As with large & random time ranges, querying many titles in a single request makes individual requests more expensive & hard to effectively rate-limit, and defeats CDN caching. It runs counter to the philosophy of REST APIs, which are focused on discrete resources & low per-request costs.

I would recommend to look into providing convenient client-side library abstractions for batch operations instead. These can benefit from CDN caching and parallelism, while providing similar convenient ergonomics for users.

See also:

• T134524: Pageview API: Limit (and document) size of data you can request
• https://en.wikipedia.org/wiki/Representational_state_transfer#Cacheable

I agree, and there's support for this in the python client I maintain, along with other stuff that we don't want to do on the server like month-level granularity for article-views. I share your opinion of being true to the spirit of REST, @GWicke.