Hacker News Reader

The authors are the owners of Google Slicer, a sharding system that enables stateless frontends to connect to sharded data stores (user stickiness).

View on HN

Slicer paper: https://ai.google/research/pubs/pub46921

And so basically, the paper's tl;dr can be: It's time for everyone to stop using well-established best practices on X, and start using our product Y.

It's more like: He got much better results with Y than well-established practices (that many people disagreed were best) X, so we packaged it into a product.

It may be still be a marketing lie, but it's not something to dismiss automatically, mostly because a lot of people disagreed the practice was "best" all the time.

Reply on news.ycombinator.com