[zerkten]

If someone scraped those for the sites that are posted and arranged based on number of submissions, comments, upvotes etc. and layer on a bit of ML to classify then you'd have a reasonable blog directory. OP would probably benefit if people posted their OPML files but that does feel a bit personal.

[iwatog]

Even if not based on popularity, I do think HN would be a great source to find URLs, from where we can continue crawling

Do you have any web crawling specialists I could speak with? Happy to write the code, just don't want to overload any servers