Reliability as improving the odds, not a fixed effect

## Spark

Lorin Hochstein's "Reliability as a Game of Improving the Odds" (highlighted this week) reframes all reliability work as shifting a probability and estimating its effect size, rather than buying a fixed guarantee. It anchored a wider SRE reading cluster this window: the coming coordination calamity, what makes incidents different from routine work, bring back ops pride, expertise and overload.

## Why it could be interesting

Short signal-boost shape — Hochstein's quote plus a paragraph of my own. The claim worth adding: **the odds-framing is exactly the right lens for solo infra.** I can't buy nines, and I have no incident volume to learn from, so every control (audit logs, Renovate bake periods, client-pushed backups) is an odds nudge whose effect size I'm mostly guessing at. Naming "what does this actually move, and by how much" is the antidote to cargo-culting controls because an org does them. Pairs naturally with the security-posture work this same week.

## Open questions

- [ ] Is "estimate the effect size" tractable at scale-of-one, with no incident base rate to calibrate against?
- [ ] Best as a standalone short, or folded into the personal-infra security-posture post as its framing?

## Source material

- https://read.readwise.io/read/01kseegq0y86q0rrr0yx6xj0h1 — Reliability as a Game of Improving the Odds (highlighted)
- https://read.readwise.io/read/01ksesdasf2d9erztzy4ecx3hg — The coming coordination calamity
- https://read.readwise.io/read/01knm7eq5erk6gp9704mn85rv3 — What Makes Incidents Different to Routine Work?
- https://read.readwise.io/read/01kq5bhf52rnxbxsx36erzt6gk — Bring Back Ops Pride