Each bright star is one of my studies. Each faint one is a paper that a study cites. Studies that cite the same papers get pulled together, so the clumps you see are the ones that really do share a literature. Hover a star to see what that study is about; click it for the full entry.
Where I file each study. The constellations are real ones (Ursa Major, Cassiopeia, Orion, Lyra, Corvus, Crux), and the grey points are the places still open in each cell.
I took the reference list of every piece of work I have finished: 13 of them, citing 752 different papers between them. I matched the lists against each other, so a paper cited by two studies becomes one star rather than two. Then a physics simulation places the studies. Studies that share references pull on each other and every study pushes the others away, so once it settles, the ones that share a literature end up close. Each paper then sits in a halo around the study that cites it, or between the studies if more than one does.
Only 32 of those 752 papers are cited by more than one study, which is why most stars sit in a single halo. Those 32 are the pale gold ones, and a line is drawn between two studies wherever they share a reference, and the more they share the heavier the line.
The paler haze behind the halos is one step further back. I looked up the papers I cite on OpenAlex and read their own reference lists, which so far has worked for 273 of the 752. Each point in the haze is a work that at least 3 of my studies reach, each study through a different paper. At the first step, 32 of the 78 pairs of my studies share a reference. One step back, at least 54 pairs share 3 or more works that they reach through different papers.
Among the works the most of my studies reach are Algorithm appreciation (Logg et al., 2019), What can machine learning do? (Brynjolfsson & Mitchell, 2017), Algorithm aversion (Dietvorst et al., 2015), Human trust in artificial intelligence (Glikson & Woolley, 2020) and Machine behaviour (Rahwan et al., 2019), each by at least 8 of the 13. Hover the brighter points in the haze for a few more. The tracing is still going, between 9% and 91% of each study’s references so far. Only a traced reference can lead into the haze, so a study I have traced less shows up in it less. The haze moves no star and thickens no line, and every count one step back is a floor that should rise.
The proportion is the part worth standing back for. A working life of research is a small, well-lit patch of somebody else’s field. Every study is also listed in plain words on the research page.