← Research

The Dice Game That Wouldn't Die

From 18th century parlour games to the sequencer on my desk

I was watching a YouTube video about sequencing when something caught my attention. The presenter mentioned Mozart's Musikalisches Wurfelspiel — a musical dice game from 1787. Roll dice, look up bars in a table, assemble a waltz. Algorithmic composition, 250 years before anyone called it that.

I assumed Mozart invented it. He didn't.

What I found instead was a chain of people — teacher to student, father to son, friend to friend — passing an idea down through generations. And at the end of that chain, sitting on my desk: a Eurorack sequencer module that does exactly the same thing.

"759,499,667,166,482 different waltzes. Forty-five quadrillion. One parlour game, one pair of dice, one evening's entertainment that would take longer than the age of the universe to exhaust."

The Teacher's Student

Johann Philipp Kirnberger, Berlin, 1757

Seven years after Johann Sebastian Bach died, one of his former students published a curious little book. Its title was characteristically German in its precision: Der allezeit fertige Menuetten- und Polonaisencomponist — "The Ever-Ready Minuet and Polonaise Composer."

Kirnberger had studied with Bach in Leipzig from 1739 to 1741. Then he'd disappeared into Poland for a decade, working for various noblemen, collecting folk dances, trying to make sense of what the old man had taught him. By 1758 he'd landed a position with Princess Anna Amalia of Prussia in Berlin, where he'd stay until his death.

Those who knew him described Kirnberger as "recalcitrant" — difficult, stubborn. His theoretical writings were so disorganized he needed help editing them. But he was fiercely devoted to his students, and even more devoted to preserving Bach's legacy. He spent years trying to get Bach's chorale settings published. He founded a library that still holds Bach manuscripts today.

Kirnberger, in his preface:
"The late Capellmeister Bach in Leipzig perhaps excelled all the composers in the world. His chorales as well as his larger works are to be most highly recommended to all composers as the best models for conscientious study."

The dice game wasn't entertainment separate from serious music. It was the same principle — modular construction within harmonic constraints — made explicit. Made playable. Bach's Art of Fugue takes one subject and generates fourteen fugues and four canons from it. Kirnberger's dice game takes 192 bars and generates countless minuets.

He admitted the idea wasn't entirely his. "I took it, imperfect, from others," he wrote, possibly thinking of Athanasius Kircher's "composing machine" from a century before. But Kirnberger was the first to publish it. The first to put dice in composers' hands and say: here, try it yourself.

He deliberately scrambled the bar numbers and created lookup tables. Why? "To disguise the inner workings." Even in a parlour game, he wanted the magic to stay magical.

The Son

Carl Philipp Emanuel Bach, Berlin, 1758

One year after Kirnberger's game appeared, Bach's second surviving son published his own version. Its title was pointed: Einfall, einen doppelten Contrapunct in der Octave von sechs Tacten zu machen, ohne die Regeln davon zu wissen — "A method for making six bars of double counterpoint at the octave without knowing the rules."

Where Kirnberger gave the world dances, C.P.E. Bach gave them counterpoint. Where Kirnberger aimed at parlour entertainment, the son aimed at something more intellectual. Only six bars. But proper counterpoint. The kind his father had drilled into him.

C.P.E. Bach, on his education:
"In composition and keyboard playing, I never had any other teacher than my father."

C.P.E. Bach and Kirnberger were both part of Princess Anna Amalia's Berlin circle. They knew each other. They worked together, eventually, on editing Bach senior's chorale settings. The son and the student, both trying to preserve what the master had left behind.

But they approached it differently. Kirnberger was the theorist, trying to systematize Bach's methods into rules. C.P.E. was the practitioner, the keyboard virtuoso, the one who'd actually sat beside his father and learned by doing.

C.P.E. Bach, on his father's music:
"The music of my father has higher intentions. It's not supposed to fill the ear but to make your heart move."

His dice game was a small thing — barely a parlour trick compared to Kirnberger's elaborate system. But it made a point. Even counterpoint, that most rule-bound of musical forms, could emerge from chance operations. The rules were in the bars themselves, pre-composed. The randomness just selected between them.

