The Reflection Queue: Pin Hot Thoughts, Process When Attention Is Cheap

Choose your door

• Builder — Builder Track: start your queue today with three real pins.
• Architect — Architect Track: the attention-economics model.
• Translator — Bridge below.
• Validator — Gates at the bottom.

The story this comes from

Mid-build, good questions arrive at the worst time: "should these repos share a naming scheme?" pops up while you're three layers deep in a phone-number normalizer. Chasing it costs the context you're holding; ignoring it loses the question forever. Both options are bad.

The answer is a single file you always keep in the same place (call it reflection-queue.md): when a hot thought arrives mid-build, pin it — one or two lines, timestamped — and return to the work. Then, at a deliberate reflection session (a "Reflection Day"), drain the queue with full attention. It proved itself the same day it was created: the naming question got pinned mid-build instead of derailing the session, and became its own queued working session.

Translator Bridge

A teacher mid-lesson gets a great tangent question. Stopping loses twenty-four other students; brushing it off loses a real question. The seasoned move is the parking lot — "great question, it's going on the board, we're hitting it Friday" — and the crucial part isn't the board, it's that Friday actually comes. A parking lot that's never revisited is just a polite trash can.

Builder Track (DO FIRST)

1. Create the file (or page, or note) — one queue, always the same place. Splitting queues kills them.
2. Set the pin format: date + one line + enough context that future-you can reconstruct why it mattered. Example pin: 2026-07-11 — do the repos need a shared naming scheme? (came up during a cross-repo cleanup; feels like drift risk).
3. Pin three real items right now — things you've been carrying in your head. Feel the unload.
4. Put the drain on the calendar. The queue is only alive if drain day exists. During a drain: take ONE item, give it full attention, end with a decision or a named next action — not a re-pin.

Architect Track (MODEL FIRST)

The model is attention arbitrage. A question costs wildly different amounts depending on when you pay it:

So the queue is a buy-low policy: capture at the cheap moment, think at the cheap moment, never pay retail. It's also load-bearing for honesty (this tier's theme): unqueued questions don't disappear — they either interrupt (costing the build) or evaporate (costing the insight, silently, like Lesson 7's unchecked claims). The queue makes the loss visible and deferred instead of invisible and immediate.

System notes: the queue lives with your cross-project notes (Lesson 6's top layer) because it spans all projects; it feeds the diary/knowledge-base flywheel (raw pins → refined doctrine); and it needs exactly one invariant — every pin is eventually drained or consciously killed. Pins that silently rot turn the queue into the polite trash can.

Validator Gates

1. Gate 1: What are the two bad options the queue replaces, and what does each one cost?
2. Gate 2: Why must the drain session be scheduled rather than "when I get to it"?
3. Gate 3: What's the one invariant that keeps a queue from becoming a trash can?
4. Gate 4 (own words): Describe the difference between build-mode attention and reflection-mode attention in your own working week — where does each live?