Management People

Dinosaurs and the Intelligent Assistant

Matthew Weiner describing his show Mad Men:

“The story to me is about the onset of a subversive ethnic point of view that has not yet poked through to Sterling Cooper. They’re dinosaurs.”

When was the last time you’ve seen a secretarial pool?

A good secretary can perform a range of functions from typing letters, to booking travel arrangements, and picking up lunch. And it used to be that every middle manager in an office had one. And yet, despite the wide range of “activity flexibility” a secretary offers, how many do you work with today?

The cost of employing a human being was slowly offset by good enough technology: the copy machine, the fax machine, the personal computer, the internet, and the smart phone. Technology isn’t better than a secretary, but it’s good enough and doesn’t require health insurance or a pension!

I believe the role of the Assistant Editor is undergoing a similar change. Fifteen years ago the ratio of Editors to Assistant Editors was 1:1 or 1:2. Today I see production companies that have two or three Assistant Editors spread out over two, three, or four shows! Advances in Machine Learning and Deep Learning are only going to accelerate this trend.

If you only read one thing this week, I highly recommend Philip Hodgett’s post: Maybe 10 Years is Enough for Final Cut Pro X. In it he articulates how machine learning and deep learning are already transforming tools like Premiere; making them “good enough” for hundreds of thousands of editors around the world. These tools are only going to get better and better. At what point is “good enough” going to replace the cost of health insurance and a pension?

Media Theory

Computational Video Editing for Dialogue-Driven Scenes

I finally got around to reading the Computational Video Editing for Dialogue-Driven Scenes paper from Stanford. For those who don’t remember this technology from last July I’ve linked to the video below, followed by my thoughts. Interesting stuff.

  • It’s interesting to see interfaces that expand the NLE beyond the “Source/Record” model. The idea of interacting with your footage based on idioms is an interesting one because it creates a “natural language” way of interacting with your footage. As the importance of visual literacy grows it is fascinating to imagine a world where people edit without a traditional timeline.
  • I think this technology is an example of ways that the edit suite could change in the future; and put additional downward pressure on the Assistant Editor. I’ve long argued that tools that eliminate technical quantitative work are going to create huge pressure for producers to operate more efficiently. (That’s why only the very few have dedicated secretaries.) A tool that performs sequence prep and allows fast ideation is compelling indeed.
  • That said, while interesting work, I’d guess that 99.9% of all the video that’s edited worldwide is unscripted and therefore not applicable to their work. That the idioms of unscripted editing are an order of magnitude harder than scripted is completely logical when you think about it.
Management Software

IBM Creates First Movie Trailer by AI

The title is a little misleading, but this is a good example of how machine learning is going to revolutionize post production:

Utilizing experimental Watson APIs and machine learning techniques, the IBM Research system analyzed hundreds of horror/thriller movie trailers. After learning what keeps audiences on the edge of their seats, the AI system suggested the top 10 best candidate moments for a trailer from the movie Morgan, which an IBM filmmaker then edited and arranged together.

As I pointed out in a previous post; IBM has created a potential replacement for the Assistant Editor. It doesn’t need to be perfect, it just needs to be good enough, and cheaper than $1922.80/week.