MGT.
Wimbledon · interactive proof of value

The Microscopic Graph Twin — running on real signal-box data.

Replaying 16,497 discrete relay transitions captured at the Wimbledon workstation across a single morning peak (28 Oct 2019, 07:00–10:00 UTC). Signal aspects, track circuits, points and TRTS plungers stream live onto the topological graph. Use the scrubber to navigate; tap a speed.

Events / 3h peak
16,497
Channels live
217
Track-circuit events
10,322
Signal aspect events
4,435
Sub-threshold δᵢ
2,531
Wimbledon · Live Schematic · Replay
2019-10-28 07:00:00 UTC
Stylised view, panel-mirrored (Waterloo right). Track-circuit IDs from WASC Wimbledon (W) diagram.
PLAT 1PLAT 3PLAT 5PLAT 8XDXFXGXHNSNTXKXLXMXNXPXQJWJXJYBABBBCDJDLDNDODPDR160183168166191193195197176172201369371373375UF →US →← DS← DF→ WATERLOOSURBITON ←
clearoccupied (TPR dropped)no data in windowplatform face
0% · cursor 0/16,497
Decoded relay stream0 of last events shown
TimeChannelAsset
Sub-threshold δᵢ histogram0 / 454 events under TRUST threshold
Per-track-circuit dwell variance vs each channel's own median SRT, morning peak 07:00–10:00 UTC. Bars fill as the replay progresses; the first bin (0–10s extra dwell) holds the vast majority — exactly the variance TRUST cannot see.
0sθTRUST = 180s300s
Top track circuits by sub-threshold incidence
RouteRequired clear (TCs)Conflicts lockedObserved
160 → exitXL · XM · XN · XP · XQ195, 19729
Signal 160: 1 candidate route; 29 observed clearances in the 07:00–10:00 window.
What you are looking at

The schematic is a simplified topological view of the central Wimbledon throat — Up & Down Main, Up & Down Slow, four platforms, and the principal signals (160, 168, 176, 369, 371, 373, 375…) and track circuits (OP, EP, ES, OM, AA–AE, XD–XK…) that appear in the relay dataset. Real events from the CSV drive each channel: an RGPR relay drop turns a signal green; a TPR drop fills a track-circuit edge red. The decoded stream below translates each raw channel transition into plain English. The sub-threshold histogram is the headline finding: thousands of small dwell variances per peak, each invisible to TRUST, each contributing to cascading network delay.