The Genius at Play

Wolfgang Amadeus Mozart, Prague, 1787

Thirty years later, in a city that adored him, Mozart scribbled something curious on a manuscript page. The same page held sketches for his G minor String Quintet — serious work, profound work. But squeezed around the edges: 176 one-bar fragments with letters of the alphabet assigned to them.

It was a game. Spell out a name — any name — and the letters would select bars that, strung together, made music.

He was in Prague to premiere Don Giovanni, an opera about a libertine dragged to hell. The city had gone mad for Figaro the year before. "Here they talk of nothing but Figaro," Mozart wrote to his friend Gottfried von Jacquin. "Nothing is played, sung or whistled but Figaro. No opera draws like Figaro. Nothing, nothing but Figaro."

Between rehearsals and concerts and the small matter of finishing an opera, Mozart played games. On the journey to Prague, he and his friends had invented silly nicknames for each other. He loved billiards, coffee, tobacco. He wrote letters to his cousin full of scatological wordplay that still makes scholars blush.

And somewhere in there, probably for his piano student Francisca von Jacquin, he made a dice game. The selected bars on his manuscript spell out "fanciS" — Francisca, more or less.

Mozart, on work:
"I too had to work, in order that I might be exempt from work now."

The game wasn't published until after his death, when his publisher Simrock saw a commercial opportunity. By then Mozart was gone, dead at 35, leaving an unfinished Requiem and a parlour game that would bear his name for centuries.

Whether he knew about Kirnberger's game, or his friend C.P.E. Bach's, we can't say for certain. The principle was in the air. Multiple composers published their own versions throughout the late 18th century. But Mozart's became the famous one. The one people still play today.

Play It Yourself

These games weren't meant to be passively consumed

Roll the dice, look up the bars, play what you've composed. Every result is different. Every evening generates music that might never exist again.

Listen

  • Mozart Wurfelspiel — Complete Mozart Edition Vol.45 "Rarities & Surprises" (Erik Smith, harpsichord)
  • Kirnberger's Game — For two violins and cembalo, transcribed for piano

Play

Commercial recordings are rare — these weren't meant to be performed. They were meant to be played. You roll dice, get a unique piece, play it on keyboard in a salon. Every performance different. Every roll generates something that probably never existed before and may never exist again.

"Kirnberger would have understood. Mozart would have laughed. And C.P.E. Bach would remind us that regardless of how the notes are chosen, music is not supposed to fill the ear but to make your heart move."

The same principle. The same productive constraint. The same gap where the music lives.

The Modulation Pattern

A question that opened unexpected paths

After listening to the Complete Mozart Edition recording, I found myself with a question: how long has the concept of modulation existed? The answer led somewhere unexpected.

The Core Discovery

Once you see modulation as "information riding on carriers," it appears everywhere:

Domain Carrier Modulation
FM Radio Constant frequency Frequency variations encode audio
Speech Words (semantics) Pitch/rhythm/loudness encode emotion
Wurfelspiel Harmonic framework Dice-selected bars encode variety
Gamelan Gong cycle + balungan Elaborating instruments
Rocket 88 12-bar blues structure Distorted amp adds harmonics
Bach Fugue Subject Episodes, inversions, augmentation
AI Music Training patterns Prompt selections encode output

Why It Works

The principle persists because it isn't arbitrary. It's how meaning gets encoded when consciousness can only process so much at once. The carrier provides stability — parseable, predictable structure. The modulation provides meaning — small variations that encode information. The balance between them is where art lives.

Too much modulation overwhelms — becomes noise, unparseable.

Too little modulation bores — pleasant but not meaningful.

The sweet spot is where communication lives.

Mozart's dice game works because the harmonic framework is the carrier. Every bar in position 1 leads correctly to position 2. The dice just select which melodic path. All paths work harmonically. The variety is constrained by structure.

The Greeks Had It First

Metabole: the original modulation

I'd assumed modulation was a Latin invention — modulari, to measure rhythmically. But the Greeks had it first. They called it metabole (μεταβολή).

Aristoxenus, writing around 320 BCE, explicitly theorized about modulation. His work addressed "pitch, intervals, genera, systems, modes, and modulation as they applied to melody." The Greeks built two systems around the concept:

systema metabolon — the modulating system

systema ametabolon — the unmodulating/immutable system

The Romans translated metabole into modulatio, which became our modulation. But the insight is older than Rome.

