The gender gap in scienceHow long until women are equally represented?

Thirty-six million authors. That is the scale of what Luke Holman, Devi Stuart-Fox, and Cindy Hauser achieved. They drew from PubMed and arXiv — the two largest open bibliographic databases in science — covering more than six thousand journals, over one hundred countries, and roughly fifteen years of publishing history across nearly every science, technology, engineering, mathematics, and medicine discipline. The result is the most comprehensive picture we've ever had of who is actually writing the science that gets published. The technical move at the heart of this was probabilistic name-to-gender assignment. The team queried genderize dot io, a service built on a database of more than two hundred thousand given names with country-specific associations for around eighty countries. For each name, the service returns the proportion of people with that name who are men or women. For instance, fifty-nine percent of people named Robin in the database are women, while only seven percent of people named Chris are. To keep the analysis clean, Holman and colleagues applied a strict cutoff: they only retained names associated with one gender at least ninety-five percent of the time. That decision still preserved ninety-two point four percent of the dataset — thirty-five point five million PubMed author assignments and about one point one million arXiv authors. When they validated this against a manually curated set of three hundred seventy-two authors, there was exactly one misclassification. That's an error rate of zero point three percent. These numbers are solid. The results are available through an interactive web app the team built, where anyone can drill down by journal, discipline, country, or authorship position. That matters because what they found is not a single uniform gap — it's a landscape with steep terrain. Imagine walking through a house where each room hosts a different scientific discipline. Some rooms are approaching balance. Most are not. Across one hundred fifteen disciplines, eighty-seven had significantly fewer than forty-five percent women authors. Only five had significantly more than fifty-five percent women. Twenty-three were within striking distance of parity. The most welcoming rooms — nursing, midwifery, and palliative care — are in the health and life sciences. The most closed off are physics, computer science, mathematics, surgery, and chemistry. These aren't close calls. These are fields where women remain a small fraction of the authorship, and the doors have been slow to open. The pace of change matters too. Many disciplines are making steady progress. But for the strongly male-biased fields, the trajectory under current trends is bleak. Holman and colleagues don't just say it's getting better slowly; they model it. Their time-series model fits each discipline's gender ratio over time, estimates the current rate of change, and projects how many years it would take to reach a forty-five to fifty-five percent window — what they define as parity. The results are striking. In the arXiv physics category, women comprised around thirteen percent of last authors in two thousand sixteen, and that proportion was increasing by approximately zero point one percentage points per year. Feed those numbers into the model and you get a projected time to parity of two hundred fifty-eight years, with a ninety-five percent confidence interval running from one hundred ninety-four to three hundred eighty-three. Some fields don't even get a finite projection — the data don't show a clear upward trend at all, meaning parity may never arrive under current conditions. That number — two hundred fifty-eight years — deserves a moment. A researcher entering physics today would need to live until the twenty-third century to see senior authorship approach parity in her field. That's not a pipeline problem that patience will solve. The gap doesn't just exist between fields. It compounds within them, and it compounds with seniority. Holman and colleagues map authorship position carefully: first author typically signals an early-career researcher; last author typically signals the senior scientist who led the lab or project. Women are relatively well-represented as first authors. They become progressively rarer as you move toward last author, sole author, and invited author — the positions that carry the most weight for reputation and promotion. This is the leaky pipeline made concrete and statistical. Women enter the system in reasonable numbers in some fields, and then they disappear from the top. Journal prestige amplifies this further. Higher-impact journals have fewer women authors. Review journals — which tend to be prestigious and to publish invited contributions — have fewer women authors than non-review journals. Open-access journals, by contrast, contain proportionally more women. The pattern is consistent: the more elite the venue, the wider the gap. Next comes the finding about invitations, which is the most pointed mechanism in the paper. Holman and colleagues ran simulations on three thousand sixty-seven papers in their dataset that were identified as invited submissions. Men were roughly one point seven to two point one times more likely than women to be invited to submit. The paper summarizes this as men being invited at approximately double the rate of women. This isn't about who applies. It's about who gets called. Editors are making decisions to reach out to certain researchers, and those decisions systematically favor men. The pipeline leaks partly because someone is turning the valve. Now flip to geography, and the picture gets genuinely counterintuitive. You might expect that wealthy, research-intensive nations would show the smallest gender gaps — more resources, more progressive institutions, more support structures. The data say otherwise. Japan, Germany, and Switzerland are specifically named as wealthy countries with fewer women authors than many poorer nations. The team's mixed-model analysis put a number on this: countries one standard deviation above the average per-capita gross national income had five point five percent fewer women authors after controlling for discipline, journal, authorship position, and date. Country-level effects accounted for about eleven percent of the variation in women's representation overall. Holman and colleagues stop short of definitive causal explanations — they're honest about the limits of their data on that question. But they point toward cultural and sociological factors, as well as features of academic structures, as the kinds of influences worth investigating. The implication is clear: money buys equipment and salaries. It doesn't automatically dismantle the norms and institutional arrangements that push women out. So where does this leave us? The paper's conclusion is direct: the gender gap in STEMM authorship will not close without further reforms. Holman and colleagues are specific about what those reforms need to target. Education pipelines need to be reshaped to improve entry and retention of women, particularly in the fields that are most male-dominated. Mentoring structures need to actively support women toward senior roles — not just provide nominal access to them. Journal and editorial practices need scrutiny: double-blind review, the appointment of women editors, and explicit accountability for invitation practices. Conference organizers and hiring committees are named directly as actors who need to aim for representative gender ratios. The web app and archived dataset are framed as accountability tools — ways to track whether any of this is actually moving. This paper does more than diagnose a problem; it names levers and provides data and tools. It gives us a map — now someone has to walk the territory. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.