The centre is from Mars, the edges are from Venus

Introduction

I remember in the 1990s, there was a book on my parents’ bookshelves — and, it seemed, on all parents’ bookshelves — called Men Are from Mars, Women Are from Venus. I must say I’ve never read it, and I doubt my parents have either, but AI summaries tell me it’s as daft as it sounds. The idea is that men and women see the world so differently, and speak such different languages, that they might as well be from different planets. The author, John Gray, argues that a good relationship must start by recognising these differences, and then work with them to make them productive.

The book has been widely discredited, but I imagine it sold well because the premise rings true. We see all the time that people hold profoundly different worldviews — in our workplaces, in politics, and in romantic relationships — and this shows up in the way we communicate.

Indeed, if you push aside the more essentialist claims — “women are like waves” with their “shifting moods”, and men will “retreat to their caves” — you’ll find a deep academic literature on these topics. There are sociologists, like Ludwik Fleck, who argued that we each have our own “thought style”, an intricate and internally coherent mental world, which we inhabit, and which shapes not just our judgements but all cognition. There are also studies in thinking and meaning, from disciplines like neuroscience and semiotics, which show that language and thought are deeply intertwined, and even mutually constitutive. When we speak differently, we think differently, and vice versa.

For the past few months Kinship Works and Joseph Rowntree Foundation have been running an initiative called the Centre for the Edge. We are exploring the gap between the institutions at the centre of public life, which often seem so tired and outmoded, and the vibrant innovation that is happening all around the edges. More specifically, we are asking: how could the centre support and spread promising alternatives at the edges?

As we’ve gone about this work, the issue of divergent worldviews has never been far away. Indeed, it has often come up when we’ve debated other, more tangible topics. Two people might seem to be arguing over different evaluation methods, or different ways to structure an institution, but really they have deeper disagreements — about what constitutes evidence, or the kind of thing government is, and should be.

In this post I want to address more directly the differences in worldview between people in the centre of government and people innovating around the edges. Can we get beyond generalisations to name these differences? Is there evidence we can bring to bear on this? And can we develop concrete proposals for how to bridge incompatible worldviews, or for making differences less of a barrier to renewal? Or perhaps we can even find ways to harness differences as an energy source for public sector reform.

Part 1: Different, but how?

It’s all too easy for a conversation about worldviews to be anecdotal. We end up making sweeping claims based on little more than our own prejudices. To avoid this, we’ve used a method called discourse analysis to compare the language people use at the centre and at the edges, to see what this reveals about the ways people think. I’ll describe the method and its limits, and then I’ll share the headline findings, before making some recommendations. As ever, this is all shared in a spirit of debate and I’d love to hear feedback.

We started by choosing around 30 organisations that are typical of the centre — HMT, Cabinet Office, National Audit Office, and so on — and a similar number that reflect innovation at the edges, such as Platform Places, the Doughnut Economics Action Lab and Capacity. We then collected around five documents for each organisation, mostly formal publications and, in some cases, text used to self-describe an organisation’s work or governance. We used a script run locally to convert these documents into a corpus of raw text files. Then we used Anthropic’s Opus 4.7 large language model to compare the text in each set of documents.

We focused on differences that are likely to play out in the working relationship between the centre and the edges. We started by scanning for metaphors people use to describe their work and the systems they work in. Then we scanned for the language people use to describe six key concepts:

• Change and scale

• Success

• Failure

• People

• Accountability and legitimacy

• Knowledge and evidence

Three other themes also emerged more organically from the analysis, which we then dug into. These relate to:

1. The emotional register of each side of the corpus.

2. Examples of the centre or the edges adopting each other’s language.

3. The degree of mutual awareness between the two.

Before I share the findings, I want to add some caveats. I would describe this as an experimental analysis. The version I’m sharing here is something like a public beta; we are publishing it to stress-test the findings and to improve the method. We have already gone through several iterations. For example, the early cuts were prone to confirmation bias, so we adapted the method. We separated a first stage of raw analysis — counts of words and phrases; curation of verbatim quotes — from a secondary interpretative phase, to make the results more objective and auditable.

We also added layers of checks: checking the validity of the raw data, and then cross-checking the analysis by running it blind through another LLM, OpenAI’s ChatGPT 5.5. The results were essentially identical, but this test revealed an issue with the way quotes were extracted, which we corrected. We have therefore also made everything public, saving the data and documentation in an open repository on GitHub, so that people can check and improve the method, see if the findings replicate, or run complementary analysis.

