Thread by @mipsytipsy

Btw, people who accuse me of policing the definition of observability to fit only honeycomb features are incorrect.

(Not about the policing part; oh hell YES I do that. Words mean things, and that goes double for technical terms. Quit trying to confuse people!)

They are close, but backwards. First came observability:

* years of painfully working thru the problems
* months of painfully trying to figure out how to talk about them
* stumbled over the definition of observability in the mechanical engineering sense, had my mind blown

* researched control theory and controllability/observability, sat with it for a long time
* spent months working out the ramifications -- what properties would a software system need to have to be observable?

.... and then we built honeycomb to that spec.

You all have a vested interest in this squabble over the definition of the word "observability", btw.

Because you are all on the verge of running into this problem. Where your unknown-unknowns outstrip your known unknowns, and it's high cardinality turtles all the way down.

Either we have a technical vocabulary that lets us articulate the nature of the problem and reason about it, and compare and evaluate solutions...

or we don't, and we waste loads of time on voluble descriptions to try and avoid our eternal, ever-present confusion.