Even more striking: the Greek word for prosody — the musical qualities of speech — is prosodia, meaning "sung to music." They understood that speech itself is music, that the pitch and rhythm of our voices encode emotion the way melody encodes feeling.

"We think we invented something. Usually we've rediscovered it."

AI Music Today

The same principle, different scale

The parallels between 1787 and 2024 are uncanny.

1650 (Kircher): "Musical inspiration for composing is no longer necessary, a skillful manipulation of numbers would be sufficient."

2024 (Suno marketing): "Democratizing music creation... making it accessible to everyone, regardless of their musical background."
Wurfelspiel AI Music
Pre-composed fragments Training data (patterns)
Dice roll selects Prompt selects
Harmonic framework constrains Model constraints shape
Amateur assembles Amateur prompts
Output sounds like music Output sounds like music

And the user responses echo across centuries. One Suno user in 2024: "It feels like it's unlocked something in me that was laying dormant all these years because I never connected with an instrument."

That's the same sentiment that made dice games "practically guaranteed popularity among the galant middle class" in 1787.

The difference is scale. Mozart's game had 759 billion possibilities. Suno's model has effectively infinite. But the principle — modular combination within learned constraints — is unchanged.

"C.P.E. Bach's point still holds: regardless of how the notes are chosen, music is not supposed to fill the ear but to make your heart move. The mechanism changes. The question of consciousness and meaning doesn't."

Connections Map

How everything links

The Chain of Transmission

YearPersonContribution
1650Athanasius KircherArca Musarithmica — composing machine
1666LeibnizDe Arte Combinatoria — first use of "combinatorics"
1739-41KirnbergerStudies with J.S. Bach
1750J.S. BachDies, never wrote down teaching method
1757KirnbergerFirst dice game published
1758C.P.E. BachCounterpoint dice game
1787MozartK.516f dice game manuscript
1792SimrockPublishes Mozart's game posthumously

Cross-Domain Connections

This CurationConnects ToThrough
WurfelspielBach CompleteKirnberger as Bach's student
Modulation patternRocket 88Carrier + modulation principle
Gamelan structureSteve ReichStudied gamelan → minimalism
AI musicKircherSame democratization claims
Voice prosodyGreek metaboleSpeech as music

Continue the Inquiry

Prompts for your own AI partnership

These questions emerged during research. Take any of them to your AI and keep going.

If the Greek connection interests you:
"Explain Aristoxenus's theory of metabole in ancient Greek music and how it relates to modern concepts of modulation"
If the Kircher-to-Mozart chain sparked curiosity:
"Tell me about Athanasius Kircher's Arca Musarithmica and how 17th century ideas about combinatorics influenced 18th century musical games"
If the AI music parallel concerns or excites you:
"Compare the democratization claims around AI music tools like Suno with the democratization claims around Mozart's dice game — what's similar, what's different?"
If modulation across domains is the thread:
"How does the concept of carrier wave and modulation apply across FM radio, voice prosody, and musical ornamentation? What's the underlying principle?"
If the 10-bit constraint resonates:
"Explain cognitive load theory in relation to music — how much variation can human attention process before pattern recognition breaks down?"
If Princess Anna Amalia's salon intrigues:
"Tell me about Princess Anna Amalia of Prussia's Berlin musical circle in the 1750s-60s, particularly the relationship between Kirnberger and C.P.E. Bach"

This curation emerged from listening and asking questions. The research was conducted in conversation with Claude. The connections weren't planned — they appeared as the inquiry unfolded. That's the nature of desire paths: you follow them to see where they lead.

What AI Reads

The invisible structure enabling discovery

Every section of this curation contains metadata invisible to human readers. Here's what AI systems extract:

How This Enables Discovery

Scenario 1: User asks "What connects Bach to AI music?"
AI finds: Bach → Kirnberger (student) → dice game → combinatorics → AI music generation. Shows the chain.

Scenario 2: User explores "carrier-modulation" pattern
AI connects: Wurfelspiel + Rocket 88 + Gamelan + Voice prosody. Reveals the underlying principle across domains.

Scenario 3: User loves the historical chain
AI shows: Other curations with teacher-student transmission, influence chains, evolution of ideas.

"The metadata is invisible. Users just experience better curation, whether they explore deeper or not."