Tool
Random Pair Generator
Split a list into pairs — with an honest answer for what happens on an odd count.
Random Pair Generator
Split a list into pairs, with an explicit decision about what happens when the number is odd — a group of three, or one person named as sitting out.
The odd case is the whole reason this is a separate tool. Every pairing problem hits it, and the standard failure is a tool that silently drops the last person off the end, which is precisely the person who most needs to know.
Whichever option you choose, the tool states the outcome. Nobody disappears from the result.
How the Random Pair Generator works — and why it's fair
The list is shuffled with Fisher-Yates and then read off two at a time. Because the shuffle is uniform, every possible pairing arrangement is equally likely — pairing the shuffled list positionally is the standard way to sample a random perfect matching, and it needs no additional randomness beyond the shuffle itself.
The odd handling happens before or after the pairing depending on which rule you choose. Selecting a bye pops the last name off the shuffled list first, so the bye is a genuinely random person rather than whoever happened to be typed last. Selecting a trio pairs everyone until three remain and then groups those three, so the trio is likewise a random set.
That ordering matters and is easy to get wrong. Removing the bye from the unshuffled list would give the bye to whoever was pasted last, every single time — a bug that is invisible in the output and would only be noticed by the person who kept sitting out.
Nothing here tracks history. Two consecutive runs over the same list can produce the same pairs, and over a term the same two people will be paired repeatedly by chance. If avoiding repeat partners matters, the lunch partner picker takes a list of previous pairings and works around them.
The number of possible pairings grows fast and is worth knowing when someone claims a pairing was fixed. Eight people have 105 distinct ways to be split into four pairs; ten people have 945; twelve have 10,395. Getting the same pairing twice out of ten people is a one-in-945 coincidence, which is rare but not remarkable across a term.
When the Random Pair Generator is fair — and when it is not
What it does guarantee
- Every possible pairing arrangement is exactly as likely as every other.
- The bye or the trio is drawn at random, not assigned to whoever appears last in your list.
- Nobody is dropped from the output — the odd person out is named explicitly.
What it does not
- It has no memory. Consecutive runs can produce identical pairs, and often will over a short list.
- It cannot keep two named people apart, or together. Constraints are the lunch partner picker's job.
- A trio is a bigger group, not a fairer one — three people share work differently from two, whatever the draw says.
Two worked examples
8 people, even
- Four pairs, drawn from 105 possible distinct pairings.
- Every one of those 105 arrangements is equally likely.
- The same arrangement twice in a row has a probability of about 1%.
9 people, trio rule
- Three pairs plus one group of three.
- The three in the trio are the last three of a uniform shuffle, so any three people are equally likely to form it.
- Switching to the bye rule instead names one random person as sitting out and makes four pairs.
Where pairing needs saying out loud
Classroom lab and project pairing is the most common use, and the odd-number rule is why: a class of twenty-nine is the normal case, not the exception, and how the twenty-ninth student is handled is a decision the teacher should make rather than discover.
Sparring, practice drills and language exchange all need genuine pairs and all hit the odd case constantly. A named bye who is explicitly rotating out is a completely different social experience from a person who silently was not in the output.
Peer review and buddy systems use it where the pairing should visibly not have been chosen by the organiser. The uniform shuffle is what makes that claim true, and it is worth stating to the group.
For pairing that avoids previous partners or crosses team boundaries, the lunch partner picker applies those constraints. For splitting into groups larger than two, the team generator handles arbitrary group sizes, and the balanced team splitter matches groups on strength rather than randomising them.
Frequently asked questions
Who gets the bye with an odd number?
A random person. The list is shuffled first and the bye is taken from the shuffled order, so it is not whoever you typed last.
Trio or bye — which should I choose?
A trio keeps everyone participating and changes the dynamic for those three. A bye keeps every pair a true pair and leaves one person out. Both are legitimate; pick before you generate.
Does it avoid repeating last week's pairs?
No, it has no memory at all. For history-aware pairing, the lunch partner picker takes previous pairs as a constraint.
Is every pairing arrangement equally likely?
Yes. Positional pairing of a uniformly shuffled list samples uniformly from all possible matchings.
How many possible pairings are there?
For eight people, 105. For ten, 945. For twelve, 10,395. The count grows very quickly, which is why repeats are rare in larger groups.
Can I force two people to be paired?
No. Pair them by hand and run the tool on everyone else — that is simpler than any rule the tool could offer.
Can I make groups of four instead of two?
Not here. The team generator splits a list into groups of any size, and the group role assigner handles roles within them.
Does the order of the pairs mean anything?
No. It reflects the shuffle only, and carries no ranking or priority.
What if I only enter two names?
You get one pair, which is a correct if unhelpful answer. The tool needs at least two names to do anything.