What it is
A set of private pages, built on the work a person has published. There is nothing to install, nothing to sign up for and nothing to enter: the observatory arrives finished, and from then on it keeps itself up to date.
It starts from a public identifier — the one under which the researcher is already recorded in the literature — and from there it takes their work, their field, and those who cite them.
A lifetime's work, divided into seasons
A lifetime of output cannot be told all at once: it is cut into seasons, and the seasons are not decades. They are decided by the rhythm at which that person actually published — they tighten where the papers grow dense.
Across a real career of fifty-one years and 1,717 papers, the observatory produced eight seasons: fifteen years for the first seventy-three papers, three years for the last two hundred and seventy-one. 1,715 papers out of 1,715 ended up on a map: none left out, none counted twice.
The maps of the field
Beside one's own work there are the maps of the themes being worked on, built on the literature of that field — with one's own papers marked inside, to see where one stands in relation to everything else.
There is also the world map: where in the world this research is being done today, which institutions, and where the citations come from.
The questions, and the courage not to know
You put questions to your own field and receive answers that cite only real papers, with the reference beside every sentence.
What we are proudest of is what happens when the answer is not there. In one real case, asked which assay demonstrates that a substance is carcinogenic, the system replied that none of the papers it held described that kind of assay — instead of assembling a plausible answer.
And it did the thing that actually helps: it proposed widening the field, with the topic already drafted. Once the topic was approved, within minutes 195 new papers and 22 full texts came in; the same question, asked again, answered with twelve sources.
The library
Of the papers that make up a field, some have their full text freely available: the observatory fetches those itself and keeps them in house, so that answers come from whole texts and not from abstracts alone.
Across that 1,717-paper career, 843 — 49% — were open access. The proportion varies a great deal between disciplines, and we say so upfront: in some fields publishers open almost nothing.
Retracted papers
For every paper cited, the observatory checks whether it has since been retracted. Across a career with almost ten thousand references it found sixteen.
It is not a number anyone finds by hand, and none of the researchers it happened to were aware of it.
Every day, and every week
A daily radar of new papers published on one's own themes. A weekly watch on who is citing that work, and on what is moving around it.
And a workbench for preparing figures, diagrams and presentations from one's own material — with diagrams that advance when the speaker decides, not before.
What it is not
It is not a search engine: it is not for searching all of science, it is for looking down on a single field.
It is not a general-purpose assistant: it answers only on the papers it holds, and when it does not hold them it says so.
It is not an archive to fill in: the researcher enters nothing. They receive.
Who it is for
Today we have built and tested it in biomedical research, and that is where we know it works. Outside that area we have an attempt, not a proof: we prefer to say so, rather than promise something we have not yet seen.
The pages belong to whoever receives them
They open with a private password, and no researcher's data enters another's observatory: that is a condition verified by an automated check on every release, not a promise.
They also work without a network, from a memory stick, with a double click — and on an older tablet, or with three-dimensional graphics switched off, they still open, in a lighter version that arrives on its own.