Speculative Design for Policymakers: Prototyping Policy for the Decades It Will Govern

Speculative design gives policymakers and public servants a way to see a policy from inside the decades it will govern. It takes the futures work government already does — horizon scanning, scenarios, the future wheel — and adds the making: a concrete artefact from a future in which the policy is in force, such as the letter a citizen receives. Backcasting from that artefact shows which decisions sit in the department's own hands.

Government takes the future seriously and has the guidance to prove it. What it rarely does is make anything. A policy is a design for how part of life will work, written in prose, which hides how it will feel. Ministers read submissions; nobody holds the thing the policy will become.

This guide covers what is particular to practising speculative design inside the state. The full sequence of stages is set out in the workflow we teach.

What government already has: horizon scanning, the Futures Toolkit and the Wales Act

The front half of the method is already published as government guidance. The Government Office for Science's Futures Toolkit (2024 version) sets out twelve tools for policy officials and analysts, among them Horizon Scanning, Scenarios, Futures Wheels, Policy Stress-testing and Backcasting. Officials who have used it already know most of this guide's vocabulary.

In Wales, thinking about the future is a legal duty. The Well-being of Future Generations (Wales) Act 2015 requires public bodies, in acting on its sustainable development principle, to take account of five ways of working: long term, prevention, integration, collaboration and involvement. Welsh Ministers must publish a future trends report within twelve months of each Senedd election.

Then there is the ordinary machinery: the green or white paper, options appraisal, the impact assessment, the consultation document, the ministerial submission, the post-implementation review. Each asks what a policy will do; none asks it in a form a person can pick up. Speculative design does not ask government to adopt a new discipline, only to add one step between the scenario and the appraisal: make the artefact.

The horizon problem: policies outlast parliaments

A pensions settlement, a land-use framework or a building standard will be administered by governments not yet elected and felt by people who cannot yet vote. Its scrutiny is short. Ministers move on, spending reviews cover a few years, and the political calendar runs to the next election.

Appraisal compounds this, legitimately. HM Treasury's Green Book discounts future costs and benefits so they can be compared with present ones. In October 2026 the Treasury lowered the discount rate from 3.5% to 3% for the first thirty years, falling to 2.5% from year 31 and 2.25% from year 76. Even at the new rate, a benefit arriving in year thirty counts for about 41% of the same benefit today. That is defensible. It also means that whatever a policy does in its third decade arrives in the submission small.

Speculative design does not argue with the discount rate. It puts the third decade into the room as something specific, so the people deciding can see what the arithmetic has shrunk. The window we recommend, ten to twenty-five years out, is where most policies are still in force and few of the people who decided them are still in post.

Where speculative design fits in the policy cycle

Before options close. Once a preferred option is in a submission, futures work tends to become justification, which is a different and less useful activity.

A worked example: a pension letter for 2046

Take a hypothetical team of analysts and policy officials in a UK pensions department, gathering evidence for the next review of State Pension age. The Pensions Act 2014 requires a new review report within six years of the last, with the Government Actuary advising whether people reaching pensionable age can expect to spend a set proportion of adult life in retirement.

The brief. Rising state pension ages (the source of change), people in physically demanding work in their sixties (the domain), 2046 (the horizon), for ministers, employers and the people now in their forties whom the decision will reach.

Signals. Four dated events:

• •On 16 March 2023 the French government used Article 49.3 of the constitution to pass, without a vote in the National Assembly, a reform raising the minimum retirement age from 62 to 64.
• •On 13 September 2024 China announced that its retirement ages would rise gradually over fifteen years from 1 January 2025, to 63 for men and to 55 or 58 for women.
• •In May 2025 Denmark confirmed that its state pension age will rise to 70 by 2040.
• •On 6 May 2026 the first people born after 5 April 1960 reached UK State Pension age, at 66 and one month. The rise to 67 legislated in 2014 had begun.

Read together: three countries raising pension ages within three years, one of them forcing it through without a vote. The arithmetic travels between countries easily. Consent does not.

Future wheel. At the centre: a state pension age of 70 is normal across Europe. The first ring is the fiscal case: longer working lives, a pension bill that grows more slowly. The second ring is where the policy problem lives. People in physically demanding jobs, in care, construction and warehousing, cannot all work to 70, so some move onto sickness and disability benefits in their sixties and part of the saving reappears in another department's budget. In the third ring the "proportion of adult life in retirement" becomes a political question about place and class, because a national average says nothing about the people below it. Younger workers, having seen the age move more than once, stop planning around it.

The concept. For policymakers the most useful artefact is usually a document, because documents are how the state reaches people. The team writes one:

Bridge Years — "Your pension, when your work says so." From 2046, anyone with thirty years of recorded work in occupations on a published list of physically demanding jobs can draw a reduced State Pension for up to three years before State Pension age. Eligibility comes from work records, not a medical examination. The list is revised at every statutory review.

