Building a Programme Committee That Actually Reviews

A programme committee determines what the conference is. It is usually assembled from whoever the organiser knows, given no brief, and asked to score submissions on a scale nobody defined.

The failures are predictable and preventable. For a concrete operational example of how a workplace tool approaches a related coordination problem, see learn more.

Assemble it around the load

Work out the arithmetic first. Expected submissions, reviews per submission, reviews per person.

Nobody should review more than about a dozen abstracts. Beyond that, quality falls and lateness rises, reliably.

Two reviewers per submission is the usual minimum, three where decisions will be contested.

So: 200 submissions at two reviews each is 400 reviews, which is 35 people at a dozen apiece. Assembling eight people and hoping is how committees produce fifty reviews written in one evening.

Over-recruit. Some will drop out and some will not respond.

Who is on it

Beyond your own network. This is the decision that determines whether the conference reproduces itself. A committee drawn from one department, one country or one methodological tradition selects from that tradition, and the programme then attracts the same people next year.

Mixed by seniority. Early-career researchers frequently write the most careful reviews, and being on a programme committee is genuinely valuable to them. Invite them and say why.

Mixed by subfield, matching the range you want to attract rather than the range you already have.

People who will actually do it. A distinguished name who does not review is worse than an unknown one who does — it consumes a slot and produces nothing.

Ask directly. "This is roughly twelve abstracts, over two weeks in March. Can you commit to that?" A clear ask gets a reliable answer.

The brief

Committees are usually given a deadline and nothing else. Give them the rest.

The criteria, explicitly. What are you selecting for — novelty, rigour, fit with the theme, likely quality of the talk, range across the programme? These are different and they trade off.

"Interesting" is not a criterion. It reproduces whatever the reviewer already knows about, which systematically disadvantages work outside their subfield.

What accept and reject mean here. A 30% acceptance rate and a 70% one require different judgements, and a reviewer who does not know the rate calibrates arbitrarily.

Whether to score, and what the scale means. Define each point. Without definitions, one reviewer's 3 is another's 5 and the aggregate is noise.

How long a review should take and what it should contain. Two sentences of justification per abstract is a reasonable ask and it makes the committee meeting possible.

Conflict of interest rules. Same institution, recent co-author, own student, own submission. Say it explicitly and provide a mechanism to declare it.

Running the process

Send the brief with the assignment, not before.

Give a real deadline and one reminder. Then chase individually — a general reminder to the whole committee is ignored by exactly the people who need it.

Have a plan for non-response. Some proportion will not review, and reassigning at the last minute needs a reserve list.

Watch for the abstract with two contradictory reviews. That is where a third reviewer or a chair's decision is needed, and it is where most of the real decisions are.

Do not average scores mechanically. A submission with one enthusiastic and one dismissive review is more interesting than one with two lukewarm ones, and an average hides that.

The selection meeting

Whether to hold one depends on size. Under about 100 submissions, a meeting produces better decisions than a spreadsheet.

Go through the borderline cases, not the clear accepts and clear rejects.

Look at the shape of the programme as well as the individual scores: whether one subfield dominates, whether the accepted set has range, whether early-career submissions are being systematically ranked lower.

Check that last one specifically. It happens, it is usually unintentional, and it is visible if anyone looks at the accepted set by career stage.

Decide the borderline cases against the theme, not against the highest scores. A conference is a programme, not a ranked list.

Communicating decisions

On the date you said. People are booking travel and applying for funding. A late decision costs them money and it is the most common complaint about conference organisation.

Send acceptances and rejections at the same time. Staggering them means the rejected work it out from other people's news.

Give reviewer comments where you can. Even two sentences. For an early-career researcher, an abstract rejection with no explanation is uninformative and discouraging in equal measure.

Say clearly what was accepted as what — full talk, short talk, poster — and what that means practically.

Be honest about the acceptance rate if asked.

See reading a rejection for the other side.

Afterwards

Thank the committee properly, by name, publicly if they are willing.

Tell them what happened — how many submissions, what the acceptance rate was, how the programme came out. Reviewers rarely learn the outcome of work they did.

Record who reviewed well and who did not respond, for whoever organises next time. This handover note is worth more than any other document you leave.

The short version

Do the arithmetic first. Submissions × reviews ÷ twelve is the committee size.

Recruit beyond your own network, or the programme reproduces it.

Give criteria, a defined scale and the acceptance rate, or the scores are noise.

Send decisions on the date you promised — people are booking travel.

Check the accepted set for shape, including by career stage, before you finalise it.

For an independent authoritative reference, consult ACM conference resources for professional guidance on academic events.