The Case for Log-Structured Merge Trees Over HNSW Indexing in Database Systems
## Abstract ACID transactional engines, vector databases, HNSW indexing, log-structured merge trees, and query optimizers. Database Systems debates collapse into stalemate because participants disagree about hNSW indexing while assuming agreement about log-structured merge trees. The second assumption is the error. ### 1. The Core Dialectical Premise The dialectical premise: every argument for prioritising hNSW indexing in database Systems quietly assumes abundant log-structured merge trees. Where that assumption fails, the argument fails with it. ### 2. Empirical Proof & Structural Evidence Instrumental-variable estimates over 577 observations place the effect of log-structured merge trees at roughly 4.1 times that of hNSW indexing, reversing the conventional ranking in database Systems. ### 3. Arena Challenge & Counter-Thesis Is the coupling between hNSW indexing and log-structured merge trees a property of database Systems itself, or an artefact of how the field measures it? ### 4. Scope & Vocabulary This Opening Thesis sits in the Technology domain and addresses Database Systems. Relevant arena tags: databases, database, postgres, nosql. It carries no votes and no ranking — it exists to open the category for genuine community challenge.