With these caveats, how does language differ between the centre and the edges? We have published the raw results without commentary here, but here are the headlines that jump out to me:

• The analysis seems to confirm that there are differences in worldview between the centre and the edges, at least as reflected in language. The two groups don’t just have different preferences within a shared frame; the differences go deeper. The centre and the edges use different metaphors to describe the systems they work in, which imply different logics of public action. The centre’s logic is one of production; work is “delivered” — this word is 2.7x more common in centre documents compared to edge documents — inputs are turned into outputs, and the goal is to find and optimise answers. For the edges, the system is a “community” or “ecosystem” — the word “community” is 7x more common in edge documents; “relationship” 5x more common; “ecology” 30x more common — and the goal is to cultivate, learn and spread; to make the system healthier; to bring about shifts in power and agency; or to build movements.

• There are stark differences in the way the centre and the edges think about scaling. The centre uses phrases like “roll out” — 3.4x more than the edges — as well as “milestones” — 5.7x — “targets” — 3.9x — “pilot” — 2.0x — and “implement” — 1.9x. By contrast, organisations at the edges talk about scaling in a language of “systems change” — 3.9x more than the centre — “movements” — 3.9x — “community-led change” — 23x — “paradigms” — 5.7x — and “field-building”, which appears only at the edges.

• There are also differences in attitudes towards accountability. Although both the centre and the edges use the word “accountability” at almost the same rate — 4.17 per 10k words in the centre, versus 3.96 at the edge — each has something different in mind. The word “audit” appears far more often in centre than edge documents — 38x — and at the edges the word is used mostly in reference to the centre’s audit machinery. The centre also talks about accountability using words like “inspection” — 103x more than the edges — “assurance” — 12x — and “scrutiny” — 4.2x. The edges use a different vocabulary; edge organisations talk about “power-sharing” — 40x more than the centre — “democracy” — 19x — “legitimacy” — 10x — and “deliberation” — 9x.

• There is a big difference in emotional register. Here, it really does feel like the centre and the edges are from different planets. In the centre, the vocabulary is anxious and often pertains to control: we see words like “risk” — 2.8x more than the edges — “error” — 13.6x — “compliance” — 10x — “protect” — 5.9x — “monitor” — 2.7x — and “manage” — 2.0x. Edge documents, meanwhile, have an emotional register that is more positive and aspirational: we see words like “inspire” — 4.9x more than the centre — “hope” — 10.5x — “imagination” — 10.9x — “belonging” — 8.9x — “celebrate” — 5.1x — “love” — 3.9x — “dignity” — 3.7x — and “flourish” — 2.2x. This reflects a deeper pattern; centre organisations seem to focus more on “how” things are done, while edge organisations focus on “why”.

• Both the centre and the edges refer to expertise and empirical authority, using words like “research” and “data” at similar rates. However, each appeals to different epistemologies. In the centre, confidence comes from retrospective evaluation; people ask: did the “intervention” or the “policy” work? At the edges, expertise, insight and capability are built over time through practice, relationships and community knowledge. The word “learning” is used 2.7x more often at the edges, and it is understood differently too: rather than being post-hoc — “did it work?” — learning at the edges is an iterative and social process, and is concurrent with action.

• The edges are far more interested in the centre than the centre is in the edges. Edge documents are full of references to the centre, and edge organisations often position themselves in relation to the centre. The words “Whitehall” and “Westminster” appear more often in edge documents than in centre documents. Edge organisations talk a lot about the centre’s processes and qualities; the word “bureaucratic” appears more at the edges — 2.8x — as does “top-down” — 4.6x — and “siloed” — 3.1x — implying that edge organisations carry most of the burden of translating or bridging between the centre and edges. The centre, by contrast, barely mentions edge organisations or techniques, even when these are considered mainstream: “co-production” is 6x more common at the edges, “place-based” 10x, “lived experience” 9x, “relational” 5x, and even “engagement” is 4x, while “asset-based” is almost exclusive to the edges.

• The two groups of organisations describe people in different ways. In the centre, people are defined by their relation to government. People are “taxpayers” — used 64x more than at the edges — or “claimants”, which appears only at the centre, but also “consumers” — 6.8x. At other times, people are called simply “members of the public”, which is implied as being distinct from, or outside, government. At the edges, people tend to be described in ways that relate to places or engagement: “residents” is 8x more common at the edges, “participants” 8.6x, and holders of “lived experience” 9.5x. The word “citizens” appears 3x more often at the edges, and means something different: in the centre, citizens “receive better outcomes”; at the edge, citizens offer insight and participate.

