ᔑᐸᐱᑚᕬᐱᑎ @skanman@qoto.org

> Should not the individual user control his own algorithms?

To the extent a user is capable of using different curation algorithms, the meaningful practical control they have over every one of those algorithms is in terms of how well they can understand the learned parameters and query the algorithm. I mentioned improvement in granularity; replacing one algorithm with entire another is too coarse a control for my ideal.

Case in point, Twitter allows two options: "top tweets" and "latest tweets". We probably agree that's not a lot of control. How many algorithms is enough control? How much should the algorithms differ each other? How do you describe those differences? You end up with the same problem.

Learning based algorithms are costly to train, in all of labor, money and time. I prefer one big algorithm which can produce different feeds by complex queries, than many algorithms which you still have to pick which one by complex queries.

To that end, for example, I consider the "Topic" feature of Twitter a step towards more complex query. You can choose to subscribe to a topic, you can also choose to exclude a topic. I agree with complaints that the topics are not precise or not granular enough.

Mastodon is great and all, but what is a sustainable model for financing the infrastructure, the management of all the instances, the risk is people, mainly volunteers, will crash under the load of growth in users, instances will dissappear etcetera, any good ideas or articles on this topic ? Where is the money 🙂 ? #mastodon #finance #money #infrastructure #sustainability