Alternate data doesn't even need to be as sexy as satellite photos, hell you almost certainly want the data that isn't sexy, the stuff people haven't thought of because it's too boring. Alternative data vendors above all want sales, and even the funds themselves want things to show off to clients. This gives you great opportunities to look at the alternative data they aren't touching.

Given this is a predominantly a web development community it always surprises me how little creativity there is in the articles on investing. Neural networks and machine learning sound cool but the reality is almost none of the readership would be able to make any money off them.

Simply tracking how many sales or users exist in databases by watching sequential IDs should be the go-to method for any web developer trying to get an edge. I would have expected HN to have articles where people are getting creative on that, ie trying to use measures of entropy on usernames to get rough subscriber numbers etc.

Even plain scraping of prices etc, is often full of great insight that is ignored. If a grocery store drops their prices in profitable categories against their competitors, that could be the signal about an incoming price war for an entire sector. There's a lot more information in that than social media feeds and all the other sorts of sexy data that get coverage in the media.