning ecosystem: at the local level, and for learning and innovating the “how to” for Moving Fast and Fixing Things. Additionally, drawing on public administration rather than public policy focus, in universities like Glasgow, Sheffield, Birmingham, and how they work hand in hand with local authorities and devolved administrations on building innovation practices as a framework for how the NSGPS at the national level might learn from this successful practice. 

Getting moving

In his speech, Darren Jones also noted that people were right to be sceptical about politicians claiming to fix the state and then offering incremental change. He suggested that the Move Fast and Fix Things approach is a possible antidote to this scepticism. We agree, and in this blog we’ve offered three key ideas for how to get started.

First, start by really understanding the problem, from all angles, not just the Whitehall perspective, before deciding what ‘fixing it’ really means. Second, make the most of AI’s ability to accelerate change by bringing together and augmenting expertise instead of replacing or automating it. Finally, create a National School of Government and Public Service that rapidly fosters the kind of learning that this new approach to policy requires.

We don’t claim that these three ideas are the only ones required to Move Fast and Fix Things. But we do believe that they will at least allow the government to get moving in the right direction.