The sketch is the letter: one page, dated 2046, to a 63-year-old care worker. It lists the years the department has on record, the three dates she could start from and the weekly amount at each. Near the bottom, one line explains that four of her years do not count, because no employer recorded them.

Implications. Start with the person reading it: told by the state that her work has worn her out, in a letter that also tells her four years of that work did not happen. Then outwards: which jobs make the list, and the lobbying to get onto it; whether delivery riders and the self-employed are in or out. Politically, France is the warning that a pension-age change can be lawful and still corrode trust. How it goes wrong: the list freezes the labour market of the year it was drawn, and Bridge Years becomes the department's most appealed decision.

Backcasting. Working back from 2046, the rungs include a threshold (State Pension age passing 68) and a run of legislative and institutional changes: primary legislation for an early partial pension; a reliable record of what work people did, not only what they earned; an agreed list of demanding occupations and a body to keep it; an account that sets disability spending against pension savings across departments. In most speculative projects these are the rungs where futures die, because the people doing the work do not hold them. This team holds several. The 2014 Act lets the Secretary of State add factors beyond life expectancy to the review, and the department drafts that advice.

What changes this quarter. The team drafts advice asking ministers to add differences between occupations to the review. It commissions analysis of how many people in their early sixties are still in physically demanding work. It takes the 2046 letter into user research with people in their forties in care, construction and logistics, asking what they would do the day it arrived and what is wrong with it. And it writes the backcast into the submission, with the disability-benefit effect named as a risk and the other department's name next to it.

Where speculative design goes wrong in government

Consultation theatre. A polished future shown to the public after the preferred option is chosen is not consultation. It is communication with better typography. Future artefacts belong in engagement while options are open, with questions that can change them and a published record of what changed.

Futures without a mandate. Backcasting is normative: you choose the future you reason back from. Inside government that choice is not the unit's to make. A small team deciding which future a population should want is standing in for politics. Offer more than one future, say whose each is, and involve the people it would happen to. In Wales involvement is one of the Act's five ways of working; elsewhere it is still the condition for legitimacy.

The scenario that becomes a forecast on the way up. Language hardens as it climbs. "One plausible future" in the workshop becomes "the expected position by 2046" in the submission. The case against forecast-led strategy applies with extra force when the reader is a minister. Write the caveat into the artefact itself.

The futures report nobody owns. Horizon scanning produced in a strategy unit and never connected to the appraisal or the bill team. This is the absorption problem we describe in the seven ways speculative design fails inside organisations. Every lever marked on the backcast should have a team's name next to it.

A first exercise for a policy team

1. 1.Pick a policy you are working on that will still be in force in twenty years.
2. 2.Write down that year, and count the elections likely between now and then.
3. 3.Collect three signals, each with a who, a when and a what, preferably from jurisdictions that have already moved.
4. 4.Run a future wheel on one of them, assumed fully realised. It is in the Futures Toolkit, so a colleague may already know it. Push it to the second ring at least.
5. 5.Write one artefact the state would send to a citizen in that year under your policy: a letter, a decision notice, a form. Use your department's real template. One page.
6. 6.Show it to three colleagues, one from operational delivery, one from legal and one from outside your team. Ask what would have to be true for it to be sent, and who would refuse to sign it.
7. 7.Turn the answers into rungs, mark the ones in your department's gift, and put one into your next submission or options paper as a named risk or sensitivity.

Where public servants can learn speculative design

Speculative Design Basics is self-paced and online, with lifetime access, and it assumes no background in design. For policymakers the parts that matter most are the Process chapter on time horizons; the Tools & Methods chapters on backcasting, future prototyping, signals and future wheels; and the Creating Value chapter on how speculative design relates to foresight and strategy. The Ethical Imperative chapter matters more here than for most roles. Policy rarely stops at the document, and our guide on speculative design for service designers picks up where a letter like Bridge Years becomes a service. Other roles are collected in speculative design for every role.

Key takeaways

• •Government already owns the front half of the method, futures wheels and backcasting included. What it lacks is the step where someone makes the artefact.
• •Policies outlast parliaments, and appraisal shrinks their third decade: at the Green Book's 3% rate, a benefit in year thirty counts for about 41% of the same benefit today.
• •The most useful policy artefact is usually a document, because that is how the state reaches people.
• •In backcasting the regulatory rung is where most futures die. For policymakers it is the rung they hold.
• •Use future artefacts in engagement before options close, or it is consultation theatre.
• •Offer several futures and say whose they are. Choosing the public's future is not a futures unit's job.