Podcatr.com bills itself as “the podcast directory powered by machine learning”.

It’s new to me. A few tests shows it’s pretty good as a categorisation and discovery tool for podcast shows. Better, in terms of relevance, than the increasingly screwy Listen Notes. The latter has now gone from only allowing four pages of search results before accusing the hapless user of trying to pirate ‘their’ listings, to now allowing no public results at all for either ‘Episodes’ or ‘Podcasts’ — unless you’re logged in…

Podcatr on the other hand is public and it makes a useful choice in how it presents results within the topic categories for shows. The shows are “sorted exclusively by freshness”. Thus rewarding timeliness rather than popularity. This feature seems to work best if you feed Podcatr’s search-box something more specific than a single keyword, e.g. “H.P. Lovecraft” rather than just Lovecraft. In effect it’s a sort of subtle “what’s new” that you might drop in on every few weeks, just to keep in touch with who’s covering a topic. Thus potentially enabling you to back off from the ‘firehose effect’ of taking a daily look at everything new for a keyword at the episode level, in date order, just in case something is missed.