Science and Research

The state should be a major driver of scientific research.

Institutions such as the CNRS, universities, national laboratories, and public research organisations exist precisely because many important areas of research require:

• stable funding;
• long time horizons;
• scientific freedom;
• permanent expertise;
• large infrastructure;
• continuity between generations of researchers.

Research policy should therefore provide public institutions with sufficient permanent funding, staff, equipment, and infrastructure.

Science should not depend excessively on short-term project funding.

Public laboratories should receive substantially more recurrent funding.

Researchers should not be forced to spend excessive amounts of time preparing, submitting, reviewing, and administering competitive grant proposals simply to maintain ordinary scientific activity.

Competitive funding has a role for specific ambitious projects, but it should complement stable institutional funding rather than replace it.

A functioning laboratory should be able to:

• maintain equipment;
• retain technical staff;
• support doctoral students;
• purchase routine materials;
• conduct exploratory work;
• maintain software;
• preserve data;
• develop long-term research programmes;

without having to rebuild its financial existence every few years.

Public research requires continuity.

I therefore support more permanent scientific, engineering, technical, and support positions.

Excessive dependence on short-term contracts creates instability and encourages researchers to organise their work around the duration of grants rather than the scientific problems themselves.

A research system that continuously trains highly qualified people and then forces them into repeated temporary contracts wastes both human and public investment.

Scientific careers should remain selective, but precarity should not become the default organisational model of public research.

Fundamental research should be financed even when no immediate industrial or commercial application is visible.

The value of scientific knowledge cannot always be predicted in advance.

Many major technological developments originate from work that initially had no obvious economic application.

Public institutions are particularly important for this reason: they can fund research according to scientific importance and long-term potential rather than short-term market demand.

Researchers should have sufficient freedom to investigate questions that are scientifically important even when those questions are not currently fashionable or commercially attractive.

Scientific institutions should protect researchers from inappropriate political, commercial, or managerial pressure.

Scientific freedom does not eliminate accountability.

Public research should remain transparent about funding, methods, conflicts of interest, and the use of public resources.

Scientific knowledge produced using public funding should be openly accessible.

Access should be free for readers.

Publication should also be free, or close to free, for authors and their institutions.

Public institutions should not finance research, provide the researchers, perform much of the peer review, and then pay private publishers large sums to read or publish the resulting work.

Scientific communication should increasingly become public infrastructure.

Europe should develop a major public scientific publication infrastructure.

This platform could be supported by European institutions, national research organisations, universities, and agencies such as the CNRS.

It should provide:

• peer review;
• persistent identifiers;
• long-term archiving;
• versioning;
• citation infrastructure;
• indexing;
• metadata;
• data repositories;
• software repositories;
• links between articles, data, and code.

Publication and access should be free at the point of use.

The objective should be to make publicly funded scientific communication itself a public good.

Scientific institutions should progressively reduce their dependence on large commercial publishers.

Private publishers can continue to exist, but they should not control access to publicly funded scientific knowledge.

Where a credible public infrastructure exists, universities and research organisations should be willing to move publications, peer review, archives, and scientific communication toward public or non-profit platforms.

Scientific prestige should not depend on a small number of private journals.

Research data produced using public funding should, by default, be made openly available when this is legally, ethically, and technically possible.

Exceptions are legitimate for:

• personal data;
• medical confidentiality;
• national security;
• industrial confidentiality;
• environmental or ecological sensitivity;
• other clearly justified constraints.

The default, however, should be openness rather than secrecy.

Data should be accompanied by sufficient metadata and documentation to allow others to understand and reuse it.

Software developed using public research funding should normally be released as free software.

Scientific software is part of the scientific method.

If computational results depend on software that cannot be inspected, modified, or executed independently, reproducibility is weakened.

Research code should therefore be:

• published;
• documented;
• versioned;
• licensed clearly;
• archived;
• linked to the corresponding scientific work.

Where possible, strong free software licences should be preferred.

