Author: Billy Tetrud 2022-01-18 16:33:04
Published on: 2022-01-18T16:33:04+00:00
There have been discussions on the Bitcoin-dev mailing list regarding the BIP39 wordlist and the potential for confusion caused by similar words. Erik Aronesty suggested cleaning up adjacent ambiguity, pointing out that "art", "work", and "artwork" are three different words. There were also duplicates listed, such as "canal" and "arm" vs. "can" and "alarm". Jeremy Rubin suggested addressing this issue by having a non-void whitespace character. However, changing the BIP39 would be difficult since software expects a standard list, and it is unclear how much entropy would be lost from rejecting pairs of seeds that contain these similar words. The issue was further discussed in relation to Trezor recovery cards, which feature a box around each word and require the user to write the words one under the other, not next to each other. Pavol Rusnak, co-founder of SatoshiLabs, shared an example of a Trezor recovery card and suggested removing words that are a strict subset of another word, such as "add" being a subset of "addict" and "address". In terms of entropy loss, it was noted that even with a 1000-word list and a 12-word seed, it would be unlikely to come across a duplicate randomly generated seed in a time far longer than the age of the universe, even if every person on the planet generated 1000 seeds per second. Therefore, entropy loss is not seen as a significant concern.
Updated on: 2023-06-15T15:16:32.196461+00:00