• Sometimes it seems the centre adopts edge language, while retaining a wider centre vocabulary. For example, the word “mission” appears 478 times in centre documents, reflecting the UK Labour government’s commitment, during the period covered, to mission-driven working. Despite this, the word “mission” remains surrounded by the centre’s default operational vocabulary; words like “implementation” and “milestones”. Missions are “specified” and “delivered”.

• There is a common core of language shared by the centre and the edges. Words like “innovation” appear with near-parity across the two sets of documents — a centre-to-edge ratio of 0.98 — as do words like “transformation” — 1.28 — and “wellbeing” — 1.39 — which skew slightly to the centre, and “partnership” — 0.70 — and “governance” — 0.51 — which skew slightly to the edges. In some cases, however, the same word has different connotations. For example, “innovation” in the centre tends to refer to digital technology, while at the edges “innovation” more often relates to novel ways of organising work, such as different forms of institutions.

What should we make of these findings? To me, they are quite generative, and I’ll explore where they could take us in a series of recommendations in a moment. First, I would anticipate a couple of pushbacks. Some people might say the findings are unsurprising — of course centre organisations have different worldviews to those at the edges; they have different histories, face different pressures, and employ different kinds of people. Others might say we have biased our analysis by choosing edge organisations that are non-traditional, and centre organisations that are traditional.

To respond to these concerns, it’s worth returning to the point of the analysis. The idea is not to generate shock at a surprising gap in worldviews between the centre and the edges. The point is to bring specificity and evidence to differences that people often observe, but which are often stated in vague or anecdotal terms.

We are aware that the differences we find will be a function of many things, including the different contexts and constraints faced by organisations, their different histories, and the backgrounds and training of staff. Indeed, it would be interesting to explore these drivers more. We are intentionally focused on innovators at the edges, as opposed to looking simply at small or local organisations, because our ultimate concern is institutional renewal.

Our hope ultimately is that, with a better understanding of the differences in worldview between centre organisations and edge innovators, we can develop a more intentional relationship between the organisations at the centre of public life and those at the edges. The goal is to be less swept up in our differences — like a couple arguing in a failing marriage — and instead name the differences calmly, assuming best intentions on all sides. We can then develop strategies for working together towards institutional renewal: bridging our differences, or reducing the friction they cause, or perhaps even leveraging them as an agent of change.

So let’s turn to what these strategies could be.

Part 2: What can we do?

I’m going to make seven recommendations, based on the insights above. I’ll try to be clear and specific, but I hold the suggestions loosely and see them as starting points for discussion.

The first two are about increasing empathy between the centre and the edges, and finding ways to share the load that each “side” is currently carrying.

1. The centre should take on more of the burden of translation and bridging

The centre is not investing enough in its relationship with the edges. It leaves the edges to do most of the work of translation and bridge-building. One way to correct this is for people in the centre to pay more attention to the edges — making an active effort to see the work that is happening, but also to become attuned to its success conditions. What does edge innovation need to succeed?

Every government department should make it clear whose job it is to lead on this work. In all departments there should be functions tasked with mapping the work at the edges and sharing what they learn. More importantly, understanding the edges should be part of everybody’s job. All civil servants should spend far more time at the edges of the systems they work in — in education, healthcare, criminal justice, the environment, and so on. Leaders should encourage this, and mandate it if needed. Civil servants should be guided in how they approach these visits: they are about listening, not broadcasting. Sharing insights from visits and edge conversations should be made a prominent part of the working week.

It is worth a special mention of the best existing efforts to achieve this kind of attunement. Changing Futures is one example, and the current government’s programmes Test, Learn and Grow and Pride in Place are attempting a similar act of translation or bridge-building, albeit often resisted by the system around them.

The goal ultimately should be for all civil servants to see that being attuned to the edges of the system they work in is a central aspect of high performance.

2. People at the edges should help with the centre’s challenges, especially on accountability

People working at the edges should more actively engage with the constraints people face in the centre, especially with regard to accountability and related processes: risk management, procurement, delivery and reporting architectures, and so on.

The goal is not for people at the edges to internalise these pressures and constrain their own practice — work at the edges is powerful partly because it is somewhat protected from pressures at the centre. Instead, understanding the accountability environment in the centre can improve work at the edges; for example, it can mean it docks better into the system and is easier to spread.

More importantly, people at the edges could help to solve the centre’s challenges, by sharing the contemporary methods that are common at the edges, but rare in the centre: contemporary ways to manage risk, agile delivery, dispersed and iterative approaches to learning and evidence, and so on.