Reproducibility should be a central principle of scientific research.

A scientific result should provide enough information for an independent team to understand how it was obtained and, where technically possible, reproduce the analysis.

This means publishing or documenting:

• raw data where possible;
• processed data;
• source code;
• algorithms;
• experimental protocols;
• software versions;
• computational environments;
• parameter values;
• calibration procedures;
• uncertainty analysis.

Scientific publications should explain not only the final result but the path that produced it.

Computational research should increasingly use reproducible environments.

Relevant dependencies, versions, compiler settings, data transformations, and execution procedures should be documented.

Tools such as containers, package lock files, environment descriptions, reproducible builds, and automated workflows can help ensure that results can still be reproduced years later.

The specific technology may change over time, but the principle should remain: the computational environment is part of the scientific experiment.

Scientific publishing should make more room for negative results, non-replications, and failed hypotheses.

A research system that mainly rewards positive and novel results creates publication bias and encourages unnecessary duplication.

Knowing that an approach does not work can itself be valuable scientific knowledge.

Public publication infrastructure should make it easier to preserve and publish such results.

Scientific evaluation should not be reduced to publication counts, journal prestige, impact factors, citation counts, or h-index values.

These indicators can provide information, but they are poor substitutes for scientific judgement.

Research should also be evaluated according to:

• originality;
• methodological quality;
• reproducibility;
• scientific importance;
• software contributions;
• datasets;
• instrumentation;
• teaching;
• supervision;
• long-term technical work;
• community contributions;
• replication studies.

The objective should be to evaluate actual scientific contribution rather than optimisation of bibliometric indicators.

Scientists should have a stronger role in public decision-making when policies depend on scientific or technical knowledge.

This does not mean replacing democracy with technocracy.

Scientific expertise can establish constraints, quantify risks, analyse evidence, and compare technical alternatives.

Political institutions must still decide between competing values and social priorities.

However, political decision-makers should be expected to explain clearly when they choose policies that contradict the best available scientific evidence.

Parliament and government should have access to permanent, independent scientific advisory structures.

These institutions should bring together researchers and technical experts from relevant disciplines.

Their analyses should be:

• public;
• documented;
• traceable;
• explicit about uncertainty;
• explicit about disagreement;
• transparent about conflicts of interest.

Scientific advice should support political decision-making without becoming an opaque source of authority.

Parliament should have strong internal scientific and technical expertise.

Members of parliament should not be forced to depend primarily on ministries, lobbyists, consultancy firms, or private companies to understand complex technical legislation.

Independent parliamentary scientific offices should be capable of evaluating issues such as:

• energy;
• cybersecurity;
• healthcare;
• artificial intelligence;
• industrial policy;
• climate;
• digital infrastructure;
• biotechnology;
• defence technology.

A legislature cannot exercise meaningful democratic control over technical systems it does not understand.

Scientific expertise used in public decision-making must be transparent about conflicts of interest.

Relevant financial, industrial, institutional, and personal relationships should be disclosed.

A conflict of interest does not automatically invalidate scientific expertise, but hidden conflicts damage trust and make independent evaluation more difficult.

Public institutions should publish the sources of funding and relevant interests of experts contributing to major policy decisions.

Uncertainty is a normal part of science.

Public debate should distinguish between genuine scientific uncertainty and the deliberate manufacture of doubt.

Scientists should communicate uncertainty honestly.

Confidence intervals, limitations, competing hypotheses, model assumptions, and unknowns should be made explicit when relevant.

Political institutions should not demand artificial certainty from science.

At the same time, uncertainty should not be used as an excuse to ignore strong evidence.

Europe should coordinate more scientific research at continental scale.

Many scientific programmes require infrastructure, funding, and expertise that are difficult for individual countries to sustain alone.

European cooperation should be strengthened for:

• supercomputing;
• synchrotrons;
• particle accelerators;
• telescopes;
• satellite programmes;
• large laboratories;
• scientific databases;
• genomic infrastructure;
• metrology;
• scientific software;
• research networks.

