Primer · Metascience and Decentralized Science

Scientific Roadmapping

David Hilmer Rex · June 2026

Summary

A scientific roadmap is a structured document that charts the path from the current state of knowledge or capability to a defined scientific or technological goal - identifying the gaps that must be closed, the dependencies between them, and the sequence in which they should be addressed. Distinct from a project plan (which assigns tasks to people) or a strategy document (which sets direction), a roadmap is primarily an epistemological object: it represents what needs to be true, discovered, or built, and in what order, for a goal to become achievable. When done well, roadmaps coordinate effort across labs and funders, surface the most important bottlenecks, and make it possible to identify leverage. Operating upstream of or alongside roadmaps are related formats: vision papers establish the why before the path is clear; mission maps coordinate multiple actors across sectors around a shared societal goal; grand challenge statements identify the most important open problems.

What is a scientific roadmap? ↑

The term “roadmap” has been used loosely enough that it covers everything from a glossy priorities document to a rigorous decomposition of technical dependencies. The useful definition is narrower: a roadmap identifies a goal, the current state of knowledge or capability relative to that goal, the gaps between the two, the dependencies among those gaps, and - ideally - the metrics or criteria that would signal that each gap has been closed.

What a roadmap is not: a list of research priorities, a funding wish list, a project plan, or a prediction. A roadmap is not a forecast of what will happen; it is a structured account of what would need to happen for a specific outcome to become possible. That distinction matters because it makes a roadmap falsifiable and revisable: as progress is made or surprises emerge, the map can be updated.

The practice has deep roots in semiconductor development (the IRDS and its predecessor the ITRS set the template for coordinating an entire industry around shared technical targets)¹ but has been substantially retheorised for biomedical science, climate technology, and other fields where the product is a public good or a platform rather than a commercial device.

A note on scope. Technology roadmapping is a broader, older field with its own established methodology - the Science-Technology-Product (S-T-P) layer structure, workshop-based rapid roadmapping processes, and a substantial academic literature.¹⁴ This resource focuses on the scientific branch: contexts where the output is knowledge or capability rather than a commercial product, and where the intended audience is a research field, a funder, or a public-good mission rather than a market. The two traditions overlap (the IRDS sits in both), but the concerns, methods, and literature differ enough to treat separately.

Functions and uses ↑

Roadmaps serve different functions depending on who is building them and for whom. The main uses:

Coordination across a fragmented field

When a scientific goal requires contributions from many labs, organisations, and disciplines that do not naturally communicate, a roadmap creates a shared reference frame. Teams working on different sub-problems can see how their work connects to others’ and where the current bottlenecks lie.

Surfacing bottlenecks for funders

A funder who wants to accelerate a field needs to know not just what is being worked on but what is not being worked on that everything else depends on. Roadmaps make dependency structure visible, allowing funders to identify the places where a targeted grant would unblock the most downstream progress. This is the logic behind Convergent Research’s Fundamental Development Gap Map.²

Scoping a new organisation or programme

FROs, DARPA programmes, and philanthropic initiatives are often built around a specific gap in a roadmap. The roadmap provides the justification for the scope: why this problem, why now, why at this scale, and what the organisation will hand off when it is done.

Agenda-setting and field-building

A published roadmap from a credible source signals to researchers, funders, and policymakers that a problem space is tractable. It attracts talent and shapes the questions that grant programmes are built around. Renaissance Philanthropy’s work on agenda-setting treats roadmaps as a core tool of field creation.³

Types of roadmap ↑

Technology roadmaps

Chart the evolution of specific technical capabilities toward defined performance targets, typically over a 5–20 year horizon. The IRDS is the canonical example. Spec Tech’s roadmaps apply the same structure to frontier deep-tech domains.⁴

Key feature: Quantitative performance targets at defined time horizons. Dependencies shown between capability areas.

