https://github.com/nostr-protocol/nips/blob/vending-machine/vending-machine.md#appendix-2-job-types

There's a lot of new stuff on the Data Vending Machine NIP that I've been adding over the past few days.

Today I added something I'm very interested in:

Algorithmic feeds

Why am I interested in this? Didn't we escape twitter because of it's damn manipulative algorithms?

No. The issue is not with the use of algorithms (hell, chronological sorting *is* an algorithm): the issue is with opaqueness, misaligned-incentives, universality, and non-voluntary nature.

Data Vending Machine feed algorithms are not only fully voluntary, they also have a very organic long-tail design, where thousands of algorithms can be easily designed (because Nostr has solved the discovery problem!) and where Data Vending Machines could serve a niche so specific that is composed of only 5 or 10 people and still be worthwhile running it.

Next I'm going to work on building one more reference implementation where I will have a data vending machine calculate a feed for nostr:npub1w0rthyjyp2f5gful0gm2500pwyxfrx93a85289xdz0sd6hyef33sh2cu4x that would be truly impossible to calculate on the client (lists that contains most zapped highlights within a specific domain, e.g. AI).