Virtues of smaller versus larger spreads
Question
Virtues of smaller versus larger spreads? Should I branch out, or be happy with what works?
Answer
The short answer
Smaller spreads are sharper, larger spreads buy coverage you usually don't need — and the real gain from "branching out" isn't more detail, it's a different set of *seats*. The number of cards isn't the quality lever; whether every seat lands on something you could go look at is.
| Spread | Seats | Built for | Wrong when |
|---|---|---|---|
| `single` | today | one posture for today, a practice pull | you need to compare options, or seat two people |
| `three` | past / present / future (or you / them / field) | one decision, one timeline | more than one actor or a rhythm over weeks |
| `diamond` | core / root / obstacle / potential / advice | stuck — you know the problem, not the snag | you only want today's line |
| `moon` | new / first quarter / full / last | a month, a habit, an emotional cycle | a yes/no that needs an answer tonight |
| `horseshoe` | 7, incl. *external influence* | a stretch of time plus outside pressure | one card would do |
| `celtic` | 10, incl. *hope & fear* and *outcome* | multi-actor, tangled, long-running | anything you could ask in three — don't default to it for completeness |
What each size actually buys
Small (1–3). Forces the middle seat to be the hinge and the read to be one causal chain rather than three separate opinions. Equally important, it forces the reading to end in *one* move you can take. The failure mode of small isn't thinness — it's asking it a question with two people in it.
Mid (4–5). `diamond` adds two seats a timeline has no room for: what's blocking you, and what's already available but unused. That's usually the most useful upgrade a stuck question can get. `moon` is the only spread here about rhythm, and it reads as *intent → action → seeing → letting go* — not as a climax in the full-moon seat.
Large (7–10). Real virtues, honestly: `horseshoe`'s *external influence* seat puts a system, an institution, or other people in a place a three-card timeline structurally cannot. `celtic` separates *hope & fear* (position 9) from *outcome* (position 10) — and that's a genuine gift, because 9 is psychology and doesn't rewrite 10, and 10 doesn't prove 9's fear is real. Read 5→6 as one near timeline: what just happened sliding into what's about to.
The cost. Every seat past what your question actually contains is padding, and padded readings are vaguer — you start picking the impressive-sounding card and skipping the one you can't act on. A ten-card spread is not a more serious reading. It's a bigger surface to avoid one hard seat on.
Your question: branch out, or stay
If most of your questions are about you and one decision at a time, what works *is* the right size — that's not stagnation, it's fit. Branch out when the question itself outgrows the seats: a second actor shows up, an outside body has a hand in it, or you're asking about weeks rather than days. Go up one size, not to ten. And the test of whether the bigger spread earned its keep is whether the extra seat produced something you could go check — a document, a person, a date, a record, something that recurs. Seats you can't narrate that way are seats you didn't need.
The one step
Before your next pull, write the question as two lists: *who is in it* and *over what span of time*. Then pick the smallest spread that gives every one of those a seat, and write the spread name down before you draw. That's a 2-minute habit and it decides the spread for you, instead of you deciding it by feel.
Want to test it on a real question of your own — name the question, and I'll pull whichever spread your two lists call for.