SE Radio

The proposed split is four layers: ingest and routing, storage, query and compute, and visualization. The parallel Eric draws is to business intelligence, where the warehouse and the dashboard stopped being one product. A coupled observability vendor sells all four together, and that bundling is where both the lock-in and the shape of the scaling bill come from, not from any one layer being bad.

Portability is a schema problem before it is a storage problem. OpenTelemetry earns its place in this architecture mainly as a shared schema: if telemetry is written in a form more than one backend can read, then changing the query layer is a configuration change rather than a re-instrumentation project across every service.

Query and compute is where the difficulty concentrates, because observability data wants real-time analytics behavior rather than batch. That drives the discussion into Druid's indexing choices and into returning one unified result across hot and cold storage, so an engineer chasing an incident does not have to know which tier their time range happens to live in.

The first cut is between ambiguity that is reducible and ambiguity that is inherent. Reducible ambiguity yields to more context, and the categories worth separating are lexical, syntactic, and pragmatic, with voice adding homophones and accent variation on top. Inherent ambiguity does not yield to any amount of context, and treating the two the same is how systems end up confidently guessing.

Personalization runs on revealed preferences rather than explicit feedback: what a user keeps, edits, or throws away says more than a thumbs-up they were never going to click. That behavioral signal is also the antidote to output that regresses to the mean, where a model tuned to be acceptable to everyone writes prose that sounds like nobody in particular.

Surfacing uncertainty is a product decision, not just a modeling one. Flagging every low-confidence span turns the interface into a proofreading assignment, so the useful question is which uncertainties are worth a user's attention. Evaluation follows the same split: offline sets for whether ambiguity was resolved correctly, online signals for whether the resolution held up in real use.