Climbing My Academic Family Tree

I found the Mathematics Genealogy Project online, typed in a name out of idle curiosity, and lost an entire evening.

If you have not met it before: MGP is a database of doctoral advisors. Not papers, not citations, not h-indices — just who supervised whom, going back centuries. Every doctorate is a node, every supervision an edge, and the whole thing is one enormous directed graph of intellectual descent. For someone who spends his working life on graphs and data, this is catnip.

So I did the obvious thing and started climbing my own branch.

The Belgrade generations

I defended on 10 May 2021 at the Faculty of Organizational Sciences in Belgrade, on integrating domain knowledge into the development of data mining models. My advisor was Milija Suknović, with Boris Delibašić chairing the committee and Zoran Obradović as a member.

Milija defended in 2001, on a methodology for group decision support. His advisor was Milutin Čupić, who defended in June 1981 — and here the tree quietly changes buildings. Čupić’s doctorate was not in organizational sciences at all. It was at the Faculty of Mechanical Engineering. He went on to become one of the country’s authorities on decision theory and quantitative analysis.

Čupić’s advisor was Dobrivoje Jovanović, 1972, also Mechanical Engineering, on social progress and the development policy of industrial communities.

Two generations up from a thesis about data mining, and we are firmly among the machines.

A committee to be reckoned with

Jovanović’s advisor was Vukan Dešić, and Dešić is where the story gets good.

Born in Azanja in March 1900, Dešić was one of the founding figures of the scientific organisation of work in Yugoslavia. He established the Chair for Scientific Organization of Work at the Faculty of Mechanical Engineering, and in 1949 introduced a course called “Methods of Scientific Organization of Work” — which is, more or less, the intellectual seed that the Faculty of Organizational Sciences later grew from. He read Fayol, Taylor, and Weber and adapted them rather than importing them wholesale, building out methods for assessing how well an enterprise was actually organised. He was also, at various points, Director General of the Yugoslav State Railways, assistant minister of transport, and a consultant to Energoinvest, Zastava, and Elektronska industrija Niš.

He defended his thesis in March 1953 at the Mechanical Institute of the Serbian Academy of Sciences. And then I read his examination committee:

• Milutin Milanković
• Anton Bilimović
• Vladimir Farmakovski
• Slobodan Dobrosavljević
• Pavle Stanković

I had to read the first name twice.

Yes, that Milanković

Milutin Milanković — the man behind Milanković cycles, the astronomical theory of the ice ages. The one whose calculations of how the Earth’s orbital eccentricity, axial tilt, and precession combine to drive long-term climate are still the backbone of how we understand glacial periods. NASA has listed him among the fifteen most significant contributors to geophysics. He also proposed the calendar reform adopted at the Congress of Constantinople in 1923, which is why the revised Julian calendar tracks the astronomical year better than the Gregorian one does.

He was, on top of all that, a genuinely good writer — his Canon of Insolation of the Earth and the autobiographical Through Distant Worlds and Times are not the prose of a man who merely tolerated language.

And in 1953, he sat on the committee that examined my academic great-great-grandfather.

Further up, into probability

Milanković himself defended at the Technische Hochschule Wien in 1904, with a dissertation titled Theorie der Druckkurven — the theory of pressure curves. Here the Mathematics Genealogy Project picks the trail back up, listing him with two advisors: Johannes Brick and Emanuel Czuber.

Brick was a bridge engineer, professor of bridge construction and structural mechanics, twice rector at Brno and later rector and dean in Vienna — a specialist in getting iron and reinforced concrete to span things.

Czuber is where it becomes almost too neat. He was one of the founders of modern probability theory and actuarial mathematics, and his 1876 habilitation in Prague was titled Theorie und Praxis der Ausgleichsrechnung — the theory and practice of adjustment calculus. Which is, in the language of 1876, least squares.

I make a living fitting models by minimising squared error. Six defences up my family tree, someone was writing the book on it.

