Of course, in a live broadcast, ideas for the original game to build could come from viewers.
I could make pastel sketches faster than he makes oil paintings but they'd look like they were drawn by a kid and I'm sure I would come across as naive, inarticulate, and ignorant of art technique though I am sure I'd get better at it if I did an episode every week for years.
Making a good show of it would involve having a plan, framework, and really knowing what you're doing.
I would look to this guy's video series on building CPUs based on FPGAs
https://www.youtube.com/watch?v=2fNBkUCjhcE&list=RDCMUCBcljX...
Pretty clearly he is starting at zero and teaching himself as he goes along and it's entirely unrealistic that he's going to livestream developing his Zork machine. However, he approaches it in a way that is really pedagogical and anybody who follows this series will learn a lot about how simple CPUs work, how you make logic on an FPGA, etc.
He eventually regroups and starts a CPU building process he really can finish. It's a good example of an ambitious development project on YouTube I think.
For any given creator, I doubt I’d watch more than 45-90 minutes unless it was exceptionally interesting (which seems not likely).
That will tell you what you want to know
tl;Dr, to be successful you have to produce a certain type of content that can loosely be called programming. It is primarily entertainment, not educational