10 Ethics

“A mathematician does something on a piece of paper & then lo & behold a big explosion may occur.”

– Stanislaw Ulam, scribbled in the drawer of his desk at Los Alamos¹

science

The systematic study of the structure and behaviour of the physical and natural world through observation, experimentation, and the testing of theories against the evidence obtained (Oxford Dictionary of English 2010 s.v. “science”).

ethics

Moral principles that govern a person’s behaviour or the conducting of an activity (Oxford Dictionary of English 2010 s.v. “ethics”).

big science

Scientific research that is expensive and involves large teams of scientists (Oxford Dictionary of English 2010 s.v. “big science”).

blue-sky

Creative or visionary and unconstrained by practicalities (Oxford Dictionary of English 2010 s.v. “blue-sky”).

10.1 Two guiding questions

Ethical issues in scientific work tend to cluster around two questions:

what science should be conducted and
how science should be conducted.

The first question asks about purpose and proportionality. We weigh the value of the knowledge against the costs and risks, including the possibility of misuse. In mathematics and physics this often means checking whether a model, algorithm, or optimisation method could be applied in ways we did not intend, thereby steering policy, finance, surveillance, or defence. If unintended consequences are identified, then one must decide what safeguards or limits on release are appropriate. It also asks who benefits, who bears burdens, and whether scarce time and resources are being directed fairly.

The second question focuses on conduct: about integrity and care in everyday practice. As professional mathematicians and physicists, we commit to honesty in analysis and reporting; clear attribution of ideas, data, and tools; and enough transparency that a peer can retrace our steps. In quantitative work, this includes stating assumptions, limits of validity, and uncertainty; keeping a reproducible trail for computations; and avoiding exaggerated claims. Good conduct extends to collaboration, such as fair authorship, respectful review, prompt correction of errors, and to communication, where we present results responsibly without hype.

These two questions, what and how, form a habit: pause before starting, and reflect before sharing. They help keep powerful methods pointed toward good ends, and they keep trust in our work intact.

10.2 What work should be done

Big scientific programmes raise questions of priority and fairness. Big Science projects (costly, long-horizon efforts run by large teams) promise major advances, but they also centralise decisions about which fields get funded and who gets to participate, including whether researchers in the developing world can share equitably in the work and its benefits.

By contrast, Blue-Sky research pursues ideas with no immediate payoff. Such research can look speculative in the moment yet prove foundational later. For example, the early laser was famously dismissed as “a solution looking for a problem,” before becoming ubiquitous in science, medicine, and communications. The ethical question is whether we sustain such inquiry despite uncertain returns, and how we justify that choice to the public.

To weigh these options, one should use ethical lenses: a virtue perspective asks what kind of scientific character we cultivate; a Kantian view tests duties we owe regardless of outcomes; a utilitarian calculus weighs overall consequences. Importantly, for any ethical decision, one must guard against hindsight bias: as Herodotus reminds us, a decision should be judged by the evidence available when it was made, not by how it turned out.² This keeps the evaluation of risks anchored in reasoned judgment rather than outcome luck.

A quick ethics checklist

The following ethical considerations might be taken into account for most research projects.

Purpose & stakes: Who benefits? Who could be harmed?
Scope & assumptions: Where does your result not apply?
Uncertainty: What’s the error budget/sensitivity?
Transparency: Can a peer reproduce your work?
Attribution: Have you credited data/code/ideas?
Dual-use: Could this be misapplied, and how will you mitigate?

10.3 How we conduct ourselves

Not all errors are equal (Hamner 1992). Some errors are reputable errors; these are mistakes, made in good-faith, that are documented, reproducible, and fixable. These represent a deviation from methodological norms and include negligence, such as sloppy methods or record-keeping that breach methodological standards even without intent. On the other hand, some errors are disreputable, such as fabrication, falsification, and plagiarism. These represent a deviation from moral norms. The practical ethic is to design your workflow so that reputable error is possible and correctable (clear notes, version control, open materials) and disreputable error is impossible (independent checks, audit trails, explicit authorship and citation norms).

Moral norms

Robert K. Merton’s classic account of the scientific ethos names four ideals: universalism, communality, disinterestedness, and organized skepticism. His essay on these moral norms first appeared in 1942 and has been reprinted with revised titles reflect changing values in the scientific community (Merton 1973). The current resource frames openness and research transparency today. Think of these norms as routine habits that counter the very pressures that produce negligence and fraud.

