Virtual Holiday Day Thirteen—The Big Loop#

Bullas, Zarzadilla de Totana, Aledo, Canal, Jardin de los Molinos, Gebas, Fuente Librilla.

Last year#

Strava 2025^🔗:

Changes for this ride#

No changes. The aim with this ride is to see if performance degrades over what will probably be a four-hour ride. I mean—seriously? Four hours on an indoor trainer. Are you mad?

Later…

Well, I said no changes, and that was the plan. However, I decided I would fit a cadence sensor. We picked this one up from Halfords, which is a strange UK retail chain that sells car parts and bicycle parts. Shopping there can be a very random experience. The staff are always really helpful but the stock is just… weird. I once failed to find a 700C inner tube but picked up a Shimano Ultegra chainset for £150. “Are you sure about this price?” (I did query it, I’m honest me) “Sure—it’s end-of-line, ten years old.” Sold!

This year#

Strava 2026^🔗:

Post-ride analysis#

First off—the cadence sensor worked. I don’t know if it is cadence sensors in general, this specific cadence sensor, the way I am reading the data or the way I am sending the data to the main program—but it seems very keen on reporting zero cadence when I stop pedalling for a moment and then taking its time (about three seconds) to report when I start pedalling. This is not a criticism—merely an observation.

I think I might be on to something regarding the ‘spikes’ (where the time per frame jumps from a normal 30-40 ms to an abnormal 6500 ms).

On this ride there is a café stop in Bullas. To increase my immersion I often get off the trainer and have a ‘comfort break’ in the usual places. Therefore, I stopped in Bullas to have a cup of tea.

When I resumed the ride it went into ‘spike mode’. Every time I crossed an AS30 boundary there was this terrible lag. It was of course massively annoying but I’m here to work these things out, not get annoyed—so I pressed on. The image below shows a particularly bad ‘deviation’ when it was lagging as I tried to turn. I actually went off the road by several hundred metres before I could get back on track.

This is not good, but I kept going while I thought about it. My working theory was that it was the number of trees. More trees = more lag. However, I know this isn’t true because quite early in the ride I had 60,000 trees loaded and I was crossing AS30 segment boundaries with only the slightest lag—spikes around 120 ms. Perfectly acceptable.

My current theory is that it is something to do with trees—but not the absolute number of them. I think it is the way they are placed on M30s and AS30s and specifically the way they are moved off an M30 when a new AS30 is created. The key insight was that it happened after getting off. I think this gave the system time to search wider for trees and place them on distant M30s. Normally, as I am moving, new trees are added to AS30s that are being created every couple of minutes or so.

I will examine the code. Make some changes and see what happens tomorrow.

A Final Observation#

I am seriously impressed that I managed to complete a 150 km ride on an indoor trainer. Prior to starting this Virtual Holiday experiment, I doubt that I had used the damn thing for more than six hours in total.

But here’s the caveat: I think this works well because I know the route. My mind is filling in the gaps as I ride. And it is doubly interesting because I am constantly doing a ‘compare and contrast’. There is a little frisson when I spot something that triggers an “Oh—it really does look like that” moment, but conversely the “Hmmm, that’s odd” moment is also interesting—in a different way.

I’m nervous about what it will be like to ride a route I don’t know. I’m actually questioning if that is what this system is for.