I personally liked the series Connections that showed off that many of these guys did not live in a vacuum. https://www.imdb.com/title/tt0078588 https://www.i

I personally liked the series Connections that showed off that many of these guys did not live in a vacuum. https://www.imdb.com/title/tt0078588 https://www.imdb.com/title/tt0078588 Basically mostly everything is fairly incremental but every once and awhile you get someone who connects 3 or 4 different fields together and something really cool happens. If you work by yourself you will become odd and stale. Bouncing ideas off others helps weed out junk and maybe even help you get a better understanding of what is going on. Or they may ask a question you did not think of and woosh you found something even more cool. I had an idea a few days ago that most ML seems to be mostly curve fitting (except when it isnt). But what if you could turn that network back into a formula instead of a set of nodes with weights and particular follow rules. Those sets of formulas could yield some interesting ideas, I think, if you could compare them. Like there may be a set of formulas that are just hiding there but are smeared across 1000 nodes. Not even sure where to begin on that. But an interesting idea none the less. One of the smart ones could see the idea and also figure out where to start and tease out the idea until it is done.