I have improved the stability of my dev environment. The windows partition needed to be a bit bigger because honestly Windows and everything that installs to it is huge. But because of the primitive way Windows 10 handles OS partition resizing I had to free up adjacent space on the disk before I could extend it. Luckily the adjacent partitions were just a recovery partition and a data drive so I was able to backup & delete those. Windows already feels happier, boots up faster and has way more space (went from < 50 gb avail to 200+). The solid state drive also arrived, however it is too advanced for this particular laptop. It does have an m.2 2280 socket - however, all 2280 means is the physical dimensions of the chip is 22mm x 80mm. The architecture of this SSD is NVMe. It took some digging to find out with ASUS but the architecture of this laptop's .m2 bus is SATA. Weird that they would change the bus architecture but use the same connector. But I don't have the heart to return the new SSD so it sits there now mocking me, daring me to build a new computer with it.

In the meantime. My first experiment with horizontal collision detection was unsatisfactory. The Android VM that I am using as a debug environment really seems to struggle with the Xamarin Mono runtime. Just rendering the frames with no collision logic it seems to top out at ~20ish FPS. (Oddly, when I connect an actual Android phone to debug with I get 1 FPS??). If I add a naive collision check (X actors x Y obstacles, or quadratic complexity) the framerate dropped by as much as ⅓ in the VM while Bob was moving. So there's no escaping it: I need a better algorithm, and I need to multi-thread this loop. In addition to this, the collision detect didn't work - Bob just kept on sailing choppily through. The collision checks work in unit testing, but the game is not a unit test. I've still got a lot of work to do on this one.