zlacker

Debatable - supporting a small number of users on the public API may be a legitimate technical debt issue, and a running cost as the API can't change without a lot of documentation, release planning to support all those third party stakeholders etc. Your future internal work has to remain compatible with legacy design choices if you don't want to shutdown/change the existing public end points - the list of issues has potential to be pretty big. Public APIs by their nature can't introduce major change too often without upsetting existing communities.

If the API is solely for your own consumption, this can be simpler, and of course third party clients are harder to monetize as the kinds of ads you can serve are going to be restricted to what you can force a third party client to receive and render.

If the number of users on third party clients is really low, all of the above can carry more weight in internal business case style discussions too.

>>giobox+(OP)
Seems to me just better to entirely stop supporting the public api than to make the costs so ridiculously high. I mean then you're _still_ supporting it, yet you've basically scared almost all customers away. Charging a ridiculously high amount seems maybe like the worst approach of all.

>>Cogito+l3
I think you've probably described exactly whats happening - they do want to stop supporting the public API, but only for third party clients. There are other API access use cases they want to support. If the pricing kills third party clients but not the new use cases, that seems like a design choice to me.

They would instead rather charge far more money for data access for things like AI training etc, Twitter have also made similar changes to their own API to prioritize high bills for AI training use cases, not third party clients. That's at least how I see this change. The high pricing for these customers also removes the need to worry about the ad tech situation as is the case in the third party clients - you can just offer them an ad free feed at these prices for the training requirements.

I suspect the internal at Reddit desire to have less third party clients may well predate the AI discussion too, given almost all companies in this position eventually want to wind down those clients as history has shown again and again, for all of the reasons discussed in this thread.