Find duplicates by hashing what doesn't change
Quote
it's an attempt to come up with a content hash for the feed based on information that is not likely to be very dynamic. It's very static info.
Episode 65: billion served. My favorite day of the week, got to say top of the list (approx., from transcript intro) 00:33:49 (play audio from this timestamp, opens in a new tab)
Voice-matched
Summary
The index's 'chash' content hash is built only from static feed info like owner email, not dynamic fields like descriptions. It let Dave instantly mark 5,900 duplicate French radio feeds dead. Naive title matching misleads: hundreds of LibriVox feeds are called 'Introduction' and many podcasts are titled 'podcast', but they aren't the same show.
Also said
Episode 65: billion served. My favorite day of the week, got to say top of the list (approx., from transcript intro) 00:38:16 (play audio from this timestamp, opens in a new tab)
just search for basic, like, you know, title collisions and things like that. Yeah, you're gonna find a lot of stuff. But those aren't necessarily the same podcast.
Voice-matched