Appendix
Measuring Engagement, 2014: model of the office
Part of Measuring Engagement, a 2014 study by the Iota team at SapientNitro. How the walk is drawn gives the method.
This appendix. It records the 3D model of the office floor where the walk starts and ends, built in 2026 from the agency’s floor plan and TJ’s corrections, and how the Glass video was placed in it. The placement gives where the walker stood, to a foot or two where it can be checked, and the direction their head faced.
Views
Layers
The model: drag to turn, scroll to zoom. It is a measuring tool and not a portrait, so it has no textures or people.
What it is built from
- The floor plan. The agency’s own plan of the floor, its lines fitted to the team’s 2014 floor outline to a median of 4.5 in. It is drawn on the model’s floor, and TJ, who worked there, calls it an accurate record. The small offices, the long block, the corner office, the rooms to the north and the benches are raised from it.
- What TJ remembered and corrected. The wall between the office and the elevator lobby is glass. Foamcore panels made a corridor, in the same plane as the corner office’s north wall. There is a walkway between the electronics cabinets and the wood tables in front of them, and their ends line up. A long white table stands just north of the corner office.
- TJ’s photographs, which are not shown here. Four cameras are solved from points the plan fixes. They give the mural, the teal tile, the kitchen’s counter and cabinets, and the tables, most to about a foot. The corridor photograph’s two nearest door jambs land 5 in from the plan’s.
- The video’s frames. The big screen comes from its four corners in four frames, about 52 by 30 in. The long white table comes from one sharp frame: about 5 by 12 ft, running north to south, good to about a foot.
How the video is placed in it
The way out takes the method in how the walk is drawn, a stretch at a time. Between tied frames the head’s direction comes from points tracked from one frame to the next.
The way back was harder, and it was redone. Its rebuild, at ten frames a second, bends between parts of the floor, so each part takes its own fit. The desks and the corner office take theirs through an earlier rebuild that holds frames of both the way out and the way back, so they are tied by the way out’s edges alone.
How well it fits
| Check | Result |
|---|---|
| The way out, reads against the model, by stretch | Median 2 to 8 pixels in a 1280-pixel frame, depending on the stretch. The worst single read is 102 pixels, a desk edge low in the frame. |
| The way out, where two separately tied stretches meet | 12 to 36 in apart at the same moment, and in two places perhaps 20 and 40 in further than walking explains. This is the best sense of how good the place is. |
| The way out, the first frame against TJ’s photograph from the desk | 11 in apart, though TJ need not have stood in the same spot. |
| The way back, fit one visit to the corner office’s side, judge the other | 1 to 1.6 ft, and 1.5° of head direction. |
| The way back, eye height, which no fit is told (TJ’s is 69.25 in) | 65 to 76 in there; 56 to 72 in along the desks. |
| The way back, against walls and furniture, which the fit is not told | Keeps at least 18 in east of the long table, and crosses the corner office’s wall within 4 in of the door’s jamb, going in and coming out. |
| The head, with 8 s of known frames hidden | Way out: median 3°, with one trial in twenty worse than 11°. Way back: median 7°, with one in twenty worse than 30°. A fast turn the known frames miss can be lost. |
All of the way back’s checks rest on the same edges from the way out, so they test the solving and not those edges.
What went wrong along the way
- Agreement proved little. One version of the way back was tied to hand reads from the way back itself. They agreed with each other to 0.3 ft, yet put the walker on the long table and through the corner office’s wall. TJ pointed at the table. Those ties are no longer used there.
- The table was drawn the wrong way. It ran east to west in the model, from a thin rectangle on the plan, until a sharp frame measured it.
- The corridor’s rebuild tilts about 8°. That ran its height to over 10 ft before it was levelled. Its size then comes from one wall end, which three frames read alike.
- Reads by eye can be wrong. Three head directions read by eye were up to 18° from the later rebuild, and where they differ the shelves in the frame side with the rebuild.
Estimated, not measured
- The corridor’s far end is a few feet uncertain.
- The kitchen and the corridor’s way back are tied to their own sharp frames, so nothing checks them across visits.
- Straight lines. About a ninth of the way back, 15.6 s of 139, is a straight line between known places, and from 1:54 to 2:00 and 2:15 to 2:21 the way out is too.
- Seven head directions on the way back are read by eye, good to 5 to 8°.
- Leaving the corner office, the walker crosses the foamcore’s line about 2 ft inside its east end, which a wall would not allow.
- Heights that were not measured: the open area’s ceiling at 108 in and the desks at 29 in. The corner office’s furniture and the sofas are placed roughly from frames, and small cabinets on the wood tables are from TJ’s memory.
- The north of the floor is the plan and no more: the frames see it only from the corridor’s end.
The walk and its instrumentation were a team effort by Iota, at SapientNitro. This model was made in 2026, from the agency’s floor plan, TJ’s memory and photographs, and the Glass video. The photographs are not reproduced here.