The Steam Machine did not sell well, however it gave Steam the bandwidth to work on Proton and SteamOS which ultimately lead to the deck, that did not flop.
Might I interest you in the much lesser well known Steam Link? Also a Desktop-Console-type device. It gave Valve the experience to give us the ability to Stream games to other computers on the network via Steam, thus saving distributed space.
Or the Steam Controller, which is also referenced in the article? They learned lessons and fixed the problems with it for the Deck control scheme, which is now being used to create a better product.
Or the Valve Index, which is also referenced in the article? They learned many lessons by using this as a test bed, that allowed them to make the Steam Frame.
All you have done in this entire thread is indicate that you don't understand iterative development, and that a company would be willing to make mistakes to learn from them and improve.