What is the best score value to be used to hold the transactions?

[GrahamMcBain]

When using the watchman library what is the best score value to ensure you don't get too many false positives while also not missing any names that should be flagged?

[GrahamMcBain]

(These are responses from the slack thread about this question)

"Its a give and take on how many false positives that you want to deal with. At 85% you will get a lot of false positives but will catch everything bad. At 92% you won’t get many false positive and you “should” catch everything but, maybe."

[GrahamMcBain]

I was reading some papers recently to get an academic view of how to set this. Wanted to cite those and write something up for our docs/FAQ

I think the largest factor is your input length. Longer names will match better and using as much of the name as possible.

e.g. john doe doesn’t match as well as john clancy doe

Also ?name=…&address=… is a special code path in watchman where it’ll combine those scores.
?q=… isn’t going to score the same

(from @adamdecaf )