Research roadmaps / gap maps

Chart the scientific knowledge that needs to be generated. Identify the specific questions that remain open and the experiments or observations that would close them. Convergent Research’s Fundamental Development Gap Map is a living, crowd-sourced version of this for biology and adjacent fields.

Key feature: Questions framed as gaps; dependencies between questions made explicit. Closer to a literature review than a Gantt chart.

Mission roadmaps (backward-from-goal)

Start from a specific desired outcome and work backward to identify what must be true for it to be achieved. The Apollo programme is the archetype; in contemporary science, the DOE Fusion Roadmap⁵ and ARPA-H’s⁶ programme structure work this way.

Key feature: End state defined first; all intermediate steps derive their justification from their necessity for that end state.

Field roadmaps

Map the state of an entire scientific domain: what is known, what is contested, what is unknown and tractable, and what is unknown and currently intractable. Often produced by working groups or consortia. The EBRC Engineering Biology Roadmaps are a good example.⁷

Key feature: Field-wide scope; produced collaboratively; updated periodically.

Roadmapping across scales ↑

What counts as a good roadmap depends heavily on who is building it and for whom. The appropriate scope, process, governance, and level of detail differ sharply across scales - from a researcher mapping their own dissertation problem to a transnational coalition coordinating a field spanning dozens of countries. The same method applied at the wrong scale produces either noise (too granular) or vapour (too abstract).

Individual

A researcher mapping their own scientific agenda: identifying the most important open questions in their area, the dependency structure among them, and where they could have the highest leverage. Common uses include PhD topic selection, postdoc strategy, scoping a grant proposal, and deciding what to read next.

At this scale, a roadmap is primarily an epistemic tool for the individual - clarifying what needs to be true before a desired result becomes achievable. The Tiling Tree Method⁸ is most legible here: one person can hold the whole structure in view and update it as they learn. The failure mode is conflating “I find this interesting” with “this is the highest-leverage gap.”

Key discipline: Individual roadmaps often over-index on the researcher’s existing skills and under-index on what the field actually needs next.

Organization (lab, institute, FRO, company)

A research organisation mapping the technical problem space its work addresses: dependencies between capabilities, gaps that are unoccupied, the specific bottleneck the organisation is built to address. Common uses include FRO scoping and founding documents, programme design at DARPA-model agencies, and R&D portfolio management.

This is the scale at which most current methodological development has occurred. Convergent Research’s Fundamental Development Gap Map and Spec Tech’s deep-tech roadmaps are both organisational-scale instruments. The roadmap at this scale defines what the organisation does and does not do - especially important for mission-driven organisations where scope creep is a recurring failure mode.

Key tools: Target Product Profiles set the end state; Tiling Tree maps the path; gap maps locate the organisation’s work in the broader field.

University (or department / research centre)

A university or department mapping its strategic research priorities: where it has distinctive capability, where it lacks it, what partnerships it needs to build. Common uses include strategic plans, cluster hiring decisions, and research centre formation.

University roadmaps face distinctive constraints: academic governance is slow, faculty autonomy is high, and the political cost of excluding areas is large. The result is often plans so inclusive they provide no real guidance. The most effective university-level roadmaps are produced by smaller units - a department, a centre, an interdisciplinary institute - where governance is simpler and problem scope is narrower.

Key risk: A strategic plan that lists every faculty member’s existing research as a “priority” is a catalogue, not a roadmap. The discipline of identifying what the unit will not do usually separates a genuine roadmap from a political document.

National

A government or national research system mapping priorities for a field or sector: where to invest relative to other nations, what capabilities to build, which bottlenecks to address through policy or public funding. Common uses include national science and technology strategies, sector-specific R&D roadmaps, and national academies reviews.

National roadmaps are produced at longer cycles (typically 5–10 years), require multi-year stakeholder consultation, and must navigate political as well as scientific constraints. Germany’s High-Tech Agenda, the UK’s R&D Roadmap, and Denmark’s Innomissions are recent examples at this scale.

