pull down to refresh
We only allow alphanumerics and underscores in nyms, so WeAreAllSatoshi implemented a Damerau–Levenshtein distance IIRC.
But time is better spent removing friction right now given how friction-full the onboarding is.
reply
The problem with levenshtein distance is that it treats changing O to K the same as changing O to 0, it's more about number of operations distance than visual similarity
reply
We added a dictionary for visual similarity, but you're right we should use an algorithm that is purpose built for visual similarity.
reply
Regarding #2, ChatGPT pointed me to this: https://unicode-org.github.io/icu-docs/apidoc/dev/icu4j/com/ibm/icu/text/SpoofChecker.html