Mount Vernon Trail · Arlington & Alexandria, VA
The dumping slowed. The devices still did not leave.
On the third pass through this study, I saw far fewer newly abandoned scooters and e-bikes. But the corridor was no emptier. A smaller group had become a more persistent one: seven of 17 reconstructed devices stayed at least a week, and three stood in the northern corridor for the full window.
- reconstructed devices, at minimum
- 17
- new devices per pass
- 0.6
- stayed a week or more
- 41%
- mean observed stay
- 4.9 days
- device-days of obstruction
- 83
The result
New dumping nearly stopped. The backlog remained.
Across 16 observed passes on nine riding days, 130 GPS-tagged sightings resolve to at least 17 devices. New arrivals fell again, from 3.3 per pass in period 1 to 1.3 in period 2 and 0.6 here.
The standing count stayed almost flat, averaging 8.4 devices per pass in the first half and 8.0 in the second. Mean observed dwell rose from 1.5 days to 2.9 and now 4.9. Seven devices stayed a week or more, yet the total obstruction remained 83 device-days, exactly as high as period 2.
Seven of this period’s 14 locations were occupied in all three surveys. Four held a week-plus device in both periods 2 and 3; the recurring Four Mile Run location at 38.84155, -77.04786 did so in all three.
Rider’s-eye view
What the corridor looked like in period 3
The reel shows the fixed commute corridor as I rode it during this observation window. It provides visual context; the measured findings come from the timestamped sightings and reconstruction described below.
Live map
Fourteen places carried the remaining backlog
Circle area tracks recorded sightings. Color shows the minimum observed stay: red 6+ days, orange 3–5 days, ochre 1–2 days, gray under a day. Basemap: OpenFreeMap, © OpenMapTiles, data © OpenStreetMap contributors.
Circle area follows sightings and color follows minimum observed dwell. The public map uses clustered locations, never the timestamped commute log.
How I counted
The same method, with two transparent adjustments.
I re-recorded each obstruction on every observed pass. Sightings within 40 metres are grouped into a location, and an unbroken run across the ordered passes is treated as one occupancy episode. A one-pass internal gap is bridged as a likely missed observation; longer gaps are treated as departures.
Two trips home, September 22 and 24, were ridden but not observed. They are absent from the sequence rather than treated as empty passes, so they cannot make devices appear to vanish.
On September 23 I passed the bridge-deck device without logging it and confirmed it was still present. The pass is marked occupied for reconstruction, but that confirmation is not counted as a sighting. It changes no headline result and only lowers the strict upper bound from 27 to 26.
What the count can establish
Seventeen is a floor. On nine passes, one location produced multiple simultaneous sightings, implying at least nine more devices the location-based reconstruction cannot separate. Devices that arrived and left between passes are invisible.
A recurring location demonstrates a recurring place, not one continuous device. The 17-day gap before this survey was unobserved, and only operator telemetry can identify individual units across time.
Requested action
Explain the collection gap.
- Clear the recurring locations and account for why seven devices remained a week or more.
- Compare operator telemetry with all three windows, including the apparent Four Mile Run clearance and the three persistent northern devices.
- Geofence the trail and bridge crossing, require end-of-ride parking verification, and add the corridor to a daily sweep.
- Make removal standards enforceable and measure dwell time rather than complaint volume across jurisdictions.
The series continues
The effect of intervention can be measured.
The complete timestamped datasets are available on request to an operator or agency that wants to compare them with internal parking and collection records. They are not posted publicly because they reveal a repeat commute pattern.
Contact Jordan about the studyData notes & limitations
Read these alongside the findings.
- Sampling: I made 16 observed passes over nine riding days on a fixed route. Devices outside my line of sight, or placed and removed between passes, are absent, so every count is biased downward.
- Skipped trips: I rode home on September 22 and 24 without observing. Those rides are excluded rather than treated as empty passes, so they cannot make every device appear to leave.
- Confirmed presence: on the evening of September 23 I passed the bridge-deck device without logging it and confirmed it was still there. That presence helps reconstruct the episode but is not counted as a sighting and changes no headline figure.
- Device count: 17 is reconstructed from 130 sightings, not directly counted. Splitting every gap gives 26, while nine passes with co-located sightings imply at least nine additional devices the location-based method cannot separate.
- Positional accuracy: phone GPS fixes were taken while moving. The 40-metre clustering radius may merge nearby parking spots, especially around the bridge.
- Episode identity: a sequence at one coordinate cannot distinguish one stationary device from successive devices left in the same place. Only operator telemetry can resolve that ambiguity.
- Cross-period matching: locations are matched within 60 metres. A match means the place recurred, not the device; the 17 days between periods 2 and 3 were unobserved.
- Departures: a device that stopped being recorded left by means this study cannot identify. The data cannot distinguish collection, a rider trip, or movement by someone else.
- Dwell times are lower bounds. A device may have arrived before its first sighting or remained after its last.
- Operator attribution: brand is my contemporaneous visual observation and is not encoded in the public location data. The map and table make no per-device operator claim.