Key tension: Political cycles (4–5 years) are shorter than scientific ones (10–20 years). Building independent institutional structures - national academies, dedicated mission organisations - is the main mechanism for insulating long-horizon commitments from short-horizon politics.

Transnational / Global

Roadmapping that spans multiple nations, where the problem exceeds any single country’s capacity and the benefits of coordination are large: shared scientific infrastructure, global challenges, standards-dependent technology fields. Common uses include international research coordination, global technology standards, and shared infrastructure for science.

The IRDS is the archetype: a rolling consensus document produced by the global semiconductor industry coordinating R&D investment and standards across dozens of companies and research groups worldwide. Horizon Europe’s missions represent a transnational approach to societal challenges. At this scale, a roadmap becomes as much a political and diplomatic instrument as a scientific one - its authority derives from the legitimacy of the process that produced it, not only the accuracy of its content.

Key challenge: International roadmaps tend toward the lowest common denominator that all participants can endorse and update slowly because each revision requires re-running a complex consultation. A well-scoped organisational roadmap can iterate in months; the IRDS takes multiple years.

Vision papers & speculative formats ↑

A family of formats operates upstream of roadmaps: they describe what a field could become, what a problem is worth solving, or what a solution would unlock, before the technical path is clear enough to chart. Where a roadmap is primarily operational, these formats are primarily persuasive and imaginative. They establish the why, attract people to a problem, and create the shared language that makes roadmapping possible later.

Michael Nielsen’s working notes on vision papers in basic science¹¹ offer the clearest account of what distinguishes the genre: a vision paper proposes a research programme that does not yet exist. It is closer to an architectural sketch - a drawing of a building that has not yet been built, legible enough to attract collaborators and funders, but not a construction plan.

Vision papers

Describe a future state of a scientific field or capability that does not yet exist. Function as invitations: they recruit talent, orient funders, and establish the conceptual vocabulary a new field will organise around. The best vision papers are specific enough to be falsifiable but open enough to leave the technical path underdetermined.

Key resource: Michael Nielsen, Working notes on the role of vision papers in basic science. May 2022.

Grand challenge papers & problem statements

Identify the most important unsolved problems in a field and argue for why they matter and that they are now tractable. The archetype is Hilbert’s 23 problems (1900).¹² The distinction from vision papers: a grand challenge paper identifies the problem; a vision paper proposes a future state. “What should we solve?” vs. “What could exist?”

Key feature: Problem-centred. Often produced by a convened group. Functions as agenda-setting for a field or subfield.

Moonshot proposals

Argue that a specific ambitious goal is achievable within a defined time horizon. The key rhetorical move is the tractability argument: most readers accept that the goal would be valuable; the proposal’s work is to argue it is achievable. Requires engaging with the current state of the art and making the case that specific barriers are now surmountable.

Examples: Rodrigues & Marblestone, Focused Research Organizations (2020); Stebbins & Ling, Creating HARPA (2020).

Manifestos & position papers

Argue for a reorientation of a field’s priorities, methods, or values. Less focused on a specific technical goal than on how the field should operate. Metascience is particularly rich in this format - many founding texts of open science and replication reform are position papers.

Note: Position papers can be mistaken for roadmaps when they include a list of recommendations, but the orientation is different: they argue about values and priorities, not technical paths.

“What would it take” papers

A hybrid format sitting between vision paper and roadmap. Takes a specific desirable outcome and enumerates the preconditions - scientific, technical, institutional, political - without the full dependency map or timeline of a roadmap. Useful for scoping before a full roadmapping exercise.

Example: Willy Chertman’s fertility whitepaper (2023).

The progression

These formats are typically precursors to roadmaps, not competitors. The progression runs roughly: vision paper or grand challenge statement → moonshot proposal → “what would it take” scoping → roadmap → Tiling Tree and gap map. Each stage adds operational specificity. The failure mode of stopping too early is producing documents that inspire but don’t guide; the failure mode of skipping the early stages is producing roadmaps that are technically rigorous but that nobody acts on because the goal was never compellingly articulated.

