Music is the one thing on this site that has nothing to do with research, strategy, or output. It is purely for itself. I want to write about it in that spirit — not as a topic to cover comprehensively but as something personal, the way you would describe a part of your life that matters to you but does not need to justify itself by being useful.

I mostly listen to classical music and what is now called neo-classical. I also make music, slowly, in Logic Pro and FL Studio — a project that is still unfinished and may remain unfinished for a while, which is fine. This page is about what I listen to, what I find in it, and what the project is.

What I Listen To

Classical Music

I came to classical music late — not from formal training but from listening, which means my relationship to it is idiosyncratic in the way that self-taught relationships to things always are. I came in sideways, through individual works rather than through a systematic education in the canon, and my tastes reflect that. I find some of the most celebrated works in the repertoire less interesting than some relatively neglected ones, and I have no embarrassment about that.

Bach is the composer I return to most consistently. The Well-Tempered Clavier is the work I have spent the most time with — two books of preludes and fugues in all twenty-four major and minor keys, written partly as a demonstration of equal temperament (that all keys are usable, that none is more natural than another) and partly as a set of compositional studies, and somehow both of those purposes are completely invisible in the finished music. Each prelude-fugue pair has its own character, its own emotional weight. The C major prelude of Book I — a sequence of arpeggiated chords, technically simple — is among the most beautiful things in music.

What I find in Bach that I do not find in the same degree anywhere else: complete structural transparency combined with complete emotional depth. You can follow exactly what is happening harmonically and contrapuntally — the voices are all clearly defined, the modulations are logical, the subject entries in the fugues are trackable — and it does not diminish the experience. Most music asks you to either analyse or feel. Bach lets you do both simultaneously.

The Goldberg Variations are a separate thing entirely — a set of thirty variations on a ground bass, written in 1741 (according to one story, to help an insomniac count named Goldberg sleep). Glenn Gould’s 1981 recording is the one I keep returning to — his second recording, made a year before his death, much slower and more interior than his famous 1955 recording. It is the sound of someone who has lived with the piece for thirty years.

Beethoven’s late string quartets — Op. 127, 130, 131, 132, 135, and the Grosse Fuge Op. 133 — are the music I find hardest and most rewarding. They were written between 1824 and 1826, after he was completely deaf, and they sound like it — not in a negative sense but in the sense that they are indifferent to convention in a way that is only possible when you can no longer hear what people think of you. The Op. 131 quartet in C# minor, in seven connected movements played without pause, is probably my favourite single piece of music. It starts with a slow fugue in the manner of Bach and ends somewhere entirely different.

Schubert’s late piano sonatas — the last three (D.845, D.894, and the final three D.958-960 written in the last months of his life in 1828) — are music of a particular quality I find hard to describe without sounding vague: they move slowly, they repeat themselves, they seem to circle around something they never quite say, and the effect is not monotony but something closer to patience. Schubert died at 31. The D.960 in B-flat major sounds like someone who knew he was dying and had decided to spend the time well.

Debussy — the Préludes for piano (two books, 1909–1913) and La Mer (1905). Debussy is the composer whose influence on subsequent music I find most underappreciated. The whole-tone scales, the parallel chord motion, the absence of functional harmonic progression in the traditional sense — these were genuinely radical innovations, and they ran directly into the neo-classical and ambient music that followed a century later. The neo-classical composers talk about Debussy constantly for a reason.

Arvo Pärt — the Estonian composer who developed what he called tintinnabuli style in the 1970s: a method of composition in which one voice moves in a simple melodic line while another voice arpeggiate the tonic triad. Spiegel im Spiegel (1978, for piano and violin or cello) is eight minutes of complete stillness. Fratres exists in many versions — for strings, for cello and piano, for violin and piano — and each of them is different and each of them is extraordinary. Pärt is the composer whose music I find most useful to work to: it has enough structure to feel like something is happening but enough space to leave your own thinking room.

