In newer versions we now use a different order of requests to reduce the amount of needed queries per server. BTW: All the queried endpoints are valid in some versions of Friendica (which at one point hat the name Friendika) and GNU Social (which had the name Statusnet before).
The current server detection looks like this: github.com/friendica/friendica…
In current versions we search for "/.well-known/nodeinfo" first. And we only query those endpoints above only when this query hadn't been successful (and also when the "host-meta" path is present). This seems to work fine with Smithereen (I just did a quick test against a known server).
@heluecht added logging, now gotta wait for these requests to have my curiosity satisfied :D