Canonical vision papers

A. M. Turing — On Computable Numbers. 1936. Described a universal computing machine before any such thing existed.
Alexander Rich — On the Problems of Evolution and Biochemical Information Transfer. 1962. Anticipated the RNA World hypothesis — that RNA preceded both DNA and proteins as the original carrier of genetic information.
Alan Kay — A Personal Computer for Children of All Ages. 1972. Described the personal computer more than a decade before it existed commercially.
John Archibald Wheeler — Information, Physics, Quantum: The Search for Links. 1989. Articulated the “it from bit” programme in quantum information.
A. Yu. Kitaev — Fault-tolerant quantum computation by anyons. 2003. Proposed topological quantum computing; eventually the basis for major industry investment.

Mission maps ↑

Mission maps are a distinct format developed primarily through Mariana Mazzucato’s work on mission-oriented innovation.⁹ Where a technology roadmap charts the technical path to a specific capability, and an actor map charts who is doing what in a field, a mission map shows how multiple actors across different sectors - research, industry, government, civil society - must coordinate their activities to achieve a defined societal goal.

The mission concept starts from a diagnosis of why important societal challenges fail to be addressed: not because knowledge is absent, but because it is fragmented across actors who have no mechanism to coordinate. A mission creates that mechanism by defining a goal that is simultaneously ambitious enough to drive cross-sector mobilisation, concrete enough to be achievable within a defined horizon, cross-disciplinary enough that no single actor could achieve it alone, and measurable enough that you can tell when it is done. Horizon Europe’s five missions - adaptation to climate change, cancer, smart cities, A Soil Deal for Europe, ocean and water - are the current institutional instantiation.¹⁰

What a mission map shows

The mission: A precise, bounded societal goal at the top. Not “fight climate change” but “make 100 climate-neutral and smart cities by 2030.” Specificity is load-bearing: it determines what counts as progress and what kinds of actors need to be involved.

Contributing workstreams: The mission is decomposed not into a technical dependency tree (as in a roadmap) but into the distinct domains that must contribute - fundamental research, applied R&D, regulatory change, infrastructure investment, behavioural change, market creation. Each domain has its own logic and its own actors.

Actor landscape: Who in each domain is currently working toward the mission, and who needs to be recruited. The map makes visible which sectors are engaged and which are absent, and where the cross-sector connections are weak or missing.

Coordination points: Where contributions from different workstreams need to come together. These are the points where governance and coordination mechanisms are most needed.

Gaps: Domains or actor types that are missing or underrepresented. A mission map makes it visible when, for example, a scientific research programme is advancing but the regulatory pathway or the industrial manufacturing capacity to deploy it does not exist.

Key distinction from roadmaps: Roadmaps are primarily epistemological (what needs to be known or built). Mission maps are primarily sociological and political (who needs to act, how, and in what coordination with others).

The five criteria for a well-defined mission

Mazzucato’s framework¹³ specifies five criteria that distinguish a genuine mission from a vague aspiration:

Bold and inspirational: Captures the public imagination and motivates actors across sectors to engage.
Clear direction, with measurable targets: Specific enough that progress can be tracked and failure identified. “Cure cancer” is not a mission; “save more than 3 million lives from cancer by 2030” is.
Ambitious but realistic: Requires a step change, not just incremental improvement; but within reach given serious coordinated effort at a defined scale.
Cross-disciplinary and cross-sectoral: Requires contributions from multiple fields and types of actor. If a single organisation or discipline could solve it alone, it is not a mission - it is a project.
Enabling multiple, bottom-up solutions: Rather than prescribing a single approach, a mission should open the solution space to diverse actors and pathways. Top-down prescription of one route is a recurring failure mode of mission design.

Source: Mazzucato (2018), Mission-Oriented Research and Innovation in the European Union. Report for the European Commission.

Mission maps vs. roadmaps vs. theories of change

Roadmaps chart the technical dependencies toward a capability: what must be built, in what order, by whom. They are strongest on the R&D pathway.