Shostakovich’s string quartets, particularly No. 8 in C minor Op. 110, written in 1960 in three days in Dresden. The eighth quartet is autobiographical — Shostakovich quotes his own earlier works throughout, as if surveying a life. It opens with a slow four-voice fugue in the manner of Bach’s B minor prelude, and it ends in exhaustion. He wrote it, he said, for the victims of fascism and war, but he also wrote it for himself. The dedication was a way of getting it published in the Soviet Union.

Neo-Classical Music

The term is imprecise and increasingly contested — some of the composers associated with it dislike it, and it covers a range of approaches that are genuinely different from each other. What they share, roughly, is: an acoustic core (piano, strings, sometimes chamber ensemble), an influence from minimalism (repetition, gradual change, sustained attention to texture), and the integration of electronic production in varying degrees. It is music that takes classical training seriously and takes electronic production seriously and refuses to treat them as separate worlds.

Jóhann Jóhannsson (1969–2018) is the composer whose work I have spent the most time with in this genre, and whose death feels like the largest single loss to contemporary music I have experienced. Fordlandia (2008) and And in the Endless Pause There Came the Sound of Bees (2009) are his most purely neo-classical works; the film scores (Arrival, Sicario, Mandy) are where he demonstrated what the genre could do cinematically. The Arrival score in particular is extraordinary — he built the texture from recordings of actual sounds (human voices, instruments) processed beyond recognition, creating a sonic environment that feels alien without being electronic in the conventional sense. Flight from the City from Orphée (2016) is the piece I would use to introduce someone to what he was doing: delicate piano lines, subtle electronic textures, field recordings, a sense of enormous space with very little material.

He was also a deeply intellectual musician — he wrote about his compositional process with unusual precision, and his influences ranged from Medieval monophony and early polyphony through minimalism to post-rock. The breadth is audible in the music.

Ólafur Arnalds is the most technically innovative composer in this space currently. His use of the Stratus system — two semi-generative self-playing pianos controlled by algorithms that respond to his live playing — turns each performance into an improvised duet with a machine. The 2020 album some kind of peace uses this prominently. What I find most interesting about Arnalds is that the technology is never the point — the emotional directness is the point, and the technology is in the service of it. Near Light (2013) and saman (2012) are the works I return to most.

Nils Frahm is the most improvisational. Felt (2011) — recorded in the middle of the night with felt dampening the piano hammers to avoid waking neighbours — has an intimacy that feels accidental and is probably entirely deliberate. Spaces (2013, live) and All Melody (2018) show the range: from close, quiet piano recordings to full electronic soundscapes with pipe organ. He moves between acoustic and electronic instruments during live performances in a way that blurs the category distinction entirely.

Max Richter — Sleep (2015), an eight-hour work intended to be listened to while sleeping, and On the Nature of Daylight (2004), which has been used in films so many times (Arrival, Shutter Island, The Leftovers) that it is almost impossible to hear it neutrally now. Richter studied composition formally (Edinburgh and Florence) and his music shows it — the writing is architecturally clear in a way that the more improvisational neo-classical composers sometimes are not.

Ludovico Einaudi is the most popular and the most controversial within the genre — music snobs dismiss him as easy listening. I find this uncharitable. Divenire (2006) and In a Time Lapse (2013) are well-constructed and emotionally coherent. The minimalist repetition is intentional and effective. The case against him is that the repetition asks very little of the listener; the case for him is that music that asks very little can give very much.

A Winged Victory for the Sullen (Dustin O’Halloran and Adam Wiltzie) make the most cinematic music in this space — slow, massive, built from strings and piano and synthesiser drones that sustain indefinitely. The self-titled debut (2011) is the essential record. It is music that requires and creates patience.

The Connection to the Other Work

I want to say something honest about why I listen to this specific combination — Bach and the late Beethoven quartets and Arnalds and Jóhannsson — because I think it is not accidental.

The music I am drawn to has a quality I would describe as structural transparency with emotional depth: you can follow the architecture while feeling the content. Bach’s fugues are analytically visible — you can trace every voice — and emotionally direct. Jóhannsson’s film scores are built from audible materials assembled in audible ways, and they are profoundly affecting. This combination matters to me because I think it is the standard I try to apply in other domains: in writing (clear argument, genuine content), in mathematics (rigorous proof, meaningful result), in research (honest methodology, real finding).

