Apologies to anyone who has, in fact, implemented these at scale. Please tell me how it went!
- Fussy recommendation — Items must successively receive endorsement from larger and larger groups of people, to earn the right to be show to yet more people
- Participatory budgeting of collective attention — Society votes on how much attention should be spent on family, on politics, on culture etc. (though a lot more granular). Recommender system does its best to allocate quotas in accordance with the outcome.
- Reverse recommendation — People vote on what content is definitely not worth the attention of other people. Such content gets buried. Otherwise its a free-for-all.
- Recursive epistemic virtues — Some online fora (like Tim van Gelder’s YourView) award reputation based on endorsements from your outgroup. What if we took this a step further? Base reputation on endorsements from your outgroup, weighting more heavily those members of the outgroup who are respected by your ingroup, weighting more heavily those members of your ingroup who are respected by your outgroup, weighting more heavily those members of the outgroup who are respected by your ingroup , … ad infinitum. Do we just end up with PageRank? (or EigenTrust?) (or EigenMorality?)
- No recommendation — A blank feed.
- Classics-based recommendation — A recommender that recommends items in proportion to their age and how popular they are. The idea being that it only presents content with enduring, long lasting value. Would need to account for a popularity-related feedback loop.
- Meaningful recommendation — Values/meaning based recommender for spaces (à la Joe Edelman).
- Quadratic/plural recommendation — Designed by analogy with quadratic/plural voting and funding mechanisms.
- Georgist recommendation — A Georgism-based recommender. People with large followings are expected to pay rents or keep posting regularly. If they fail to do these, they start to lose followers.