Marginalia
← blog

What a team reading group is for — and when it is not worth it

A team reading group exists to give a group shared vocabulary and shared references — not to transfer skills. If you need people to be able to do something by Friday, run a workshop. If you want six people to mean the same thing by "interface", read the same book and argue about it.

Getting that distinction right is most of whether the group is worth anyone's evening.

What does a reading group actually produce?

Compressed disagreement. Six people read the same argument, disagree about it in the same room, and afterwards a shorthand exists that did not before: "that's the second-system problem", "this is a Conway thing". Six months later that shorthand is doing real work in design reviews.

That is the output. It is not a certificate, it is not a skill, and it will not show up in a competency matrix — which is exactly why groups justified as training tend to be quietly cancelled by the second quarter.

When is it not worth running?

Four cases, fairly reliably.

When the team is under deadline pressure. Reading is the first thing to go, and a group that meets twice and stops leaves people feeling worse than never starting.

When attendance is mandatory. A required book club is homework, and adults respond to homework by doing the minimum. Voluntary and small beats mandatory and large every time.

When the book is the wrong shape. A 600-page reference does not produce disagreement, it produces guilt. Neither does anything the whole team already agrees with.

When there is no fixed slot. A group that finds time each fortnight has already failed; it just does not know yet.

How big should it be?

Four to eight. Below four, one absence kills a session. Above eight, the quiet people stop speaking and it becomes a lecture with extra chairs.

If more people want in, run two groups on the same book rather than one large one — and let them meet separately. Two six-person arguments are much better than one twelve-person summary.

Who should pick the book?

Whoever is going to run the sessions, with veto rights for the group. Committee selection produces the safest possible book, and the safest possible book is the one nobody argues about.

Rotate the picker after each book. It spreads the work and it broadens the shelf, which is the second-order benefit — a year of this leaves a team with a genuinely shared reading history rather than one person's taste.

Does everyone need to finish?

No, and insisting on it is the most common way these groups die. The session should happen at the scheduled time regardless of who finished, with a ninety-second recap for anyone in the middle. The mechanics of that are in how to run a book club at work.

What does help is visible progress before the session — one shared shelf where everyone's page number sits next to everyone else's. Not to rank people; to let whoever is running it see on Tuesday that the target was too ambitious and move it, rather than discovering it on Thursday.

What about remote and distributed teams?

It works better than most remote rituals, because reading is asynchronous by nature and the meeting is genuinely discussion rather than status.

Two adjustments. Keep the session shorter — forty minutes over video is longer than forty minutes in a room. And make the shelf the source of truth for progress, because in a distributed team nobody sees anybody else's bookmark by accident.

Should the group's reading be public?

Sometimes, and it should be a deliberate choice per book rather than a default. A team publishing "here is what we read this quarter and what we thought" is genuinely useful to people outside it — and to candidates. What should never leave the group is the margin notes, because half-formed disagreement is the point of the group and people stop writing it down the moment it might be quoted.

That is why sharing a book and sharing a note are different permissions, and only the first one exists.

The short version