Ottawa fixed the permit for a party nobody can count

Ottawa city council voted on 24 June to replace its per-event permit for all-night dance events with an annual authorisation attached to a venue. From 2027, a room that meets the safety, insurance and filing conditions holds the permission for a year rather than applying every time it opens late. The category being reformed is defined by the clock: the city's own licensing pages describe an all-night dance event as one offering live or recorded music between 3am and 9am, and note that these are commonly called raves. We index event listings in eight other cities. Across 11,935 of them, 102 publish a start time inside those hours.

What the vote changed

The change arrived through a wider licensing by-law review and re-enactment that went to council on 24 June 2026. It came out of the city's nightlife council, a volunteer body set up in late 2024, and before that the 2023 nightlife economic action plan. Under the old rule an operator applied for a separate licence for every event running past 3am. Under the new one a venue applies once a year and keeps the authorisation as long as it meets the conditions attached to it.

What is permitted has not moved at all. The safety, insurance and filing requirements survive intact. The reform is entirely about how often somebody fills in a form, which sounds minor and is the thing operators actually complained about. CBC reported one Ottawa promoter who ran all-night events for around two decades and stopped after the pandemic, citing insurance costs alongside the per-event process, which is their reporting rather than ours.

The regulated hours barely appear in the public record

Every listing in the set carries a start time, so this is not a gap in the data. It is what the data says. One percent of everything scheduled across eight cities over 46 days begins in the small hours, and under one percent begins inside the window an Ottawa licence is written around.

Berlin makes the point on its own. Of its 1,916 listings, 611 count as late night by the usual rule of starting at 22:00 or later, or before 05:00. Of those 611, some 610 start at 22:00, 23:00 or midnight. Exactly one Berlin listing in six and a half weeks starts between 1am and 5am, in the city whose entire international reputation rests on the hours after that.

A listing carries a start and nothing else

The explanation is dull and it governs everything downstream. Our dataset records a start time for each event and no end time at all, because the sources feeding it publish one and not the other. A Berlin club with a midnight door and a Sunday afternoon finish files a listing that says 00:00. An event occupying the whole of 3am to 9am reaches the record as an event that began at 23:00 the night before.

So a histogram of start hours is a picture of when doors open. Ottawa's by-law regulates occupancy, meaning the hours an event is actually running. Those two measurements diverge across precisely the range the licence covers, which is why counting one tells you nothing about the other.

One caution on our own figures, since it cuts the same way. Midnight starts are the largest single bucket in several cities, at 237 in Berlin and 64 in London, and some share of those are listings that published a date without a time and landed at 00:00 by default. The public record of when cities dance is thinner than even this table suggests.

Nobody can score the reform from listings

The obvious question a year from now is whether Ottawa ended up with more all-night events. That question cannot be answered from event listings, in Ottawa or anywhere, because the field that would answer it does not exist. Anyone claiming a rise or a fall will be arguing from memory or from a handful of examples.

The city, though, will be sitting on the exact number. An annual authorisation regime generates a register: how many venues hold one, in which wards, renewed or lapsed. Publishing that register turns a policy change into something measurable at no cost, and its absence is the difference between a reform that can be evaluated and one that can only be defended.

The permission moved from the promoter to the room

A per-event licence is applied for by whoever is putting the event on. An annual authorisation attached to a venue is held by whoever holds the lease. That is a transfer, and it has not been discussed as one.

For a room running late events most weekends, the saving is large and immediate. For a promoter without a venue, the position is unchanged in law and worse in practice, because the search is now for a room that has already gone and obtained the authorisation. Nothing in the design lets a promoter get one independently.

Ottawa is not one of the cities we index, so this is a reading of how the mechanism works rather than a measurement of what will happen there. What our data does show is the shape of the rooms this kind of rule lands on. Across eight cities we counted 3,610 venues programming 11,935 events, which works out at around three events per room across six weeks. Most of a city's calendar sits in that long tail, in bars and galleries and restaurants that host a handful of things a month. An annual licence is efficient for a venue that opens late every week and close to pointless for one that does it twice a year, and the second kind is far more numerous than the first.

Three things a city could publish

• The authorisation register. Counts by venue and by year, with renewals and lapses. This is the only reliable measure of whether the reform worked, and the city generates it as a by-product of running the scheme.
• An end time on the filing. If an operator must declare that an event runs into the regulated hours, that declaration is a scheduled end time. Requiring it on the form, and releasing it in aggregate, would create the first honest picture of how late a city actually runs.
• The definition, in the words promoters use. A rule written around 3am to 9am is legible to a licensing officer and invisible to every listings system a promoter touches. Cities keep regulating a category that their own residents cannot look up.

Our counts here come from the published market dataset, captured on 17 August 2026 across a 46-day forward window, and its coverage limit applies as always: it measures the listings our sources carry for a city rather than everything happening in it. On this particular question that limit is the finding. The hours nobody publishes are the hours nobody can govern by evidence.