Why? 16 million isn’t much, maybe 10 to 20 gigabyte if compressed.

I’ve seen IRC bouncers that have more messages stored for a single user (270 million, in fact).

You could do that with the Algolia API: https://github.com/minimaxir/get-all-hacker-news-submissions...

But as noted, BigQuery is more pragmatic.

I have a script running on aws right now.