I am wondering what's behind this. Maybe it's based on Shazam but as far as I know Shazam has no public API. Coding such bot without an API against Shazam would be rather cumbersome - and I doubt somehow this is a Shazam-internal thing. So maybe it's based on another search engine - e. g. how does Mixcloud do their track recognition? Also the whole setup (i mean the user account) looks rather sloppy - looks like some 'script-kiddy' (not meant offensive) or some project in beta/test
Concerning the digital/beatport point - if someone would seriously want to build up a search index with tracks not released digitally he could quite easily find at least some more or less accurate sources. Youtube videos linked to discogs releases, record shops with online samples, ... things like this could be crawled automatically.
Totally agree that things like this reduce the fun - the whole social part of track IDing gets lost.