A Citi Bike trip isn't over until you can dock. I started with a simple question: when the nearest station fills up, where else could you return the bike? The web experiment compares captured dock history. It's the starting point for DockMe, an iOS app in development.
Finding a bike is only half the trip. Arriving at a full station means another detour when you thought you were done.
Pick a day and time in the example. Compare no open docks with almost full, then see what nearby stations offered at those same moments. A better overall average does not tell you whether a backup had room when you needed it.
These are captured observations, not live availability or an arrival-time forecast. Weather, events, and rebalancing can change the pattern. The native app is a separate work in progress; the web link opens the research that came first.
iOS app · in development
Choose a dock, keep a backup within reach, and confirm arrival. Here’s the native flow in a short demo.
UI prototype under testing; live routing and physical riding validation pending.
Select DUMBO, compare three dock choices, choose an alternative, start the demo ride, swap to the backup, and confirm arrival manually. The recording ends at the finish controls. The completed summary is shown in a separate screenshot.
The ride screen keeps the selected dock, a swap control, and manual arrival together.
Actual app capture · synthetic data · tap to enlarge
The verified summary after finishing. The recording ends at the arrival and finish controls.
Actual app capture · synthetic data · tap to enlargeAllen & Stanton · Lower East Side
Friday, 6–9 pm stood out. The destination had no open docks in 53 of 90 checks — the highest share of the 56 three-hour windows in this sample.
Select a time · New York (EDT) · darker = more no open docks
90 matched checks across 3 dates
Destination · Allen St & Stanton St
Returns paused: 3 of 90 checks. Separate status; may overlap the dock counts above.
During Friday · 6–9 pm, Allen & Stanton was unavailable for returns in 53 of 90 checks. Unavailable means zero open docks or paused returns. The backup percentages below use only those 53 checks.
219 m straight-line from destination
When the destination was unavailable for returns:
70%of those checks had space here
37 of 53 destination-unavailable checks: at least one open dock and accepting returns.
8% had 3+ open docks and accepted returns (4 of the same 53 checks).
Across all 90 checks in this window:
Returns paused: 0 of 90 checks
260 m straight-line from destination
When the destination was unavailable for returns:
36%of those checks had space here
19 of 53 destination-unavailable checks: at least one open dock and accepting returns.
8% had 3+ open docks and accepted returns (4 of the same 53 checks).
Across all 90 checks in this window:
Returns paused: 0 of 90 checks
Check Allen St & Rivington St live first. It offered a return in 70% of checks when the destination was unavailable. It had three or more docks in 8% of those same checks. Check current availability before riding. All three were unable to take a return in 13 of 90 total checks. Keep a wider backup in mind.
September 12–October 3, 2026 · America/New_York (EDT)
5,035 aligned checks per station. Historical frequency, not a prediction for your arrival. Check current availability before riding.
First capture: Sep 12, 5:46 am. Last: Oct 3, 5:37 am. Both boundary days are partial. Report generated Oct 3, 2026.
Selected window: Sep 18 (30 checks; 8 with zero docks at the destination); Sep 25 (30 checks; 27 with zero docks at the destination); Oct 2 (30 checks; 18 with zero docks at the destination).
Each weekday/hour has only about three weeks of evidence. Three-hour windows contain 89–90 checks; successive checks can describe the same full-station episode and are not independent trips. The largest capture gap was about eight minutes. Unrecorded changes between checks are unknown.
Sampling was configured for five minutes; observed spacing was about six. Coverage is limited to these three stations and this season. Weather, events and rebalancing are not controlled for.
0, 1–2 and 3+ are exclusive physical dock-count categories. “Almost full” includes 0–2. Paused returns is a separate status that can overlap any category. A backup offers a return only when it has at least one dock and accepts returns; three or more indicates additional room, not a guarantee.
Backup percentages are rounded to whole percentages and use only the destination-unavailable checks, not every check in the time window. Exact counts are shown beside each result.
The older dashboard combines 0–2 docks OR paused returns in one warning. This view separates them and compares stations at identical capture times. Distances use station coordinates; walking routes and extra walking time have not been verified.
Light cells are lower historical frequency, not a safety rating. Windows below 60 matched checks or three dates are marked as limited coverage; this is a display guard, not statistical confidence.
Public history source ↗