Mission maps chart the multi-sector coordination required to translate capability into societal impact: who needs to act, across which domains, with what governance. They are strongest on the actor and coordination layer.

Theories of change specify the causal logic from activities through to outcomes: what must happen for each step to follow from the last. They are strongest on explicit assumptions.

A complete strategy for a complex societal goal typically needs all three: a roadmap for the R&D pathway, a mission map for the cross-sector coordination, and a theory of change to make the causal assumptions explicit.

Examples

Horizon Europe Missions - The most developed institutional implementation of mission-oriented innovation at scale. Five missions, each with its own mission board, implementation plan, and cross-sector coordination structure.
Danish Innomissions - Denmark’s mission-driven green research and innovation partnerships. A national-scale application of the mission model to climate and energy transitions.
The Apollo programme - The historical archetype: a precisely specified mission (land a man on the Moon and return him safely by the end of the decade) that drove coordinated R&D, procurement, and capability-building across the federal government and industrial base.
ARPA-H - Structured around health missions rather than individual grants; each programme manager is responsible for a mission-shaped problem rather than a portfolio of incremental projects.

The Tiling Tree Method ↑

The Tiling Tree Method, developed by Ed Boyden, Nina Khera, Adam Marblestone, and Claire Wang, is the most rigorous formal procedure for scientific roadmapping currently available. It was developed at MIT and refined through Convergent Research’s FRO scoping process. The core idea: to solve a problem, you must tile it - cover its entire solution space with approaches, and identify which tiles are missing or weak.

The procedure

Define the goal precisely. Ambiguous goals produce ambiguous trees. The goal should be specific enough that you could recognise it if it were achieved.
Identify the space of possible approaches. For each sub-problem, enumerate every conceptually distinct way it could in principle be solved - not just the approaches currently being pursued.
Build the tree. Decompose the goal into sub-problems, each sub-problem into further sub-problems, recursively, until you reach the level of specific experiments or engineering tasks.
Assess each branch. For each approach at each level: is it being pursued? By whom? What is its current state? What would it take to make it work? What is blocking it?
Identify the bottlenecks. A bottleneck is a node where all available approaches are either underpursued, blocked, or missing. These are the highest-leverage intervention points.

Key texts: Boyden, Khera, Marblestone & Wang, The Tiling Tree Method (2025); Part 2: Common Pitfalls.

The most important discipline the method enforces is the exhaustive enumeration of approaches. Most roadmapping stops at the approaches already being tried; the Tiling Tree explicitly asks what approaches are not being tried and why. This surfaces the “absent branches” - solution paths that nobody is pursuing, which may include the most important ones.

Common pitfalls: conflating the goal with one approach to achieving it; stopping the decomposition too early; failing to enumerate approaches outside the current paradigm; confusing “we don’t know how to do this” with “this is impossible.”

Target Product Profiles ↑

A Target Product Profile (TPP) is a specification document that defines the characteristics a technology or intervention must meet to be useful in a specific context. Developed originally in pharmaceutical R&D, TPPs have been extended to vaccines, diagnostics, agricultural inputs, and climate technologies.

The TPP is essentially a rigorous statement of the end state for a mission roadmap: it makes explicit what “success” means in terms specific enough to drive technical decisions. Without one, roadmaps tend to drift toward vague aspirational language that cannot adjudicate between competing approaches.

Key components of a TPP

Indication / use case: What problem does this solve, for whom, in what context?
Minimum acceptable profile: The floor: what performance would make the product worth deploying at all?
Preferred profile: The target: what would ideal look like?
Critical attributes: Which requirements are non-negotiable vs. negotiable?
Key uncertainties: Which attributes are currently unknown and most important to resolve?

Key resource: Renaissance Philanthropy — Target Product Profiles playbook. WHO TPP templates for vaccines and diagnostics.

Examples ↑

International Roadmap for Devices and Systems (IRDS)

The canonical technology roadmap. Coordinates the semiconductor industry around shared performance targets. Ongoing.

