Click an exhibit to approach it. Map choices jump instantly. Arrows follow your Walk/Instant preference. Drag to turn your view. W/S or ↑/↓ walk forward/back; A/D or ←/→ strafe sideways. Recenter restores the exhibit.
Open Rooms & map to choose a destination. Walk uses a brief departure and arrival with a fade between rooms. Reduced motion makes travel instant. Flat keeps the same experiments in a reading layout.
The 3D gallery is unavailable here. All twelve experiments are ready in Flat.
01 / 1936 · ALAN TURING
A machine made
of instructions.
A tape. A reading head. A few rules.
An idea that changed what a machine could be.
Read transcript
George · AI narration by ElevenLabsWelcome to nineteen thirty-six. Before laptops, before the internet, before a printer could ruin your morning... Alan Turing asks a startling question: what does it actually mean to compute?
His answer begins on paper. Imagine a tape divided into squares, a head that reads one symbol at a time, and a small set of exact rules. Read. Write. Move. Repeat. No intuition. No inspiration. Absolutely no improvising.
It sounds almost laughably simple. But here comes the leap: Turing describes a universal machine. Give it the encoded instructions for another computing machine, and it can imitate that machine's work. One design, many different jobs. The program changes what the machine does.
And there's a twist. Turing's work also exposes limits: some mathematical questions cannot be settled by a general mechanical procedure. More patience won't fix that. Neither will a very stern email to technical support.
The sculpture here brings that abstract idea into view. It's our teaching model, not a machine Turing built in nineteen thirty-six. In this little experiment, three becomes four through three instructions. Watch each symbol change. Nothing guesses. Every move has a reason.
When you're ready, try the machine. Then step into nineteen forty-three, where the question changes: could simple switches, connected together, begin to model a neuron? Your next room is waiting.
0 / 3 instructions
Can three become four, one instruction at a time?
The reading head starts at the right. Run the first instruction.
Read 1 → write 0, move left.
Read 0 → write 1, stop.
A question becomes a conversation.
YouExplain gravity.
YouMake that simpler.
Ask a question, then change the answer with a follow-up.
What makes a logical neuron fire?
Both inputs active → output 1.
Otherwise → output 0. This switch does not learn.
Find one signal among eight equally likely possibilities.
Each balanced yes-or-no answer removes one bit of uncertainty.
“What makes a good friend?”
Reply concealed
Reply concealed
Both replies are scripted.
What if you could guide a computer in your own words? Step into the next room.
A three-cell teaching model, not a replica of a historical machine.