Follow
tech ramble
Bruh. Prolly Trees are just so much easier to use than Merkle Search Trees. Just the construction alone ends up being so much more simple.
Like, just reading the overview from Aaron Boodman is so much easier than the MST paper and construction/updating/reading is so much easier.
IMO ease + sequential reads are more important than "layers" being similar sizes and containing items.
MSTs are a step up from the CHAMP/HAMT craze though.
tech ramble
One thing I like about the MST paper is the focus on sparse replication of CRDTs which I think is crucial to consider for any sort of distributed application that wants to do CRDTs for their data model with a large ish number of "documents" to modify. One thing MSTs don't help with is compaction of CRDT data to help deal with long histories. Unsure how one would best address that tho.