From seeds to safe streams
Advanced Algorithms
Compare ordinary PRNGs, CSPRNGs, standard DRBGs, conditioning, unbiased sampling, entropy pools, and deterministic signatures.
6 instruments
One claim, three lenses.
Change depth inside any card. “Another example” adapts the lens to the situations you chose during onboarding.
One idea · three lenses
Repeatable dice and secret dice have different jobs
A game-world generator should rebuild the same mountain from the same seed. A key generator must stop someone who sees many results from finding the seed or predicting the next result.
One idea · three lenses
A small secret state can feed many requests
After collecting enough trustworthy surprise, a carefully built machine can produce a long stream without demanding a fresh physical event for every bit. Sometimes it pauses to mix in new surprise.
One idea · three lenses
Trade quantity for a cleaner stream
If a coin is a little lopsided, looking at non-overlapping pairs and keeping only head-tail or tail-head can remove that fixed lean—while throwing many flips away. It works only under its assumptions.
One idea · three lenses
Mix sources without counting the same surprise twice
Two rain gauges on the same roof do not prove twice as much rain. A pool can combine their readings, but it should only count what has an independent reason to be surprising.
One idea · three lenses
Do not squeeze ten tickets into six boxes unevenly
If you give each of ten equally likely tickets a number from one to six using leftovers, some boxes get two tickets and some get one. Throwing away the leftovers before retrying restores equal chances.
One idea · three lenses
Algorithms ask for surprise at different moments
One lock needs a secret key once. Another handshake asks for a fresh secret each time. Some signatures roll hidden dice for every message; others safely calculate that one-time value from the key and message.
Bench notes
Keep these three.
- 01
Simulation PRNGs and security CSPRNGs make different promises.
- 02
A DRBG safely expands an adequately seeded, protected state; it does not replace the seed.
- 03
Conditioning, pooling, sampling, and each cryptographic scheme need explicit assumptions.
Local field-note progress
Finished Advanced Algorithms?
This optional mark lives only in this browser. It does not score the lesson or lock what comes next.
Touch the model
Make the variables move.
The lab runs locally in your browser and is designed to expose the assumptions behind this lesson.