Just coded up a recursive relay crawler using NIP-65 events for discovery. Over the course of a few hours it discovered a total of 2,761 unique URLs for relays (after discarding a variety of malformed or otherwise unusable relay tags.) Unfortunately, actually going out and attempting a connection with each, only *800* were online and queryable! Not sure what to make of this. Clearly many people need to update their kind 10002 relay lists...