Shared European infrastructure can reduce unnecessary duplication while increasing scientific capability.

Scientific sovereignty is an essential component of technological sovereignty.

Europe should retain the capacity to design, build, operate, and maintain the tools required for scientific work.

This includes capabilities in:

• scientific instrumentation;
• high-performance computing;
• semiconductors;
• sensors;
• laboratory equipment;
• scientific software;
• measurement systems;
• data infrastructure;
• precision manufacturing;
• metrology.

Dependence on external suppliers for critical scientific infrastructure can eventually become dependence on external knowledge and technical capability.

Some scientific infrastructure should be treated as a public utility.

This can include:

• repositories;
• publication platforms;
• computing infrastructure;
• data storage;
• research networks;
• software registries;
• metadata systems;
• identity systems;
• long-term archives.

Such infrastructure should use open standards and, where possible, free software.

Scientific knowledge should not depend on opaque proprietary infrastructure that public institutions cannot independently maintain.

Research financed primarily with public money should not automatically become exclusive private intellectual property.

Where patents are genuinely useful for transferring technology or supporting industrial development, they can remain legitimate.

However, public institutions should prefer licensing strategies that preserve broad access.

This can include:

• non-exclusive licences;
• open licences;
• public-interest licensing conditions;
• royalty-free licences for research;
• obligations concerning accessibility and affordability.

Exclusive monopolies should not be granted automatically when the public has already financed the underlying research.

Public research and industry should cooperate.

Industrial collaboration can help transform scientific knowledge into useful technology.

However, cooperation should not make public research dependent on commercial objectives.

Contracts with private companies should protect:

• scientific independence;
• publication rights;
• research continuity;
• public-interest licensing;
• transparency where appropriate.

Public-private cooperation is valuable when it extends scientific capability rather than privatising publicly produced knowledge.

Scientific education should not be limited to training future researchers.

Citizens should understand enough science to participate meaningfully in a technological democracy.

Education should develop:

• quantitative reasoning;
• statistics;
• scientific methodology;
• uncertainty;
• experimental reasoning;
• source evaluation;
• basic computing;
• understanding of models;
• critical thinking.

A population capable of understanding scientific reasoning is harder to mislead through pseudoscience or manipulated technical claims.

Publicly funded researchers and institutions should have a stronger mission of scientific communication.

Scientific knowledge should be explained in clear language without sacrificing accuracy.

Universities, research organisations, museums, media, and public institutions should invest more heavily in high-quality scientific communication.

The objective should not be to demand blind trust in experts.

It should be to give citizens enough information to understand why a scientific claim is considered reliable and where its limitations lie.

Scientific infrastructure is increasingly digital and must therefore be protected accordingly.

Research organisations should develop strong internal competence in:

• system administration;
• network security;
• cryptography;
• software maintenance;
• backups;
• incident response;
• identity management;
• secure software supply chains.

Scientific organisations should not depend entirely on outsourced black-box systems for critical technical functions.

Technical competence is itself part of scientific sovereignty.

I do not currently adopt a broad political position on artificial intelligence as a whole.

However, its use in scientific research should remain compatible with the basic requirements of scientific methodology.

When AI systems contribute materially to a scientific result, researchers should document their use sufficiently to allow the result to be evaluated.

Scientific workflows should remain:

• auditable;
• reproducible;
• documented;
• traceable.

The use of AI should not turn scientific reasoning into an opaque process that cannot be independently examined.

The long-term objective should be a scientific system in which publicly funded knowledge accumulates as a common resource.

Research should produce not only papers, but also:

• knowledge;
• data;
• software;
• methods;
• instruments;
• standards;
• trained scientists;
• technical competence;
• public understanding.

These assets should reinforce one another over decades.

A strong scientific system is therefore not simply a mechanism for producing publications.

It is part of the intellectual, industrial, democratic, and technological infrastructure of society.