EBRC Engineering Biology Roadmaps

Field-wide roadmaps for engineering biology; produced collaboratively. Ongoing.

Fundamental Development Gap Map

Convergent Research’s living map of R&D gaps in biology - capabilities that would unblock multiple downstream directions.

Nanomodular Electronics Roadmap

Speculative Technologies roadmap by Michael Filler & Ben Reinhardt. February 2023.

Fusion Science & Technology Roadmap

DOE mission roadmap targeting grid-scale fusion by the mid-2030s. October 2025.

Horizon Europe Missions

Five cross-sector societal missions (climate, cancer, cities, soil, ocean). The most developed institutional application of mission-oriented innovation.

Focused Research Organizations (vision paper)

Rodrigues & Marblestone, 2020. Moonshot proposal that launched the FRO model. Example of a vision paper that generated a field.

Solid-state Far-UVC Roadmapping Workshop Report

Output from a structured expert workshop; illustrates the Delphi/workshop process. 2024.

For a broader curated list of roadmaps and related documents, see the Technology Roadmaps, Tech Trees & Agenda Setting section on the main page.

Notes

International Roadmap for Devices and Systems (IRDS). IEEE, ongoing. Successor to the International Technology Roadmap for Semiconductors (ITRS, 1992–2016), which coordinated the global semiconductor industry around shared performance targets for over two decades.
Convergent Research — Fundamental Development Gap Map. A living, crowd-sourced map of foundational R&D gaps across biology and adjacent fields: capabilities whose development would unblock multiple downstream directions.
Renaissance Philanthropy — Agenda Setting is Underrated. December 2024. Renaissance Philanthropy’s broader agenda-setting playbooks treat roadmaps and gap maps as core instruments of field creation.
Speculative Technologies (Spec Tech) — Roadmap library. Deep-tech roadmaps applying technology roadmap methodology to frontier domains, including nanomodular electronics and macromolecular additive manufacturing.
U.S. Department of Energy — Fusion Science & Technology Roadmap. October 2025. Charts the technical path to grid-scale fusion power.
Advanced Research Projects Agency for Health (ARPA-H). Established 2022. Each programme manager is responsible for a mission-shaped health challenge rather than a portfolio of incremental grants.
Engineering Biology Research Consortium (EBRC) — Engineering Biology Roadmaps. Field-wide roadmaps produced collaboratively by the EBRC community; updated periodically to reflect the current state and open challenges of the field.
Ed Boyden, Nina Khera, Adam Marblestone & Claire Wang — The Tiling Tree Method. Engineering X, 21 September 2025. See also Part 2: Common Pitfalls.
Mariana Mazzucato — Mission-Oriented Research and Innovation in the European Union. Report for the European Commission. UCL Institute for Innovation and Public Purpose, 2018. See also Mission Economy: A Moonshot Guide to Changing Capitalism. Allen Lane, 2021.
EU Missions in Horizon Europe. European Commission. The five missions: adaptation to climate change, cancer, smart and climate-neutral cities, a soil deal for Europe, and ocean and water.
Michael Nielsen — Working notes on the role of vision papers in basic science. May 2022. Hosted at scienceplusplus.org.
David Hilbert — “Mathematische Probleme.” Address to the International Congress of Mathematicians, Paris, 1900. The 23 problems Hilbert formulated shaped the course of twentieth-century mathematics; the format has since been used extensively to set research agendas in other fields.
The five criteria are drawn from Mazzucato (2018); see note 9.
The canonical academic reference for technology roadmapping methodology is Robert Phaal, Clare Farrukh & David Probert — “Technology roadmapping — A planning framework for evolution and revolution.” Technological Forecasting and Social Change, 71 (2004). Phaal’s group at the Cambridge Institute for Manufacturing developed the T-Plan rapid roadmapping process and a taxonomy of roadmap formats. See also the IFM Roadmapping Hub.