Limits of the desk
Working from a screen
So far I have built Situ from open data, files that public bodies or volunteers publish for anyone to use, such as the city’s tax roll or OpenStreetMap. I have looked at Griffintown mostly through other people’s photos and videos. This desk research, everything I can do from a screen, has not yet been checked by walking the area with a notebook, and that shapes what every system can claim.
Where the files fall short
The first limit is purpose. No file describes a city in general. Each was made by an institution to do a job, and its categories serve that job. Scholars of data make the same point: data are gathered and shaped for a purpose before anyone analyses them (Gitelman 2013; Kitchin 2014). When I borrow a file for a different question, its categories come with it.
The city’s street file is a road file. It left out the canal path and the new park paths, and 10 of the first 12 dead ends in S1 turned out to be places where walkers carry on. The tax roll is a file for taxing property. It counts a condo parking space as a property, as it counts a home, and gives a tower’s storeys only for some kinds of property.
The second is time. Every file has its own date, and Griffintown changes faster than most of them. The city’s sign post at one corner under the Wellington viaduct is from October 2025, while every photo of it is older (P03). The city’s 3D model of 2020 stops at the Ville-Marie borough line, so Griffintown, in Le Sud-Ouest, has only the model of 2016, older than many of its towers. When two files disagree, the cause is often a gap in time.
The third is coverage. The speed file has no limits for Autoroute 10 and its ramps. The traffic light file says where the lights are and nothing about their timing. No open file says where the doors of buildings are. OpenStreetMap is drawn by volunteers, so a building appears only once someone draws it.
The fourth is position. A point in a file stands in for a thing on the street, and the two are rarely in the same spot: an address point often stands at the curb, a shop’s point on its façade. At 1485 rue des Bassins, 290 dwellings share one address point 130 m from the two towers on their lot (B01). Even my own arithmetic can move things: an early version of S4 placed some building centres tens of metres off, because of the way it handled large map coordinates.
Seeing from the desk
Photos and videos let me look at the street without being there, and they settled real questions, among them most of the no-entry disagreements in S3. Mapillary photos from 2016 and 2024 showed that walkers cross Place des Aiguilleurs at the end of Rue Smith (C06). A walking video filmed on 11 July 2026 shows the signs at the entrance of Allée des Vraquiers: a pedestrian zone, closed to bicycles and to every vehicle except emergency ones.
Still, a camera goes where its owner walks, and it records one moment. Neither of the two recent walking videos reaches the end of Rue Brennan, so the footway I assumed there (C04) is still unconfirmed. A sign seen edge-on hides which street it faces, and a photo older than the sign it shows can mislead.
So every decision keeps its source and its date, and I grade the evidence. A link I assume is drawn in pink. A photo or a video can make it likely, and only a visit makes it confirmed. Videos are cited by link and timestamp, and no frame appears on this site unless its owner agrees.
System by system
| System | What open data gives | What the desk cannot settle |
|---|---|---|
| S1 Streets | street lines with their class and direction, plus paths from OpenStreetMap | whether a path is open on the ground, and where a closure or building site cuts a route; three links at the end of Brennan and Ann rest on assumption |
| S2 Scores | space syntax values, which measure how each street sits in the network I built | where people actually walk; Brennan and Ann gained walking choice only because I assumed they lead somewhere |
| S3 Rules | where each sign and light stands, and the speed limits | which way a sign faces, and how long each light stays green; 21 places remain to check |
| S4 Buildings | uses and dwellings from the tax roll with its lots, and outlines from OpenStreetMap | how tall many buildings are, and which of two towers holds which units; 61 cases remain to check |
| S5 Rhythm | counts at 7 junctions on 9 weekdays, three bicycle detectors, and opening hours for 83 of 558 places | the weekend on foot and by car, and whether an open place is busy; 17 of the 24 weekday hours rest on one junction |
| S6 Streetscape | public trees, hydrants, sidewalks and road works, with lights, cameras and crossings from photos and from mappers | trees on private land, and which lights and cameras still stand; 40 places remain to check |
| S7 Commuters | homes, work from home, ways to work and hours of leaving and arriving, from the census of May 2021, and the stops of the STM | where each commuter goes and by which street, and how much the pandemic still bends the census; nothing open says how many people visit |
| S8 Views | what can be seen from each walkable street past the building outlines of OpenStreetMap | trees, walls, viaducts and heights; buildings missing from the map open views that do not exist |
| S9 Observer | a weekday morning of commuters on foot, imagined under three route rules, set against the counts and space syntax | everyone who is not commuting, which is most of the pedestrians counted; where walkers really work; which reading is closest, with seven junctions to compare |
People are missing
The deepest limit is about people. Open data records a city’s infrastructure. Apart from the city’s movement counts at a few intersections (Movement counts) and its bicycle detectors, it says almost nothing about how people use the streets: where they linger, or when a corner fills. Situ aims to mediate and archive Griffintown’s spatial culture, which Hillier described as a distinctive way of ordering space (About), and its urban rhythm. Both live in what people do. From the desk I can describe the stage. I cannot yet describe the play.
Carrying the gaps
A value no source gives stays empty, with the reason written beside it, never a zero or a guess. Every assumption is labelled as one on the page and in the Lab, and every place the desk could not settle becomes a checkpoint with a question.
So Situ describes Griffintown as its files and other people’s cameras record it, at dates I can name. It shows how the layout and the rules shape the movement that is possible; how people actually live in the area is still beyond it. Each system says which of the two it is doing.
Most remaining checkpoints can only be closed on site. The next step is a walk through Griffintown with these questions in hand, photographing and dating each answer. Later, observation of people in place will add to S5 Rhythm, the first system about time, which so far knows only weekdays.
References
- Gitelman, L. (ed.) (2013). “Raw Data” Is an Oxymoron. MIT Press.
- Kitchin, R. (2014). The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences. SAGE.