I do not know whether this is a real connection or a post-hoc rationalisation. But the preference for structural transparency is consistent across everything I do, and music is where I feel it most directly.

What I Make

I have been working in Logic Pro and FL Studio on a neo-classical composition for longer than I would like to admit. It is unfinished. It may stay unfinished for a while. I want to write about it honestly rather than presenting it as further along than it is.

The Project

The piece is built around a piano theme — simple, slow, harmonically static in the way that Pärt’s tintinnabuli works are harmonically static, holding a single chord for long enough that you start to hear the internal structure of the chord itself rather than its function in a progression. The theme is in C minor, as it has a specific gravity that C major does not have, a weight without darkness.

The plan is to build the piece in three layers. The piano theme forms the base. A string texture — strings recorded dry and processed through reverb and subtle pitch modulation — forms the middle layer. And a synthesiser drone, sustaining throughout at a very low volume, forms the harmonic foundation that ties the other layers together.

What I am trying to do, technically and aesthetically, is build something that sounds inevitable — where each element feels like it could not be otherwise. This is harder than it sounds. Most music that aims at this quality achieves either predictability (the next note is obvious) or arbitrariness (the next note is unpredictable but also unmotivated). The target is something in between: surprising enough to maintain attention, inevitable enough that when you hear it, it sounds like the only possible choice.

The Technical Setup

Logic Pro for the main composition and arrangement. The built-in Alchemy synthesiser for the drone layer — it is remarkably good for sustained, slowly evolving textures. The Spitfire LABS samples (free, consistently excellent) for strings. The piano recordings are from Native Instruments’ The Gentleman — an upright piano with a slightly muted character that suits the close, intimate quality I am going for.

FL Studio for the electronic processing — the granular synthesis on the string recordings, the spectral processing that blurs the boundary between the string sound and the synthesiser. FL Studio’s parametric EQ and the Gross Beat plugin for timing effects. I use the two DAWs in parallel rather than choosing one: Logic for arrangement and MIDI, FL Studio for audio processing and electronic texture.

The workflow is slow by design. I am not trying to produce something quickly. The experience of working in the DAW — adjusting reverb tails by a few milliseconds, finding the right ratio of dry piano sound to room, deciding whether the string entry is a bar too early or exactly right — is part of what makes it interesting.

What It Sounds Like (Approximately)

If you know the neo-classical composers I have described above: it is closest in approach to Jóhannsson’s quieter work and to Pärt, with less of the electronic experimentation of Arnalds. There are no drums, no percussion, no conventional rhythmic structure — the rhythm comes entirely from the phrasing of the piano theme and the way the string texture swells and recedes against it.

The drone is the element I am least sure about. It works in context but I have not decided whether it is essential or whether the piece would be stronger without it. This is the central unresolved question and probably why the piece is still unfinished.

When It Will Be Done

I do not know. I am not working toward a deadline. When it is finished and I am satisfied with it, I will share it here.

A Note on Why Music Is Here at All

This site is mostly about physics, mathematics, finance, and philosophy. Music does not fit that picture neatly, and I am aware of the incongruity.

I have decided not to care about the incongruity. The people I find most worth reading and thinking about — Poincaré, Feynman, Penrose, Simons, Weil (both Simone and André) — were all people whose intellectual life extended well beyond their professional specialisation, and the extension was not incidental to the work. The different domains fed each other in ways that were often hard to trace but clearly real.

I do not want to overstate this. I do not think my neo-classical project is going to improve my Yang-Mills research. The connection is subtler and less instrumental than that. It is more like: the same aesthetic preferences that make me find Bach more satisfying than Liszt also make me find Cartan’s exterior algebra more satisfying than coordinate-based tensor calculus. Both are the preference for structure that is visible in the thing itself rather than imposed from outside.

Music is on this site because it is part of the same person.

Last updated March 2026. The composition project section will be updated when the piece is complete.