Deciding Dinner When There Are Two of You
Two people means a veto is cheap, which changes what a random pick is for.
With exactly two people, either party can veto at no social cost, so a single random answer is almost never accepted — which makes the two-person case the one where the draw should produce a shortlist rather than a decision.
With two people either party can veto at no social cost, which is why a single random answer is almost never accepted. The draw has to produce a shortlist rather than a verdict.
That is a genuinely different requirement from every group version of this problem, and it comes from the group size rather than from anything about food.
Why the veto is free at two
In a group of six, rejecting the random answer means overriding five other people, which most people will not do for a mild preference. In a group of two it means overriding one person, who is usually your partner, and the cost is close to zero.
So the answer gets rejected, and then the second answer gets rejected, and the mechanism has quietly become you choosing with extra steps. The draw did not fail — the social conditions that make a draw binding are absent.
The workable response is to change what the draw produces. A shortlist of three is not a verdict to accept or reject; it is a menu, and choosing from a menu is a thing two people will happily do.
The shortlist method
Draw three options, then choose between them by discussion or by a second draw. The what-to-watch picker implements exactly this two-stage shape for a watchlist and the same approach works for dinner with the dinner decider or a wheel.
The arithmetic is worth knowing: a three-of-n shortlist followed by a one-of-three pick gives every option exactly the same chance as a single direct draw. You have not distorted anything, only changed what gets shown.
What has changed is the psychology. Seeing the two options you did not take makes the one you did feel chosen rather than imposed, which is the entire difference between a draw that ends the conversation and one that starts a new one.
Twelve options, shortlist of three
- Each option has a 3-in-12 = 25% chance of reaching the shortlist.
- Then a 1-in-3 chance from there.
- Net 1-in-12 — identical to a direct draw, with a completely different acceptance rate.
The veto round, done explicitly
The other approach that works at two people is to make the veto formal and finite. Each person may veto one option from the shortlist, and whatever survives is where you go.
Making it explicit does two things. It caps the vetoing, which otherwise continues indefinitely; and it converts an implicit power that both people had into a stated rule, which removes the sense that one person is being difficult.
With a shortlist of three and one veto each, at least one option always survives — and if both veto the same option, two survive and the second draw decides. That is a complete rule with no edge cases, which is more than most household decision procedures manage.
- Draw a shortlist of three.
- Each person may veto exactly one.
- Whatever survives is the answer; if two survive, draw between them.
- No further vetoes. The rule is the point.
Filtering before drawing still matters
The two-person case does not remove the value of narrowing first. If one of you cannot eat something, or you have forty minutes, or it is a Tuesday and neither of you wants to cook, those constraints should remove options before the draw rather than after it.
The dinner decider applies exactly that shape — filters first, draw second, with the surviving count shown — and at two people the count is often small enough that the shortlist is most of what survived, which is fine.
The failure to avoid is filtering after the answer arrives. "Not that, it takes too long" is a constraint that should have been a setting, and applying it to the result is how a draw turns into a negotiation.
Frequently asked questions
Why does a single random answer not work for two people?
Because a veto costs almost nothing when there is only one other person to override. The draw is not failing; the social conditions that make it binding are absent.
Does a shortlist change the odds?
No. A three-of-n shortlist followed by a one-of-three pick gives every option exactly the chance it had in a direct draw.
Then what does it change?
Acceptance. Seeing the two options you did not take makes the one you did feel chosen rather than imposed.
How does a veto round work?
Draw three, each person vetoes one, and whatever survives is the answer. If both veto the same option, draw between the remaining two.
Why cap the vetoes?
Because uncapped vetoing continues indefinitely and converts the draw back into a negotiation, with the added cost of having pretended otherwise.
Should I still filter before drawing?
Yes. Constraints that would cause a veto should remove options before the draw, not after. Applying them to the result is how a draw becomes a negotiation.
How many options should be in the pool?
Enough that the shortlist is a genuine narrowing. Below about six, drawing the answer directly is simpler and just as good.
Does this apply to things other than dinner?
Yes — it is a property of two-person decisions rather than of food. Films, weekends and anything else with a free veto behave the same way.
Tools for this job
Background reading
Why Humans Are Bad at Being Random
Asked to be random, people avoid repeats, alternate too often, and dodge the edges.
How Weighted Random Selection Works
Cumulative sums, one float, and why duplicating entries is the wrong way to do it.
Randomness in the Classroom: A Primer
Where random selection helps a lesson, where it hurts one, and what teachers report.