Hacker News Reader

It's pretty much true for any B-tree based index. You could get around it by using a combined index on another timestamp value, but that's going to be a very large index. Or reduce it by partitioning by e.g. year-month. Even hash-based indices would be slightly impacted due to less locality.

Thanks, I updated the comment to mention that the link is in the context of MySQL's case.

Is this applicable to postgres?

It's applicable to the indexes that postgres uses. Postgres does not put rows in the heap in order of primary key so the issue is less serious than in innodb.