The Last Author Question: How Authorship Order Quietly Shapes Careers Across Fields

Two doctoral students publish a paper together in their third year. They contributed roughly equally, both wrote sections, both ran analyses, and their names appear side by side on the title page. Five years later, one is on a tenure track and the other has slid through two unsuccessful postdoc cycles into an industry role they did not initially want. The papers themselves were never the problem. What separated them, more often than either is comfortable admitting, was a question almost no one explains to early-career researchers in plain terms: in their fields, which position on the author list signals the work that hiring committees actually count.

Authorship order looks, from the outside, like a simple matter of credit. In reality, it is a scoring convention — and the conventions vary so wildly across fields that the same five-author paper can read as a career-making contribution in one discipline and as a forgettable supporting role in another. Researchers who understand the convention of the field they intend to enter make decisions, sometimes years in advance, that researchers who do not understand it cannot.

Same paper, different score

In experimental life sciences and clinical research, authorship order encodes a recognisable hierarchy. The first author is the person who did the bulk of the work — typically the lead PhD student or postdoc on the project. The last author is the principal investigator: the senior figure who funded the work, supervised it, and is held intellectually responsible for it. Everyone in the middle is in a less legible position; “middle author” is one of the quieter career problems in modern biomedical science, because the contribution is real but the signal is faint.

In high-energy physics and other big-collaboration sciences, the entire convention collapses. Author lists of thousands of names are now routine, and order within them is alphabetical or formulaic. No single author can be said to have “led” the paper in the way a life sciences first author leads, and CVs in those fields rely on auxiliary documents — analysis notes, internal review records — to establish what a candidate actually did.

In economics, finance, and most of mathematics, alphabetical order is the default, regardless of contribution. A footnote or acknowledgments paragraph may indicate who did what, but the author list itself carries no information about effort. A researcher whose surname begins with a letter near the end of the alphabet learns early that they will often be the last name on the line for reasons that have nothing to do with what they wrote, and the convention is well enough understood within those fields that this does not penalise them — within those fields.

In computer science the picture is mixed, drifting field by field toward a first-author-led model in machine learning and systems, while older subfields retain alphabetical or seniority-based conventions. In most of the humanities, single authorship still dominates, and a co-authored paper raises its own faintly suspicious set of questions about who really wrote it.

Author order by field

The same author list, read five different ways

Each field reads authorship through its own convention. Cross a boundary the convention does not, and the same paper changes meaning.

• Life sciences & clinical research

  First = lead contributor. Last = senior PI.

  A recognisable hierarchy. Middle positions are real contributions but read as faint signals on a CV.

• Economics, mathematics, finance

  Alphabetical, by default.

  Position carries no information about effort. Contribution, where noted at all, sits in a footnote or acknowledgments line.

• High-energy physics & large collaborations

  Alphabetical, on lists running to thousands.

  No single author “led” the paper in the conventional sense; auxiliary documents establish what a candidate actually did.

• Computer science

  Mixed, drifting toward first-author-led in ML and systems.

  Older subfields retain alphabetical or seniority-based ordering; newer ones increasingly reward visible first-authorship.

• Humanities

  Single authorship still dominates.

  A co-authored paper raises its own questions about who really wrote it; the convention is, by default, suspicion.

The signal that does not travel

The problem these conventions create is not internal to a field; each works well enough on its own terms. The problem is what happens when a researcher’s career crosses a boundary the convention does not. An economics PhD applying to a quantitative finance group is moving from a world that reads authorship alphabetically into one that, often, does not. A computer scientist with first-author NeurIPS papers applying to a computational biology lab is moving the other way. An interdisciplinary candidate working between fields with different conventions is read incorrectly by both, almost by default.

The mistake hiring committees make is rarely malicious. They read the only convention they know. A life sciences reviewer looking at a CV of alphabetically ordered economics papers sees a candidate who is “always in the middle” and quietly downgrades them. An economist looking at a life sciences CV with five middle-author papers and no first-author publications sees a candidate who has not yet led anything, and treats them accordingly. Neither reading is fair; both are common.

What you can learn before you write

The practical implication is not difficult, but it is rarely stated, and it has to be applied early — before a manuscript is drafted, not after it is submitted. Three habits separate the researchers who understand authorship order from the ones who learn it the hard way.

The first is to know the convention of the field whose hiring decisions will most affect your future, not the field your current group sits inside. If you intend to apply to medical schools, you need first-authored work, even if your group habitually slots PhD students in the middle of long papers. If you intend to enter a discipline that grades alphabetically, you can afford to be relaxed about a position that would have been a problem in a life-sciences trajectory. The convention that matters is the one your future readers will use, not the one your current corridor uses.

The second is to discuss authorship order before the writing begins, and to do it in writing. Verbal understandings about who will be first author dissolve under the pressure of a deadline, particularly when an additional contributor joins late and tips the balance. A short email exchange, agreed by all parties before the work starts in earnest, prevents the most painful and quietly common form of authorship dispute: the one that happens with three days left before submission and no one willing to be the person who lost.

The third is to make the contribution legible regardless of position. The CRediT taxonomy now used by most major journals — which lists, for each author, exactly which roles they played — is one of the better things to have happened to academic credit in a decade. A researcher stranded in the middle of an author list can use a clear contribution statement to claim methodological leadership, data ownership, or statistical work in a way the order alone would never have communicated. Hiring committees that read these statements (and more of them do, every year) will see what the author list hides.

Order is a contract, not a reward

The most useful reframing of authorship order, for early-career researchers, is to stop thinking of it as a reward distributed at the end of a project — a prize for the person who worked hardest — and to start thinking of it as a contract negotiated at the beginning of one. Rewards can feel deserved or undeserved, and the conversation around them becomes emotional quickly. Contracts can be discussed in advance, written down, and renegotiated openly if the work changes. The same conversation that feels awkward and presumptuous in week ten of a project (“can I be first author?”) is straightforward in week one (“how are we thinking about authorship if this becomes a paper?”), because in week one no one has done enough work yet to feel territorial about it.

A paper has many readers, and only the first few of them care about science alone. The rest — the committees, the panels, the future employers — care about what the paper tells them about the people who wrote it. What the author list tells those readers is not a universal truth; it is a translation, from a convention they know into a verdict on a candidate they do not. Knowing the convention they will read by, and quietly engineering your name into the place that convention rewards, is not gaming the system. It is the system. Researchers who understand that early move through their careers with a small but durable advantage over the ones who learn, two cycles in, that they were playing a different game from the one their CV is being read by.