Add ‘Minimally Unique Forms’ (MUFs)

@leewardbound had some good feedback on Twitter:

Yo I like this idea and all, but the name in your example is not really what most people would consider pronouncable or memorable. Isn't this actually more vulnerable to typos (considering all the accent letters) or worse, possible encoding/render errors in various chat clients?

My response:

Thanks for the feedback! Up till now, I’ve imagined lookups/pronunciations being done irrespective of accents. Perhaps the minimally unique form (MUF) that is presented to users should only show ‘em as a last-resort disambiguator? I’ll write a sketch to get MUFs given a user-set.

I sketched up some code this morning that still has a kink that needs to be worked out, so I haven’t committed it yet. It does generate minimally unique forms, though. It’s currently using random numbers, but the worst I’ve seen so far (of about five tests) for a first-name collision was a set of ~5700 numbers. That’s not too bad for a bunch of random numbers. I think the worst case is with two adjacent odd numbers, though (but that’s not too much of a problem since Ethereum addresses are pseudo-random). Analysis still needs to be done on this, though.

Currently, this is the order in which MUFs are determined:

Stripped first
Stripped first and last
Stripped first, middle, and last
First, middle, and last

curlysemi / nume

Add ‘Minimally Unique Forms’ (MUFs) #1

Thanks for the feedback! Up till now, I’ve imagined lookups/pronunciations being done irrespective of accents. Perhaps the minimally unique form (MUF) that is presented to users should only show ‘em as a last-resort disambiguator? I’ll write a sketch to get MUFs given a user-set.