Well, take a look at the dates of when Postgres was created and when SSDs become available. Better, find articles about internal algorithms, B-trees, times of operations like seeks etc. The Postgres was initially written with disk operation timings in mind, and the point is that's changing - and I haven't heard of Postgres architecture changing with that.

Can you share examples of new database architectures and products using them that are built for SSDs?

I'm sure we have different capabilities and constraints, but I am unaware of any fundamentally different approaches to indexes.

That blog post is very light on details can be condensed to a single line/paragraph. LSM trees are more efficient for SSDs and modern databases use them.

I don't know enough to comment yet but will go read about it.