[gnso-idng] A summary of assertions and open questions, continued, maleware

[Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>]

There is a categorical difference in the problems presented by 
near-ASCII like character repertoires like Cyrillic, relative to the 
letters-digits-hyphen subset of Latin.
In the IDNAbis WG we've covered this in dealing with the edge cases 
presented by mixing scripts.
We can adopt similar guidelines to reduce risk introduced by visually 
similar glyphs, for instance, no mixing Cyrillic and Latin script in 
the same label.
As an approach, we can take an ordered-by-risk list from who ever is 
leading the maleware charge, and see which can be eliminated for the 
class of applications we're considering. If we get to "acceptable 
risk", an as yet undefined term, but .cat, or .museum, in {Chinese, 
Yiddish, Arabic, Hindi, ...} seems like a safe lower bound on 
realistic risk, then we have something for this overarching issue too.
At present we have nothing, which is underwhelming.

Eric