An architecture should be created by the centre, to work collaboratively on the most important of these topics. The UK’s new Cabinet Secretary should sponsor a series of mixed-discipline transformation teams, staffed with people from the centre and the edges. They should start by working on modernisation proposals for the most outmoded aspects of the accountability and control infrastructures, starting with risk management and procurement. This could complement the Cabinet Office’s Test, Learn and Grow programme, which is already exploring a series of issues that constrain edge practices, and is looking at how these parts of the system could be rewired.

To complement all of this, doors should be opened at the centre so that people at the edges can spend time working there — both to learn more about the work and to make a meaningful contribution to modernising government. These efforts should be supported by the centre via a massive expansion of inward secondments. These have been used to good effect in some places, but they remain rare and are still hugely burdensome for teams to organise.

The next three recommendations focus more directly on the frictions that arise from jarring or outmoded worldviews.

3. Develop shared frameworks for scaling and evidence

There are especially stark and problematic differences in worldview in two areas: scaling and evidence. These differences should be made explicit and people should collaborate across the centre and edges to develop shared language and frameworks. This work should start by focusing on two tricky dimensions:

1. What it means to spread or scale impact.

2. What constitutes good evidence.

The goal of a shared framework is not to resolve the different worldviews and decide which is better, but to contextualise and accommodate these differences. For example, the framework would describe a range of ways to think about scaling or evidence — some more “centrist” and some more “edgy” — and their respective strengths and weaknesses. Even if the frameworks failed, the act of trying to develop them together would foster dialogue and raise mutual understanding.

4. Retire outmoded mentalities

It might sound like I’m trying to keep everyone happy by retaining all of the different mentalities between the centre and edges, and applying them wherever they fit best. I should therefore come off the fence and say that it’s clear that some of the centre’s worldviews are irresponsibly outdated. I would name as examples:

• The deep attachment to mechanical mental models: levers, apparatus, scaling as “roll out”. These ways of thinking originate with the formation of bureaucracy 150+ years ago, and are now outmoded and unhelpful.

• The reliance on outdated control mechanisms, for example for procurement and risk management, and the category error of applying these mechanisms to qualitatively different types of work, like care or field-building.

• A compliance-oriented approach to management.

These are just a few examples of mentalities and methods that are straightforwardly outmoded. They run decades behind the frontier of practice and are no longer shared by any serious scholars in the relevant academic fields. There should be energetic programmes of training and culture change to retire these mentalities, especially in important “wiring” professions like procurement, project management and risk management.

The UK’s recent decision to create a National School of Government will help, but only if the NSG leans hard into contemporary methods. To complement training and culture change, there should be an explicit drive to promote new talent more quickly into the Senior Civil Service and make space by retiring senior people with outmoded attitudes.

5. Calm down

This recommendation might sound flippant, but the anxiety and loss of confidence we are seeing at the centre of government is a first-order problem. It is manifesting in efforts to control risk through compliance, which is having exactly the effect that decisions made in panic often have, which is that they backfire.

Whether you are parenting, or swimming, or playing a musical instrument, if you panic, you will tense up, causing the very thing you are worried about. You will drown because you are panicking.

This outcome — drowning because we are panicking — is happening today, all over government. People are anxious, and so they are either flailing, or freezing up, and this is inhibiting the work we need to do to save our struggling public institutions. In particular, it means government is less open to alternatives because the first question in their mind is: what if it goes wrong?

What can we do about this? In part, we need Whitehall to do the equivalent of taking a deep breath. We need calmer and more composed Ministers and we need senior civil servants who are skilled at the art of protecting space for their teams to work in. We also need to take chronic workplace stress more seriously. If you are rushing from meeting to meeting, without breaks, it is likely you will be an anxious presence.

We need to train civil service leaders in techniques that are now used widely across the private sector — mindful leadership, and so on — to build calm, intentional working environments. And, in addition to all of this, we need to upskill the civil service in more sophisticated control mechanisms that actively use difference and delegation as a way to reduce system-level risk. This could include enhanced managerial discretion, cultivating capability at the frontline, creating functions for collective and distributed intelligence, or embracing more adaptive and agile operating models and organisational design.

The final two recommendations are about making the system more tolerant of different worldviews, so that difference doesn’t block the work of change, and perhaps even drives it.

