Markov Babbler

A demo of the p/moul/x/daily/markov library — an on-chain port of Go's classic "Generating arbitrary text: a Markov chain algorithm" codewalk. Every two-word prefix maps to the words that followed it; Generate walks that chain, picking suffixes with a PRNG seeded by the block height — deterministic, yet different each block.

Fresh sample (40 words, height 165613)

it was the age of wisdom it was the worst of times it was the age of wisdom it was the best of times it was the age of wisdom it was the worst of times it was the best

Corpus

• Words: 48
• Distinct prefixes: 33
• Prefix length: 2 words

Some prefixes → suffixes

Call Feed("your text here") to teach it new words, then reload for a new sample.