Off topic nerd alert.
I've worked out what caused these high speeds.
First of all I got on my turbo and rode the Tacx circuit myself and gave it some beans on the segment in question. I kept an eye on my simulated speed as I rode it and, while it was way quicker than I go in real life, it wasn't mad. About 40-odd km/h tops. I uploaded it to Strava. Disappointingly, my speed on that section was a realistic 40-odd km/h (20-odd mph). Also Strava immediately twigged that it was a virtual ride (It was tagged as such in the TCX file). It also wouldn't match the Tite St segment. I don't know why not.
So these riders must have done something funny when uploading, or maybe Strava has upped its virtual ride tag recognising game since they uploaded. I don't know and will never know, so I'll leave that as a mystery.
But why the high speeds? This is where it gets "interesting". Time to have a look at the raw data, so I imported the TCX into Excel.
The TCX file has a record every
second, saying what my power and virtual speed was at the time. But it doesn't have sufficient location data to tell me what my virtual location was every
second. It just has a series of relatively widely spaced waypoints, and I jump from waypoint to waypoint (unlike a real outdoor GPS track which would have location for all the intermediate points). These location points are just handy for plotting the route on a map. They are completely inappropriate for calculating speed. You have to use the speed that the trainer provides for that.
Soo ... I calculated the distance between the successive waypoint co-ordinates using a bit of spherical trig and divided by the time taken to get between waypoints and the resultant speed went bonkers. I clocked 65.2 km/h (40.5 mph) going down the segment in question, skittling virtual Telegraph readers as I went. It was carnage. Not quite 52mph, but not bad. I could probably have gone faster by lying to my turbo software about how much I weigh.
In summary, somehow or other (maybe by saving the data as a GPX, not TCX, file) and uploading to Strava, somehow Strava thought this was a real ride and for some reason it took a naive approach to calculating speed from the given waypoints. This was an invalid approach - hence nonsense data.
End of off topic nerdery
Edit. Done it. I saved my ride as a GPX and edited it to tell Strava it was a real ride. Strava clocked me at 70.9kmh (44mph) on the segment, and also reckons I was putting out 110W. In actual fact in the simulation my simulated speed was 40-odd km/h and power was a bit over 300W. I think that might give me the current KOM
I've kept this private on Strava as editing files to fool the algorithnm in order to get KOMs is, I believe, not cool. But I might uncloak it just to make a point. (I have now done so. I have a KOM. Go me!
https://www.strava.com/activities/11469737608/segments/3228307530918861610 )
View attachment 731714
View attachment 731702