True, further tests would be useful. The game would also render faster if you go for the 1080p or 720p option, thus less latency at the source. If the game can render at 120 fps at 720p, while only sending half the frames, you still have cut the render time in half. In the case of a 30 fps game at 4K, that's 25 ms saved. Also compression and decompression will be faster at lower resolutions and bandwidth, as well as the transmission time per frame. It all adds up.
Yeah, I'm sure we'll see more comprehensive tests in the coming days. After seeing these preliminary tests I'm mainly curious if they'll find a common correlation between higher end options, and more latency issues. Which also makes me wonder how spread out Stadia's data centers are.
And if one data center has hardware that only focuses on 4K streaming, while another data center has hardware that only handles 1080p streams.
Though either way, it seems quite possible that you may be matched up to a data center further away because the ones near you happen to all be busy at the moment.
So essentially, during the Washington Post PC test, the closest available data center may have been further away than the one that was streaming to them during the mobile test.
Or perhaps they used different ISP's during the two tests, which may have resulted in the data taking unnecessary extra routes, depending on how the ISP handles its data.
The interesting thing there, if it's the ISP's fault, is that it may not even necessarily be a 'problem' with the ISP per say. As in, it's not something they can, or will, fix. But rather depending on how certain ISP's handle their data (some make extra stops between a parent company and the daughter company for example), you may have a bad experience with Google Stadia with ISP's that otherwise function well for you.
So this could potentially be an issue with Stadia that you can't predict until you test it out with any given ISP first.