Open RoelN opened 4 years ago
Collecting data on the fonts themselves seems undesirable.
But we can collect some data about usage. How to determine usage with non-interactive elements though?
For each thing that's detected - each language, each ot feature, each axis tag, each axis range - you could increment a counter.
Then you can use this aggregate data to know what is more commonly detected and therefore should have more/better treatment.
Eg, if small caps are very common but slashed zero is rare, a prominent section that shows off in detail what the former looks like is going to be more worthwhile developing than the latter.
a good reason to opt-out
No, as an offline-first app that specifically states up front that no fonts are uploaded, this is a good reason to make any usage reporting opt-in.
if small caps are very common but slashed zero is rare, a prominent section that shows off in detail what the former looks like is going to be more worthwhile developing than the latter.
You could probably get a better idea of what features were rare vs. common by surveying recently released fonts. Also you might want to do the opposite: surface unusual features more prominently than run-of-the-mill features.
Neither of those points are to say collecting usage stats is bad ... I think it can be a great thing. I'd focus more on usage patterns on the page than on collecting data about the fonts people submit, and I would absolutely make it opt-in, but sure it would help inform development and UI decisions.
It would be nice to keep logs on which fonts are dropped, which types, etc. All with as little privacy-sensitive data as possible.
Twitter thread: https://twitter.com/PixelAmbacht/status/1280104631750795264
Immediately a good reason to opt-out: