The scariest aspect of all of this is short term memory modification. Let's mess with your mind's hard drive every two hours and see where you sit after a month's time. I don't think a non disclosure agreement is all that necessary when all of your employee's brains are turning into gray colored tapioca.
I'm surprised more people haven't brought this up. Semaphore making movies didn't scare me, bother me, or horrify me. But the thought of hundreds of people subjecting themselves to multiple runs of elective brain damage every day did. What the hell? I'd rather just wait for the movie to come out so I can watch it without having my brain scooped out like a gourd.
This story had me feeling annoyed throughout the whole thing. I think the biggest reason was cognitive dissonance from the premise that this process would create movies that would be considered high quality by a movie critic, that they would work on many levels, and have a strong emotional core. I can believe that this process would create summer blockbusters of the variety that appeal to masses of people who don't intend to analyze the movie, but I don't believe in any way that this would create something like a Pixar reputation.
The math just doesn't work.
I agree. There are algorithms for optimization in this day and age. But taking a movie from crap to great is more than just a few variables, and each of those variables has a continuous set of values. Throwing parallel computing at it wouldn't really do the trick either because each part of a movie ties into the other parts, there would be few independent variables so it doesn't divide and conquer very well.
Even besides the computing, the amount of human time they'd have to PAY people, and the time itself, would just be ridiculous. Rather than pay hundreds of people to watch movies that are incredibly bad most of the time (as they would be when you're iterating from a random point), you'd save much money and time to just PAY A TEAM OF WRITERS to make something that doesn't start at complete crap (and then maybe use optimization to tweak a bit here and there if you had to). A lot of the time of optimization algorithms depends on having a good educated guess for the startpoint--they should hire filmmakers to provide that start point. And all of that time, and all of that money, would produce (in my opinion) a movie that's probably somewhat lower quality than a current Hollywood blockbuster, but requiring more time and money to make.
I've come to appreciate in the last few years how much I value the writing behind any TV show or movie. You need good actors, but a good actor will give you nothing worthwhile if there isn't good writing behind the scenes.
This approach to making movies reminded me of the Infinite Monkey Theorem--that if you sat a monkey in front of a typewriter for infinity, that it would write the works of Shakespeare eventually. What's more interesting is that it would theoretically also write all of the works that Shakespeare ever MIGHT have written and so on. But the real work comes from separating the crap from the good stuff--that takes a lot of work, and it would take a lot LESS work to just write something good in the first damned place. And all of that is even assuming that a monkey's key presses would be completely random from one key to the next, but I think key-mashing where a big group of keys is pressed right after each other would be more likely, which would probably never result in anything by Shakespeare.
But I digress...
An interesting idea, though I never really bought the basic premise that this would produce critically acclaimed movies. Even besides that, I wish it hadn't been written as one giant infodump. Ken Liu can do better.