So the real innovation that's needed is a new incentive structure.

I'm sure some site has tried to implement a system where you pay only if you end up in a long term relationship, but that seems hard to enforce once two people have each other's contact details.

Another approach that's been tried is for the government to run the site, on the basis that families are beneficial for society, while profit-seeking monopolies and incompatible silos of users are not efficient.

What I don't think has been tried yet is some sort of federated system, perhaps based on ActivityPub, where different providers all share the same distributed pool of users, and therefore can't lock you into their unreasonable pricing scheme.

The success of Mastodon shows that this service could be offered at scale for people to enjoy at zero cost. Possibly there would be an expectation that if you did find someone you really liked, you would make a donation to the one or two sites that you and your partner's profiles were hosted on, and perhaps give a testimonial at the same time.

I assume the difficulty of implementing this is that people don't want bots spamming them with private messages, nor do they want their profiles to be part of some bulk download which a rogue node tries to get away with.

People are understandably quite sensitive about what they write on dating sites, and who gets to see it, and although the safety of traditional dating sites is probably somewhat illusory, a system for moderating interactions is likely to be harder to implement well in a distributed system, and harder than moderating a microblogging service.

That sounds like a minefield. I have an additional idea: what about a dual-purpose site, one that has value for couples too? Saying this makes me realize Facebook, although their thing is new, might be able to cormer this actually...