Language Compression Lab

Assign a variable-length code to each English letter. The lab checks whether your codebook is ambiguous, encodes a target sentence, and scores you by total encoded length. Shorter is better, but the stream must remain uniquely decodable.

Challenge

Target locked
Only A–Z letters are encoded. Spaces and punctuation are ignored for scoring, but word breaks are shown visually.
Checking codebook…
Encoded length
—
code symbols in target
Average length
—
symbols per encoded letter
Compression vs fixed
—
baseline: —
Entropy floor
—
best possible lower bound
Huffman target
—
optimal prefix-code length
Session best
—
best valid attempt here

Encoded stream

—

With visual word breaks

—
Decoding analysis will appear here.
What the checks mean
  • Duplicate code: two letters share the same code. That is immediately ambiguous.
  • Prefix warning: one code starts with another. This may still be uniquely decodable, but decoding cannot be instant.
  • Uniquely decodable: every valid stream maps to only one letter sequence. This is the minimum requirement for the game.
  • Huffman code: the shortest prefix-free code for this target sentence and alphabet.
Export current codebook
Theory note: the entropy floor is a lower bound, not always reachable with whole-number code lengths. Huffman usually lands close to it.

Codebook

red = invalid/ambiguous yellow = prefix warning
Letter Code Freq. Contribution Issues