Not on a boat, a train,a plane or a motorcar, I won't buy cloud games anywhere, not here nor there.

Not here not there not anywhere!  Actually I bought in Wii days....

Would you buy one at the bottom of the Mariana Trench? Lots of pressure there...

To the one that asked whether I would play The Witcher III on cartridge or a cloud version... obviously a native port. I can play it in 7 years without worrying about servers and shit and I can sell it for a few bucks if I want to. To top it off, the game looks alright, it's the bare minimum of acceptability. 

I do wish these ran better. Selfishly speaking, they don't affect me. I have a PS4 Pro and plan on getting a PS5 one day. Thus, these games are useless to me on Switch. And in general, I don't like the ideal of cloud gaming at all.

Okay I watched the video, and as the owner of pretty much all cloud games on the Switch (Control, Hitman 3, Forgotten City and Guardians of Galaxy), I don't really agree with their analysis on some points:
- they only tested demo versions, and from my experience, demo versions tends to run worse than the actual full game
- the framerate depends on your connection as well, and I had at least 3 games running at 60fps (Control, Hitman 3 and Forgotten City), and so is image quality
- Control is by far the best game through cloud, which ran (in my experience) full 60fps all the time in performance mode, and was my most played game in 2020
- I've recently finished Guardians of Galaxy by playing mostly on my Switch Lite, very solid performance pretty much the whole game

I tried Geforce now for 6 months during the beta, even for free it isn't worth using. There's nothing special about the latency numbers from Geforce now. It's adds 50ms which every cloud service does.

Either the game devs screwed up their game for Xbox or digital foundrys sponsored latency program from Nvidia screwed up the numbers. Here's some real native number i found for console:

https://attackofthefanboy.com/news/ps5-xbox-input-latency/

You do realize the input latency varies from game to game right? You do realize that services can improve at time goes on right?

I personally haven't used GeForce now myself but your examples are meaningless if you aren't gonna compare the same games in the same testing environment...

Okey, my point was that a native 60fps game should hover around 50ms, a 30fps game around 100ms and a cloud service will add additional around 50ms. If something else is happening I either blame the testing method or the game devs.

It largely depends on how it's being tested hence why you need the same or similar testing environment to make a proper comparison. Cloud won't automatically add around 50ms as it depends on a lot of factors. Sometimes it will add way more and sometimes, less. As the technology improves and so do peoples internet, cloud will get better. It's just a matter of when and Nvidia so far seems to be leading the pack.

If you think something is off, you need to find another outlet that has done similar testing in a similar testing environment that shows that it's off. From what I heard in the video that's in your example, NXgamer is "deducting" screen latency from his numbers. DF isn't deducting the screen latency from what I can tell as they are using LDAT. They are completely different testing methodologies hence why you see different numbers. Maybe next time, you should research this yourself before thinking something is off...

I say all Nintendo fans show support by buying the KH collections for $90 instead of buying them cheap on Playstation/XBOX. Cloud gaming is the future, do you see Sony and MS doing the same? No because clearly they don't understand what innovation is.