Universalism: judge claims by impersonal criteria, not by who makes them.
Communality: treat knowledge as a shared good, circulated with credit.
Disinterestedness: organize work so private interests and biases do not distort it.
Organized skepticism: subject claims to systematic, critical scrutiny before acceptance.

Practical ethic

A practical ethic turns values into routine habits. A few examples of designing workflows to ensure that ethical values are maintained are given below. These approaches make reputable error easy to detect and fix and make misconduct hard to hide.

FAIR data

FAIR is a professional baseline for sharing data: Findable (rich metadata with persistent identifiers), Accessible (clear access terms with stable links), Interoperable (open formats with standard vocabularies), and Reusable (licensing with provenance and a data dictionary). If data cannot be shared, publish a metadata-only record and a safe exemplar. See GO FAIR’s concise guidance and checklists at https://www.gofair.foundation/.

Reproducible Quarto reports

Quarto supports ethical practice by bundling narrative, code, and outputs in a single code that can be used to generate many types of documents (HTML, PDF, EPUB, etc.). This code captures parameters, seeds, and environments thereby making analyses auditable. Rendering the document from a clean clone³ verifies that results do not depend on a private setup and sharing the source code provides an explicit trail for peer scrutiny. Learn more at https://quarto.org/.⁴

10.4 For discussion

Draw a line

Where’s the line between negligence and misconduct in simulation-heavy mathematics/physics?

Together or alone?

How do Merton’s norms play out in a large collaboration vs a solo proof?

Prioritising resrouces

Should limited resources prioritise Big Science or Blue-Sky Research, and who decides?

See Clark (2017).↩︎
Cf. Solon’s advice to Croesus: “[c]all him, however, until he die, not happy but fortunate” (Herodotus (transl. George Rawlinson) 2009, bk. 1);↩︎
A clean copy of a git repository; git is a type of distributed version control system that tracks file changes.↩︎
These notes are authored with quarto. You can view the code at https://github.com/dundeemath/PH11002/.↩︎