6. Push for more pluralism in government

Government should be made more pluralist, welcoming a wider range of methods and mentalities. We tend to think this is obvious when applied to the civic sphere more generally — pluralism is accepted as a virtue — but we seem okay with government being homogeneous. This comfort with homogeneity, and intolerance of difference, is one of the most dangerous legacies of the bureaucratic age.

There are many ways to make government more pluralist. One is to substantially expand underrepresented disciplines. The civil service already has a target to reach 10 percent of total headcount in digital disciplines. It should also have a target to get to a better balance of disciplines, reducing the dominance of economics and old-school policy-making. We can also help each discipline to become more contemporary and diverse: fewer PPE graduates, more secondments from outside, and so on.

We should also make it easier for people with diverse views to work together, across differences. An example is the great work done by Policy Lab to support mixed-discipline teams, which create the conditions for productive disagreement. Departments are moving, albeit achingly slowly, to operating models that embed mixed-discipline teams as standard. There should be a much stronger push from the Cabinet Office to accelerate this work.

A final way to make government more tolerant of difference is to create alternative “modes” for the system to operate in. This helps to avoid situations where alternative worldviews are blocked because they jar with so-called “enabling” functions like finance or procurement.

For example, last year I fed into some work in New South Wales, Australia, to develop a “Mode B” operating environment. The idea is to acknowledge that different kinds of work require different mentalities and methods, and these in turn require different operating environments. If you are procuring a care service, for instance, you should bring a different mindset, and use a different procurement process, than if you are bulk-buying toilet rolls. Likewise, if you are writing a business case to build a bridge, the mentality and process should be different to if you are developing a new service to help manage diabetes.

In NSW, we explored the idea of developing a Mode B operating environment — a pre-authorised set of alternative processes, including sign-off, team formation, risk management, governance and financing, that would be activated when work requires it. If you encountered an especially complex problem, you would “activate Mode B”, meaning you would use the alternative set of processes without having to negotiate them all from scratch with the relevant teams: finance, procurement, Project Management Office. The end result would be a system more tolerant of the different mentalities that are inevitably needed, given the diversity of work government now has to do.

7. Brute force it with creative destruction

We need to recognise how difficult it is to change a person’s worldviews, or even to bridge between such differences. We should therefore complement the above recommendations by making the reform of government less dependent on having to persuade people to change their minds. This is essentially about creating dynamics that drive outdated and low-impact worldviews out of business; setting up a fair fight, and seeing who wins.

What we’re talking about here is making the system more outcome-based and dynamic. The most direct way to do this is to foster creative destruction. I won’t unpack this here because I did so at length in a previous essay on institutional diversity.

The one point I’ll make here is that this is not simply about introducing market competition. Indeed, in many cases market competition has inhibited diversity of thought in the public sector because old-fashioned firms — Capita, Serco, CGI — compete at the margin, which keeps out the real innovators. We can therefore go beyond competitive procurement and commissioning to use a range of other approaches: outcomes-based forms of accountability, dynamic finance, impact investment, social enterprise, and newer adaptive mechanisms like relational contracting, or techniques associated with purpose-driven management and missions.

Conclusion

I’ve covered a lot of ground, so it might be worth circling back to the central point. In theory, at least, there could be a very productive relationship between the struggling centre of government and innovators at the edges. The latter could provide a way of renewing the former. However, often it seems that differences of worldview block this process of institutional renewal. The two sets of work end up either feeling very separate, or interactions between them generate lots of friction and frustration.

We have shared a discourse analysis that gives us a more granular sense of the differences of worldview that exist between the centre and the edges. What we see are many differences of worldview, some of which run very deep, and which are likely to explain many of the tensions we see in day-to-day working relationships.

In a series of recommendations, I’ve shared a range of different strategies for overcoming these differences, or working with them. Some are about bridging between the two worlds with a sense of empathy and shared endeavour. Some are about accommodating views that are different but both valid, if only they are used in the right contexts; while others are about tackling outmoded views. Finally, there are recommendations to make the system as a whole more tolerant of, and capable of harnessing, difference, and less reliant on having to persuade people to change their minds.

I’d love to hear people’s thoughts on the method, findings and recommendations. With the launch of our new website, we are moving into a new phase for the Centre for the Edge, in which we’re actively seeking partners to broaden and deepen the work. We’re also considering how best to share useful resources and build practitioner communities around this work. Drop me a line if you have thoughts, or would like to partner, on any of this.

We’ll write more about our future plans soon. In the meantime, our new website has links to all of our previous essays, on the themes of evidence and institutional diversity, as well as wider material about the Centre for the Edge initiative.