The rest of that Vienna examination bench is no less impressive. Ludwig von Tetmajer founded the first Swiss institution for testing building materials — today’s Empa — and after investigating the 1891 Münchenstein rail disaster produced the Tetmajer equation, which extended Euler’s theory of column buckling into the range where real columns actually fail. Josef Finger was a pioneer of continuum mechanics; the Finger tensor in finite deformation theory is named after him.

An honest caveat

I should be careful about what this tree actually shows.

For the Belgrade generations, “advisor” means advisor in the ordinary sense. But for Dešić in 1953 and for Milanković in 1904, what the records give me is the examination committee. MGP lists Czuber and Brick as Milanković’s advisors, and Brick is separately documented as having sat on his defence — but the further back you go, the blurrier the distinction between supervisor, examiner, and the professor whose lectures shaped you becomes. Doctoral supervision as we practise it now is a fairly recent institutional invention.

So this is less a clean chain of custody than a family tree drawn by an enthusiastic amateur after a large dinner. I am fine with that.

Where the trail goes cold

Czuber’s own advisor is listed as unknown. Prague, 1876, and the line simply stops — no further edge in the graph.

That is oddly satisfying. Every genealogy has a horizon, and mine is a Bohemian mathematician working out how to reconcile imperfect measurements, one generation before the technique became so ordinary that we now hand it to first-year students.

One last coincidence

Anton Bilimović — Odessa 1907, another name on Dešić’s 1953 committee — went on to found the Mathematical Institute of the Serbian Academy of Sciences and Arts, along with its journal Publications de l’Institut Mathématique and the Yugoslav Society for Mechanics.

I have given invited lectures at that institute twice, in 2020 and again in 2025, talking about fairness in machine learning. Both times I walked into a building founded by a man who, seventy years earlier, helped examine the thesis that my own doctorate descends from.

Nobody tells you these things. You have to go looking.

A humbling footnote

I was feeling quite pleased with myself about all this. Milanković! An institute founded by a man who examined my academic great-great-grandfather!

Then my good friend Michelle González Amador, whom I know through the EAAMO community, ran the same exercise. I am not in the same league.

Like me, Michelle is not in MGP herself — she defended in 2025 — so her tree starts one step up, with her advisor Robin Cowan (Stanford, 1987), whose thesis was called Backing the Wrong Horse: Sequential Technology Choice Under Increasing Returns. Cowan’s advisor was W. Brian Arthur, of increasing-returns and lock-in fame. Arthur’s advisor was Stuart Dreyfus — of Bellman and Dreyfus, Applied Dynamic Programming. And Dreyfus’s advisor was Arthur E. Bryson, Jr., often called the father of modern optimal control, whose work with Yu-Chi Ho is routinely cited among the direct precursors of backpropagation.

So her academic line runs straight through the prehistory of the algorithm I use nearly every working day. Mine gets to least squares. Hers gets to gradients.

And then it simply keeps going. One branch climbs Emil Hilb → Ferdinand Lindemann → Felix Klein → Rudolf Lipschitz → Dirichlet → Poisson → Lagrange → Euler → Johann Bernoulli → Jacob Bernoulli → Leibniz. Another runs through Rheticus to Copernicus, and from Copernicus to Regiomontanus. Elsewhere in the same tree: Laplace, Fourier, Gauss.

The deepest branch leaves Europe altogether. Through Gemistos, Kydones, Kabasilas, Palamas, Metochites, Bryennios and Chioniadis it reaches Nasir al-Dīn al-Ṭūsī, then Omar Khayyám (1068), then Bahmanyār ibn al-Marzubān, and finally Avicenna — Ibn Sīnā — whose own three teachers are each listed with no advisor of their own. Late tenth century! That is where her tree runs out of graph.

My line stops in Prague in 1876, with a mathematician doing least squares. Hers stops around the year 1000 in Persia, with the author of The Canon of Medicine.

I am keeping Milanković. But I concede the point entirely, and with considerable admiration.

Go and look

If you have a doctorate, go and look yourself up. Worst case you lose an evening. Best case you find out you are distantly related to the ice ages — and if you are very lucky, to Avicenna.