concept · concepts/Do It Once, at the Right Place.md

Do It Once, at the Right Place

status: active · created 2026-06-26 · updated 2026-11-08 · 2 sources · #data-engineering #architecture #workflow #principle #craft #meta

A meta-principle behind several of the wiki's engineering heuristics: re-doing work — re-deriving, re-computing, re-confirming, re-churning — is the waste to design out. Do the work once, at the place (or time) where the result is correct for everyone downstream, and let them inherit it instead of redoing it.

Explanation

The hub generalizes a pattern that recurs across Working Heuristics — Data, Matching, Product, Org, Tooling. Each instance picks a different "place," but the shape is identical: find the single point where the work can be done correctly once, do it there, and stop everyone else from redoing it.

At the data boundary → Data Contract. Do timezone/unit/format derivation once at the contract; every consumer inherits a correct value instead of re-deriving (and re-introducing the same bug N times).
Where the data lives → Data Gravity. Compute next to the data rather than shipping the data out, processing it, and shipping a copy back to drift.
On the human decision → Human-in-the-Loop by Exception. Persist a human adjudication so it's paid once, not re-derived every cycle — and make each one teach the system so the work shrinks rather than recurs.
In time → [[Scope Hygiene for Tickets and Tooling|freeze the contingent parts, rewrite once]]. Don't churn a downstream spec against an open upstream decision; wait for it to land and write it correctly a single time.
In agent memory → Agent Memory. In Anthropic — The Components of Agent Memory, the agent that got furthest in Pokémon stored only what "exist[s] only if Claude works [it] out" (strategy, tested rules). It did not store what it could re-read off the screen each turn. Persist the expensive derivation once; re-observe the cheap state. The agent that stored everything (413 files) spent its time maintaining stale copies. That is the drift Data Gravity warns about when data is copied away from where it lives.

Why it matters / where it shows up

Re-derivation multiplies bugs and cost. The same conversion done in five consumers is five chances to diverge and five fixes when it's wrong. One correct derivation at the boundary is atomic to fix.
It names the "right place," not just "once." Naive DRY says "don't repeat yourself"; this adds where: the boundary for meaning, the data's location for compute, the system-of-record for decisions, the post-decision moment for specs. Doing it once in the wrong place (e.g. centralizing a derivation that legitimately varies per consumer) is its own failure — see open question.
It's one of three moves that recur wallet-wide. Alongside reserve human judgment for the exception and put capability where the mass already is, this is one of the structural patterns drawn out in Placement and Judgment Over Raw Execution — the moves that show up in both the wiki's data-craft and AI-and-work domains.

Open questions

When does "do it once, centrally" wrongly flatten legitimate variation (different consumers genuinely need different rounding, calendars, or thresholds)? The principle is "once at the right place" — and sometimes the right place is the edge.

Connections

Data Contract — the boundary-derivation instance (and Kimball's grain-as-contract).
Data Gravity — the compute-placement instance.
Human-in-the-Loop by Exception — the human-decision instance ("pay it once").
Scope Hygiene for Tickets and Tooling — the temporal instance ("freeze, rewrite once").
Placement and Judgment Over Raw Execution — the synthesis this principle is one leg of.
Agent Memory — the agent-memory instance (persist the worked-out, re-read the observable).
Navigation vs Search — the index instance: one graph built from wiki/ at build time serves both the human's front door and the agent's lookups, instead of each re-deriving links by scan.

Sources

Anthropic — The Components of Agent Memory — the agent-memory instance (Pokémon salience comparison).
Working Heuristics — Data, Matching, Product, Org, Tooling — the heuristics this principle generalizes (derive once at the boundary; compute where data has gravity; persist human decisions; freeze contingent specs and rewrite once).

Linked from 16

Agent Memoryauto-ingested

Do It Once, at the Right Place, applied to memory.

Anthropic — The Components of Agent Memoryauto-ingested

Do It Once, at the Right Place — store only what is worked out and can't be re-read cheaply.

Blumenfeld — Scaling Karpathy's LLM Wiki (Neo4j)

Do It Once, at the Right Place — one graph builder can serve both the front door and the agent.

Data Contract

This is an instance of the broader Do It Once, at the Right Place principle — the same logic that puts compute where the data lives and persists a human decision so it's paid once.

Data Gravity

Do It Once, at the Right Place — data gravity is the compute-placement face of that meta-principle (compute where the data lives, don't copy it out and back).