{{< include preamble.qmd >}} # Ethics {#sec-ethics} > "A mathematician does something on a piece of paper & then lo & behold a big explosion may occur." > > -- Stanislaw Ulam, scribbled in the drawer of his desk at Los Alamos[^los-alamos] [^los-alamos]: See @Clark:2017la. ::: {.callout-note icon=false collapse="true"} ## science The systematic study of the structure and behaviour of the physical and natural world through observation, experimentation, and the testing of theories against the evidence obtained [@ODE:2010 s.v. "science"]. ::: ::: {.callout-note icon=false collapse="true"} ## ethics Moral principles that govern a person's behaviour or the conducting of an activity [@ODE:2010 s.v. "ethics"]. ::: ::: {.callout-note icon=false collapse="true"} ## big science Scientific research that is expensive and involves large teams of scientists [@ODE:2010 s.v. "big science"]. ::: ::: {.callout-note icon=false collapse="true"} ## blue-sky Creative or visionary and unconstrained by practicalities [@ODE:2010 s.v. "blue-sky"]. ::: ## Two guiding questions Ethical issues in scientific work tend to cluster around two questions: 1. what science should be conducted and 2. how science should be conducted. The first question asks about purpose and proportionality. We weigh the value of the knowledge against the costs and risks, including the possibility of misuse. In mathematics and physics this often means checking whether a model, algorithm, or optimisation method could be applied in ways we did not intend, thereby steering policy, finance, surveillance, or defence. If unintended consequences are identified, then one must decide what safeguards or limits on release are appropriate. It also asks who benefits, who bears burdens, and whether scarce time and resources are being directed fairly. The second question focuses on conduct: about integrity and care in everyday practice. As professional mathematicians and physicists, we commit to honesty in analysis and reporting; clear attribution of ideas, data, and tools; and enough transparency that a peer can retrace our steps. In quantitative work, this includes stating assumptions, limits of validity, and uncertainty; keeping a reproducible trail for computations; and avoiding exaggerated claims. Good conduct extends to collaboration, such as fair authorship, respectful review, prompt correction of errors, and to communication, where we present results responsibly without hype. These two questions, what and how, form a habit: pause before starting, and reflect before sharing. They help keep powerful methods pointed toward good ends, and they keep trust in our work intact. ## What work should be done Big scientific programmes raise questions of priority and fairness. Big Science projects (costly, long-horizon efforts run by large teams) promise major advances, but they also centralise decisions about which fields get funded and who gets to participate, including whether researchers in the developing world can share equitably in the work and its benefits. By contrast, Blue-Sky research pursues ideas with no immediate payoff. Such research can look speculative in the moment yet prove foundational later. For example, the early laser was famously dismissed as "a solution looking for a problem," before becoming ubiquitous in science, medicine, and communications. The ethical question is whether we sustain such inquiry despite uncertain returns, and how we justify that choice to the public. To weigh these options, one should use ethical lenses: a virtue perspective asks what kind of scientific character we cultivate; a Kantian view tests duties we owe regardless of outcomes; a utilitarian calculus weighs overall consequences. Importantly, for any ethical decision, one must guard against hindsight bias: as Herodotus reminds us, a decision should be judged by the evidence available when it was made, not by how it turned out.[^herodotus] This keeps the evaluation of risks anchored in reasoned judgment rather than outcome luck. [^herodotus]: Cf. Solon’s advice to Croesus: _"[c]all him, however, until he die, not happy but fortunate"_ [@Herodotus:440BCE Book 1]; ::: {.callout-tip} ## A quick ethics checklist The following ethical considerations might be taken into account for most research projects. - Purpose & stakes: Who benefits? Who could be harmed? - Scope & assumptions: Where does your result **not** apply? - Uncertainty: What's the error budget/sensitivity? - Transparency: Can a peer reproduce your work? - Attribution: Have you credited data/code/ideas? - Dual-use: Could this be misapplied, and how will you mitigate? ::: ## How we conduct ourselves Not all errors are equal [@Hamner:1992et]. Some errors are reputable errors; these are mistakes, made in good-faith, that are documented, reproducible, and fixable. These represent a deviation from methodological norms and include _negligence_, such as sloppy methods or record-keeping that breach methodological standards even without intent. On the other hand, some errors are disreputable, such as fabrication, falsification, and plagiarism. These represent a deviation from moral norms. The practical ethic is to design your workflow so that reputable error is possible and correctable (clear notes, version control, open materials) and disreputable error is impossible (independent checks, audit trails, explicit authorship and citation norms). ### Moral norms Robert K. Merton's classic account of the scientific ethos names four ideals: universalism, communality, disinterestedness, and organized skepticism. His essay on these moral norms first appeared in 1942 and has been reprinted with revised titles reflect changing values in the scientific community [@Merton:1973et]. The current resource frames openness and research transparency today. Think of these norms as routine habits that counter the very pressures that produce negligence and fraud. * Universalism: judge claims by impersonal criteria, not by who makes them. * Communality: treat knowledge as a shared good, circulated with credit. * Disinterestedness: organize work so private interests and biases do not distort it. * Organized skepticism: subject claims to systematic, critical scrutiny before acceptance. ### Practical ethic A practical ethic turns values into routine habits. A few examples of designing workflows to ensure that ethical values are maintained are given below. These approaches make reputable error easy to detect and fix and make misconduct hard to hide. ::: {.callout-note} ## FAIR data FAIR is a professional baseline for sharing data: **F**indable (rich metadata with persistent identifiers), **A**ccessible (clear access terms with stable links), **I**nteroperable (open formats with standard vocabularies), and **R**eusable (licensing with provenance and a data dictionary). If data cannot be shared, publish a metadata-only record and a safe exemplar. See GO FAIR's concise guidance and checklists at [https://www.gofair.foundation/](https://www.gofair.foundation/). ::: ::: {.callout-note} ## Reproducible Quarto reports Quarto supports ethical practice by bundling narrative, code, and outputs in a single code that can be used to generate many types of documents (HTML, PDF, EPUB, etc.). This code captures parameters, seeds, and environments thereby making analyses auditable. Rendering the document from a clean clone[^clone] verifies that results do not depend on a private setup and sharing the source code provides an explicit trail for peer scrutiny. Learn more at [https://quarto.org/](https://quarto.org/).[^quarto] ::: [^clone]: A clean copy of a git repository; git is a type of distributed version control system that tracks file changes. [^quarto]: These notes are authored with quarto. You can view the code at [https://github.com/dundeemath/PH11002/](https://github.com/dundeemath/PH11002/). ## For discussion ::: {.callout-important icon=false collapse="true"} ## Draw a line Where's the line between negligence and misconduct in simulation-heavy mathematics/physics? ::: ::: {.callout-important icon=false collapse="true"} ## Together or alone? How do Merton's norms play out in a large collaboration vs a solo proof? ::: ::: {.callout-important icon=false collapse="true"} ## Prioritising resrouces Should limited resources prioritise Big Science or Blue-Sky Research, and who decides? :::