01
Change a condition and look
Instead of reading an explanation, raise the humidity and the wind, apply a gate to a qubit, load a model onto a GPU. Changing one thing at a time makes the reason the result moved visible.
TOPIC COLLECTION
Interactive Science
7 postsEducational experiments where you change the conditions and watch a scientific or computational idea respond. The formula, the assumptions, and the ways it differs from real observation or real hardware sit next to the result.
What this subject covers
02
The formula and the assumptions
Where the number on screen came from, and what was held fixed. Educational estimates and real measurements are never mixed.
03
Where it parts from real hardware
An idealised simulation has no noise and no error. Where it starts to diverge from real quantum hardware or real GPU telemetry is written next to the result.
Groups in this topic
Quantum Mini-Labs4 posts
Four steps from bit to interference
Weather and Climate5 posts
Public observation data, redrawn
Problems Met in Production6 posts
What to check when you are stuck at the server
- Selects, merges, sorts: request matching in nginx, Apache, and Caddy
- From the name you typed to an IP, in four steps
- The distress signal a server sends: 502, 504, 500
- Fully booked, nobody in the room: what nvidia-smi 0% means
- Press the last seat at the same time and both bookings succeed
- Press it twice to see why Redis is fast
How to read these
- The answer to the question in the title comes first. The demo is where you check that answer yourself.
- The screen marks whether a number is an educational estimate or an observation.
- For values touching health or safety, individual variation and the relevant cautions sit beside the result.
Posts
1–7 of 7
Problems Met in ProductionDemo
Fully booked, nobody in the room: what nvidia-smi 0% means
VRAM nearly full while GPU-Util reads 0%. Not a fault, but a serving engine that reserved its KV pool in advance. Loading a model, running inference, and toggling pre-reservation shows that memory occupancy and compute utilisation are separate things.
2026.07.247 min↗
Quantum Mini-LabsPart 1Demo
A qubit's first impression is just a bit
Quantum mini-labs, part one. Pressing NOT on a classical bit and X on a qubit side by side, and finding they do exactly the same thing on |0> and |1>. An idealised single-qubit simulation.
2026.07.226 min↗
Quantum Mini-LabsPart 2Demo
Measuring a qubit ten times gives you the wrong conclusion
Quantum mini-labs, part two. Measuring an H-gated qubit 1, 10, 100, and 1,000 times, and seeing how one result differs from the 50:50 distribution repetition produces. An idealised single-qubit simulation.
2026.07.226 min↗
Quantum Mini-LabsPart 3Demo
Equal probabilities do not mean the same qubit
Quantum mini-labs, part three, going one step further. An H-gated qubit (|+>) and an H-then-Z qubit (|−>) both measure 50:50, and on a 2D compass they point in opposite directions. Probability alone does not describe a state.
2026.07.226 min↗
Quantum Mini-LabsPart 4Demo
Quantum interference: a hidden phase inverts the result
Quantum mini-labs, the last part, going one step further. |0>─H─H always gives 0, and inserting one Z to make |0>─H─Z─H always gives 1. The phase that was invisible to measurement in part three meets a second H and inverts the outcome.
2026.07.226 min↗
Weather and ClimateDemo
The same 33 °C feels different because of humidity, wind, and sun
Why 33 °C on the thermometer can feel like several different temperatures, worked through as three side-by-side comparisons.
2026.07.217 min↗
Weather and ClimateDemo
100 mm of rain is fifty water bottles per square metre
How many litres 100 mm of rain becomes over 1 m², 10 m², and 84 m², worked out with one multiplication.
2026.07.217 min↗