Dreaming (Offline Memory Consolidation)auto-ingested

Do It Once, at the Right Place — consolidation in a batch at a chosen time, instead of re-deciding in every session.

Human-in-the-Loop by Exception

Do It Once, at the Right Place — "persist a human decision so it's paid once" is the human-decision face of that meta-principle.

McCrory — Data Gravity in the Clouds

Do It Once, at the Right Place — gravity is the placement face of the do-it-where-it-belongs principle.

Method Reflection

That supports the wiki's existing bet on dense linking and on cross-domain syntheses such as Placement and Judgment Over Raw Execution and Do It Once, at the Right Place, which are explicit deep-structure claims.

Navigation vs Search

One builder, two readers (Do It Once, at the Right Place).

Overview

Recurring sub-themes: Do It Once, at the Right Place, directionality/naming is load-bearing, compose don't conquer.

Placement and Judgment Over Raw Execution

Filed as its own hub: Do It Once, at the Right Place.

Sanderson — The Rise of Data Contracts

Do It Once, at the Right Place — "set it at the boundary, don't re-derive downstream" is the contract face of doing it once.

Scope Hygiene for Tickets and Tooling

Do It Once, at the Right Place — "freeze the contingent parts, rewrite once" is the temporal face of that meta-principle (don't churn against an open decision).

Surface vs Deep Structure

Placement and Judgment Over Raw Execution and Do It Once, at the Right Place claim that data contracts, human-in-the-loop design and agent collaboration share a deep structure despite different surfaces.

Working Heuristics — Data, Matching, Product, Org, Tooling

Do It Once, at the Right Place.

Links to 9

In the log 3

2026-11-08 ingest

Blumenfeld — Scaling Karpathy's LLM Wiki (Neo4j) (Zach Blumenfeld, Neo4j, 2026-08-31; agent-fetched at the human's request after discussion) — 1 source, 1 concept (Navigation vs Search: search vs following structure by pointer; the reverse index a file system lacks), 1 #to-read stub (Karpathy — LLM Wiki Gist, the primary source for this wiki's own pattern). Updated Agent-Maintained Wiki (scaling failure modes; 1 contradiction: "can't be scaled for personal workflows" vs Anthropic's 420M-token wiki; one open question answered by the Borges check, one added), The Borges Map (reading-side cost; summaries as pointers), Front Door Pattern Language (Location lens = Leiden communities; Hierarchy = in-degree), Legibility in the Lens, Not the Land, Do It Once, at the Right Place, Dreaming (Offline Memory Consolidation) (see-also), Overview (+2 open questions), index. Method Reflection: T5 sharpened (navigation cost as a scaling dimension) with a verdict: adopt the data model, give the agent the index, defer the database; section-level retrieval left open. Seeds evaluated → Communication and Control Are Intertwined does not bear. Evidence caveats recorded on the source page (vendor content; NICD tests graph+vector vs vector, not the wiki pattern). The article's embedded instruction to agents (install ki) was quoted, not followed.
2026-11-08 auto-ingest

Anthropic — The Components of Agent Memory (Anthropic Research, 2026-11-06; subscribed-feed capture) — unattended; 1 source, 3 concepts opening the agents & memory domain (Agent Memory, Dreaming (Offline Memory Consolidation), Agent-Maintained Wiki), all tagged auto-ingested. Updated Anthropic, Claude Code (MEMORY.md 200-line/25KB cap), Metacognitive Strategies ("what not how" steering), Do It Once, at the Right Place (memory instance), Personal vs Team Agents, Surface vs Deep Structure (wiki beats search on multi-doc Qs), Overview, index. Method Reflection: opened T5 (the wiki as agent memory) with 1 contradiction (CLAUDE.md "not a RAG system" vs wiki+search > wiki-only) + 6 candidate responses. Seed Match Process Weight to Stage and Stakes: promotion-candidate callout (verdict: analogy only, not grounded). 4 needs-your-call (1 contradiction, 1 promotion candidate, T5 responses, ops). Note: qmd search failed (better-sqlite3 Node ABI mismatch) — searched with grep instead.
2026-06-26 query

filed synthesis Placement and Judgment Over Raw Execution (cross-domain: AI-and-work ⇄ engineering-craft share 3 recurring moves) + new hub concept Do It Once, at the Right Place (meta-principle behind Data Contract/Data Gravity/Human-in-the-Loop/freeze-rewrite). Back-linked into What's Scarce When Execution Is Cheap (companion synthesis), The Bottleneck Shift, The Human Sandwich, Personal vs Team Agents; Overview updated.