Bikeshedding

Some meetings run three hours and end with nothing decided. For a long time the only thing anyone could say afterward was that the meeting had been inefficient. With no name for where the time went, the next meeting went the same way, and the only conclusion anyone ever reached was that there should be fewer meetings.

In 1957 the British historian C. Northcote Parkinson put a name on it. Put a nuclear power plant proposal and a staff bicycle shed proposal on the same agenda, he observed, and the plant gets approved in a few minutes while the roofing material for the shed takes an hour. Two or three people in the room can judge the plant. Anyone can have an opinion about a shed roof.

Pagedeck's weekly meeting produces the same shape. The item on moving pricing data into a new structure passed in ten minutes because nobody could question Theo once he finished explaining it. The next item, naming a new button, took an hour. Mia only realized afterward that the riskiest decision of the day had been the shortest one.

Three conditions have to line up: the number of people qualified to speak varies from item to item, no item has a time budget, and nobody in the room is responsible for judging when enough time has been spent.

1. The difference in barrier to entry starts it. Speaking to a data structure change requires knowing the current structure. Speaking to a button name requires being able to read. With eight people in the room, two can speak to the first item and eight to the second, so the total speaking time splits four to one before anyone opens their mouth.
2. On top of that sits the feeling that saying nothing in a meeting means you had no reason to be there. When an item finally arrives that people can speak to, they spend what they have been saving.
3. Without a time budget there is no point at which this stops. A meeting that hasn't written down how many minutes each item gets treats the moment discussion dries up as the moment a conclusion was reached — and on a light item, discussion never dries up.

What changed once the pattern had a name is that people can point at it. Calling it a problem with how the meeting was designed, rather than a problem with a person, makes interrupting the discussion something other than rude.

The most common use is mid-meeting, to pull the discussion back. Interrupting is hard, and naming the pattern lets you say the time being spent doesn't match the weight of the item without pointing at whoever is speaking.

It also shows up while the agenda is being built. A big item pushed to the end gets rushed through by whatever minutes are left, so items that few people can judge go first and items anyone can judge go last. Each item often gets a timebox — an agreed limit that the discussion is not allowed to exceed. Once fifteen minutes are gone the item moves on undecided and one named person makes the call.

Review sessions attract the same comment. A code review collects twenty comments on variable names and line breaks and none on the design direction. A design review spends its time on corner radius instead of the screen flow. The mechanism is the same one: anyone on the team can form an opinion about a name or a radius, while judging whether the structure will hold up next quarter takes context only a couple of people have.

Checking how reversible the decision is makes it easier to prevent. A two-way door is a decision you can undo the next day if you don't like it, and button names live there. Changing a data structure is the opposite — once it ships, reversing it takes months. Meeting time belongs on the side that's hard to undo.

"This feels like bikeshedding. Let's hand the wording to Noa and move to the next item."

The time being spent is out of proportion to the item, so give the call to one person and move on.

"Last week ended in bikeshedding too. Let's put the hard items first this time."

The same thing has happened twice, so the fix is the order of the agenda.

"Fifteen minutes on the name. If it isn't settled by then, Mia decides."

Setting a limit and a decider in advance, because this discussion is expected to run long.

Pagedeck's pricing revamp meeting was the textbook version. There were two items: how existing documents move when a workspace upgrades from free to paid, and what the paid plans should be called. The first took twenty minutes, since only Theo and Mia knew the details. The second drew opinions from all six people in the room. Pro, Business, Team and Standard all came up, each with an argument behind it.

Forty minutes in, Grace named the pattern. Plan names can be changed later, she pointed out, but a bad document migration mixes up customer data on the way out. The meeting ended with three name candidates handed to Noa, and Pagedeck agendas have carried a line about reversal cost for each item ever since.

The most frequent misreading is that small items shouldn't be discussed at all. The word "bikeshed" sounds dismissive of the bikeshed. But button names and plan names are decisions that reach revenue, and they're the first words a user reads. The problem is the ratio of time spent, not the size of the item. An hour on a plan name is fine in a meeting that gave the migration two hours first. Used as a tool for shutting discussion down, the term keeps stakeholders quiet on decisions that affect their work, and those decisions come back to be reopened later.

Blaming the loudest person in the room is the other common mistake. Whoever used the most minutes is the visible cause, so that's where people look. But the same person said nothing at all during the hard item, which is the clue that the cause isn't personal — they spoke to the item they could judge. Singling someone out just makes that person quiet, and the next meeting someone else fills the same space. What actually reduces it is the agenda order, the time budget, and meeting hygiene rules that fix the purpose and the attendee list in advance.