Ceremony as a smell
A Socratic walk-through of ceremony as a smell — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #When does a practice adopted to help a team become the thing slowing it down?
Nobody introduces a bad process. Every standing meeting, sign-off form, and status template began as an honest answer to a real problem: people were surprised by each other's work, a change went out unreviewed, a customer heard the wrong thing twice. The practice was adopted because it solved that.
Yet the complaint is universal, and rarely "we have no process". It is that the process has become an end in itself. So the question is not whether ceremony is good or bad, but this: what changes between the day a practice helps and the day it hinders, when the practice itself has not changed?
Reasoning it through
REASONING #Start with the structure of any practice. There is a problem, a response to it, and a cost of running the response. On the day it is adopted, the response matches the problem and the cost is clearly worth paying. Notice that only one of those three is written down. The response gets a name, a calendar slot, an owner, a checklist. The problem it answers is discussed once, in a room, and then lives only in the memory of whoever was there.
What follows from that asymmetry? Ask what happens as people leave and join. The newcomer inherits the response fully specified and the problem not at all. They can execute the ritual perfectly and cannot evaluate it, because evaluating it requires knowing what it was for. Within two or three turnovers, everyone present is in that position — not through carelessness, but because the justification was never stored anywhere.
Now a second drift. Problems are not permanent. The standup was invented because the team shared one codebase and stepped on each other daily; three years later they own separate services and deploy independently. The daily conflict has stopped occurring. Would anyone notice? Here is the trap: a practice that prevents a problem produces the absence of that problem, and absence looks identical whether the practice is doing the work or the problem has simply gone. The successful preventive measure and the obsolete one give exactly the same evidence. You cannot tell them apart by observing outcomes, only by re-examining the cause.
Put the two together and you have the mechanism: the response is durable and specified, the problem is undocumented and mutable, and the practice's own success makes its necessity unfalsifiable.
A third pressure makes ceremony hard to remove even when someone sees it. Practices acquire constituencies — someone owns the meeting, maintains the template, reports the metric. The cost is spread thinly across everyone; the benefit of keeping it is concentrated in a few. That is the classic shape of a thing that outlives its usefulness, and it is incentives, not bad faith.
So what is the actual test? Not "does it feel tedious" — plenty of valuable work is tedious. Try three. First: can anyone state the problem it prevents, in the present tense, without reciting the practice's own name? "We do retrospectives so that we have retrospectives" is the failure mode; "we do them because the last three incidents shared a cause nobody raised" is not. Second: what would happen if we stopped for a month? If the honest answer is "nothing we could detect", you have your answer; if it is genuinely unknown, that is what the experiment is for. Third: has the output been read? A status report nobody opens, a form filed to a folder — that is measurable, and it is the clearest evidence available.
An honesty note: only the third test is objective, and even it has a false positive, since some documents exist to be available rather than read. These are prompts for judgement, not a decision procedure. I know of no validated measure separating useful process from ceremony; the literature on ritualisation describes the drift well but hands you no threshold.
The analogy
THE ANALOGY #Think of a scaffold on a building. While the wall is going up it is indispensable and nobody argues about the inconvenience. Once the wall is finished the scaffold is not harmful — it just stands there, gets repainted with the building, gets inspected annually, and slowly becomes part of what people think the building looks like. New residents assume it is structural. Removing it starts to feel dangerous, precisely because nobody remembers the wall being built without it.
a scaffold is visibly separate from the building and could be checked against the drawings, whereas a process leaves no trace of what it was holding up — the drawings for a team practice are the reasons it was adopted, and those are the one thing nobody wrote down.
Clarifying the model
THE MODEL #Three refinements, one of which is a warning.
First, ceremony is not the same as overhead. Some necessary work is dull and produces nothing visible — backups, access reviews, incident write-ups. What marks ceremony is not the boredom or the invisible output but the missing live justification. Confusing the two is how a team removes its safety mechanisms in the name of removing ceremony, which is the most expensive form of this mistake.
Second, the reverse error is real and commoner in young organisations: mistaking a problem you have not had yet for a problem that does not exist. A team that has never had an outage will find change review pure ceremony, right up until it is not. That is why the honest form of the second test is not "nothing bad happened last month" but "we understand the failure this prevents and judge it unlikely or cheap".
Third, the useful move is almost never deletion by argument. It is a time-boxed suspension with a stated tripwire — stop for six weeks, name in advance the signal that would mean it was load-bearing, and let the evidence decide. That converts an unfalsifiable claim into a testable one, which is exactly what the practice's own success took away, and it has a political virtue: nobody has to lose the argument.
The deeper point underneath all of it: the durable fix is not pruning ceremony afterwards but recording, at adoption, the problem a practice answers and a date to re-ask. A practice with an expiry attached can be evaluated by whoever inherits it.
A picture of it
THE PICTURE #How to readLeft to right is elapsed time, not a schedule — the spans are often years. The first two stages are the practice working exactly as intended; nothing has gone wrong yet. The turn comes at the third stage, where the problem changes but the response does not, and the crucial detail is its second line: a practice that has succeeded and one that is obsolete produce identical evidence. By the fifth stage the practice has a constituency, which is why argument alone rarely removes it. The final stage is not an inevitability but the choice available at any point after the third.
What became clearer
WHAT CLEARED #A practice does not rot because anyone stops caring. It rots because the response is recorded and the reason is not, because problems move while practices stay put, and because prevention makes its own necessity unfalsifiable — the absence of the problem is equally consistent with the practice being essential and with it being redundant. Hence the useful question is never "is this meeting worth it", which invites opinion, but "what live problem does this answer, and what would we see if it stopped", which invites evidence.
Where to go next
ONWARD #- Whether recording the problem and a review date at adoption actually prevents the drift, or just adds one more artefact to ignore.
- How the same mechanism produces regulatory and compliance accretion, where suspension is not an available experiment.
Key terms
TERMS #| Term | What it means |
|---|---|
| Ceremony | a practice still performed after the problem it answered has changed or gone, recognisable by the absence of a stated live justification. |
| Ritualisation | the process by which an instrumental action becomes valued for its performance rather than its effect. |
| Time-boxed suspension | deliberately stopping a practice for a fixed period, with a pre-agreed signal that would justify reinstating it. |
Every term the collection defines is gathered in the glossary.