Two 90-minute interludes, read as one arc. Each idea links to a hands-on workshop widget and a library card.
90 minutes · how do ordinary sensors fail, and how do we quantify the failure?
The arc: your own senses are a feedback system → a sensor is a transducer with three competing virtues → a phone can weigh the Earth, if you respect its noise → and a clock can keep time so well it becomes a gravity detector.
We begin with the sensor suite you were born with — eyes, ears, skin, the vestibular system. These are not passive channels: they feed a real-time processor (the brain) that builds a model of the world, predicts, and acts. Catching a ball is the whole loop in one second — see, predict, move, correct. Every engineered sensor mirrors this: detect, convert, interpret, and (sometimes) feed back.
A sensor transduces a physical change into a signal we can read, usually digitised through an analog-to-digital converter (which already costs us resolution). Its quality is captured by three numbers that trade against each other: sensitivity (the smallest change it resolves), dynamic range (smallest to largest), and bandwidth (how fast a change it follows). No device maximises all three — a sensor is a choice of corner.
workshop drag the trade-off triangle · library anchor 1 card
A MEMS accelerometer — a microscopic proof mass on a spring — lets your phone measure gravity to about a part in 103. Do it carefully and the lesson is systematics: flip the phone and the two readings disagree in the third digit. To know what your number means you need the noise toolbox — the mean and standard error, the power spectral density, and the Allan deviation, which tells you exactly when averaging longer stops helping.
workshop the Allan / PSD explorer · library anchor 2 card
Clocks are sensors of time, and they span an astonishing range — from a planet's orbit to a single ion whose optical transition oscillates 1015 times a second (its electronic clock frequency — far faster than the ion's ~MHz trap motion). The best optical atomic clocks reach 1 part in 1018. At that precision relativity intrudes: raise the clock by a centimetre and it ticks measurably faster — so a clock becomes a gravitational sensor. The floor that limits a single clock's averaging is the standard quantum limit, 1/√N.
workshop the 1/√N wall · library anchor 3 card (the SQL)
A network of sensors uses correlations — shared information — to sharpen precision and accuracy beyond any single device. Quantum-mechanically, entanglement (stronger than correlation) takes this further, enabling collective measurements below the classical limit. That is the doorway into Sensing II.
To carry with you: a network of observers — sensors, or people — combines noisy local data into something more reliable. How can shared information reduce uncertainty, and how can discussion reduce misunderstanding?
90 minutes · what can quantum resources improve, and what do they make fragile?
The arc: a ladder of quantum resources bends the 1/√N limit → but every rung you climb is more fragile → squeezing buys a modest, robust gain by moving noise around → and the whole game is balancing what you gain against what you break.
Quantum metrology is a ladder. A coherent probe sits at the standard quantum limit, 1/√N; squeezing lowers the prefactor to ξ/√N; multipartite entanglement — GHZ (Greenberger–Horne–Zeilinger) and NOON states — reaches the Heisenberg limit, 1/N; and error-corrected entanglement aims to keep 1/N in a noisy world. The catch — the spine of the whole session — is that a stronger resource is also a more fragile one.
workshop climb the quantum ladder · library anchor 4 card
Comparing two optical clocks is precise enough to ask whether the fine-structure constant drifts, and to set limits on ultralight dark matter coupling to ordinary matter. The same spin-½ toolbox runs on nitrogen-vacancy (NV) centres in diamond — quantum sensors that work at room temperature, even inside living cells, trading some coherence for ruggedness and spatial resolution.
Quantum sensing moves signals between photons (light) and phonons (mechanical motion of trapped ions). The same recipe — prepare a vacuum, displace it, make Fock or squeezed states, then read out — applies to both; only the hardware differs. Reading a tiny motion means transducing it: ion motion → spin → fluorescence photons.
workshop photons vs phonons · library anchor 6 card
Every measurement disturbs its conjugate observable — that is back-action, and Heisenberg fixes the product ΔX·ΔP. Squeezing doesn't beat that product; it redistributes the noise, narrowing one quadrature below the vacuum (the SQL) at the cost of the other. LIGO injects squeezed vacuum at the dark port, with frequency-dependent squeezing, to push gravitational-wave noise below the standard quantum limit.
workshop squeeze the vacuum · library anchor 5 card
The frontier today: a single trapped ion resolving nanometre displacements and zeptonewton forces; LIGO resolving space-time strain of about 1 part in 1021 (a strain-noise spectral density near 10−23/√Hz); a spin-squeezed strontium clock at 10−18. The practical rule that ties them together: balance gain against fragility. A few dB of squeezing often gives the best net sensitivity today; Heisenberg-scale entanglement pays a decoherence tax unless you protect it with error correction.
To carry with you: if we could measure without back-action — completely and repeatedly — what would we lose about the world? In quantum mechanics the probe is never fully detached from what it measures; perhaps that is true of inquiry more broadly.