I thoroughly enjoyed The Checklist Manifesto. Written by surgeon Atul Gawande, the checklist is presented as a tool, not only to prevent failure, but to increase performance. The book is filled with interesting examples of how checklists are applied in diverse fields like aviation and medicine; and the difference between DO-CONFIRM and READ-DO checklists. The book also chronicles Gawande’s own experience implementing a pre-surgery checklist for the World Health Organization.
While reading this book, I was experiencing a problem of my own at the office. Our Night Assistant Editor was exporting the show episodes to the network incorrectly. Exporting a cut to the network is a complicated process prone to error, made worse by the pressure to do it quickly. This seemed like the ideal process to apply a checklist to and so I created the Show Exporting Checklist. Any time an Assistant Editor is expected to export a cut to the network my Coordinator or I will printout the checklist and leave it with our Night A.E.. They are expected to fill out it and leave it on my desk for review the following morning.
The goal of my DO-CONFIRM style checklist is to help our A.E. remember of all the steps involved in an export, and to report back any anomalies to the team. I haven’t recorded hard data, but since implementation it feels like the number of errors has decreased significantly. And the information we gained from our Night A.E.’s observations has provided the producers and editors with valuable information about their episodes as well.
I highly recommend The Checklist Manifesto and thinking about the ways a humble checklist can improve your own work.
In my enthusiasm for machine learning in post production (1, 2, 3) and my firm belief in the value proposition of automation; I may have given the impression that I don’t care about assistant editors, when nothing can be further from the truth. I got my start in ‘the business’ as an A.E.. I worked the night shift for years before advancing to Online Editor. Today, as a post supervisor, I care deeply for my A.E.’s and take great pride in seeing many of them develop into great editors.
But I believe the nature of the technology has changed the A.E.’s role and responsibilities so greatly that it no longer represents the apprenticeship path to editor like it used to. And I think 20th century rules make matters worse. And the field would do well to acknowledge the impending future.
Tapeless acquisition and Avid AMA (and previously FCP7’s reliance on file management and folder hierarchy) have transformed how Assistant Editor spend their time. On any given day an A.E. will spend the majority of their shift shuffling files from hard drives to servers, transcoding media, grouping footage, and preforming other technical skills. But at no point are they learning about the craft of editing. I’ve heard of A.E.’s working a 10 hour shift and then volunteering six additional hours just to cut a small scene. If they can even find someone to offer them mentorship.
In the meanwhile, being excellent at file organization and up-to-date on codecs has nothing to do with being good at pacing, timing, and manipulating images. However, the subordinate ‘assistant’ nature of the job means many struggle to make the jump when they’d be perfectly happy as an A.E. if it paid more and were treated with more respect.
The A.E.’s that want to make the jump to editor are further hindered by union rules that acknowledge only 2 positions: “Editor … a person whose primary skills include the actually cutting or selecting…” and “Assistant Editor … [who] at no time is he/she permitted to edit any portion of the sound or the picture … unless he/she is temporarily upgraded and works under the supervision of an Editor.” The mandatory rate gap between the two positions creates a disincentive for productions from bumping today’s A.E. to tomorrow’s Editor. The union membership would do well to find a way to create a junior editor position with a rate somewhere in between.
Many A.E. tasks can already be delegated to the software (media management if you use Avid Interplay & and Multi Grouping if you use Groupitforme.com) Ultimately I believe the cost savings of machine learning will be too great for productions to ignore. And just as we don’t keep typists and negative cutters around, I believe the A.E. role will be greatly diminished in the near future too.
I’m not happy about it, but the writing is on the wall. We’d do best to address the issue head on.
As I’ve said before, it’s not the editors that should be worried (yet), it’s the Assistants. The underlying technology means that the computer is capable of organizing the raw footage to a level not achievable by even the most OCD A.E.. And the cost savings to a production are too great to ignore.
Side note: where is Avid in all of this? Adobe + IBM; Adobe + Stanford;… are we starting to notice a pattern here? I fear that A.E.’s aren’t the only ones who are going to get left behind in the new future.
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.
Going to the recent NAB convention it was hard not to be disappointed with Media Composer. I mean look:
IBM just built a Watson powered Assistant Editor; that can be integrated directly into the NLE!! And Adobe Sensei is applying machine learning to our creative tools. Meanwhile Media Composer remains stubbornly walled off in the 20th century.
So why don’t broadcast and film editors abandon ship and leave Media Composer behind for more advanced (1) and flexible tools like many did 10 years ago when Final Cut Pro was introduced?
I think it is easy to be cynical about curmudgeonly editors and their stubbornness to learn a new tool. (I think of Paul Giamatti’s cameo on 30 Rock as Richie the editor) But I think the logical reason is that the value proposition just isn’t there, yet.
When Final Cut Pro was introduced, an Avid Meridian based NLE cost between $35k and $60k. At one thousand dollars, Final Cut Pro was a cost savings too big to ignore. But now that all NLE’s are in the sub-thousand dollar range, there is little value that new software can add to the post production process. I’d guess that the number one holdup to a more efficient television post production for most shows, is the notes process. There are few software improvements that can save more money, or speed up the process, then having stakeholders delivery quality notes on time.
The next order of magnitude in efficiency gain is going to be when the NLE is 1) able to reduce personnel costs, or 2) deliver valuable business insights.
A Production company with four or five shows that share a team of five or six assistant editors is a prime candidate for an the IBM Watson + Adobe Premiere combo, because the potential to reduce the support staff represents a savings of $100k+ each year. And Adobe Premiere’s tools are already primed to reduce the ‘technical’ work performed by A.E.’s; for example, advance Grouping and integration with Review & Approval systems like Frame.io and Media Silo.
But none of the NLE’s have introduced features for the stakeholders yet, and I believe this represents nonlinear editing software’s biggest opportunity. (Premiere’s advanced metadata support puts them in a prime position to add these features) A few prime examples, that I’ve written about before would loading footage or cast interviews. Once stakeholders are able to see what they spent money on versus what ended up on the screen; they will be able to make more informed decisions. Just like Google’s Alpha Go ended up teaching the world’s best players new strategies for the game.
What I find so interesting about the current state of affairs is that Adobe is so user forward. Their website is all about selling the experience of using their products to the people who will use their products day-in and day-out. Avid’s website has customer stories to remind you that they are the dominant software tool for film and broadcast television, but their website is clearly pitched at business people making infrastructure investment. There are no pages pouring love into the tiniest interface details of their newest release.
The people who would benefit the most from NLE analytics, are the people who would be the most receptive to buying such a tool, are already Avid’s real audience; decision makers at large companies with a couple of hundred thousand to spend. In the meantime Avid editors don’t really have any incentive to learn another tool because there is no software savings that offset waiting 3 days for notes 🙂
(1) In a highly collaborative environment (12+ users) nothing beats an Avid Technologies Media Composer / Nexis combination. But I believe that Premiere will be ready to compete within the next few years because adding collaboration is easier than adding the machine learning techniques I think about.