IMDb TEAMbuilder

better = better
All the moving parts

Inspiration

If you watch the end credits of a TV show or movie (which Marvel has increasingly trained us to do) it is amazing to see how many people work on a production. Ideas are a dime a dozen, and the burden to crank out content has never been higher. Quality is at risk.

On the one hand, the power of IMDb is that it helps me place that the voice of Grayson on Arcane that I'm watching with my kids is ALSO the voice of the villain mom from Season 4 of 24. Without the joy of IMDb, I couldn't concentrate on the current show.

But the REAL power of IMDb is in the infinite clickable links that reveal the connections and work partnerships in the media production industry. The professional "graph" of longevity and experience and collaboration should be a must-have resource for Execs and Showrunners building their production teams.

I want more high quality movies and TV shows to inspire and entertain and teach -- and since I have pretty much nothing to contribute directly above the line, I want to help those in charge of spinning the magic and draw on the wisdom of IMDb to build great teams.

IMDb TEAMbuilder jumpstarts the production team building process

A lot of times you have a producer and an idea and you have to start building out the hundreds of people required to make a show. From the thousands and thousands of industry workers it's easy to have stronger memories of who you have worked with recently, or who was on that-one-project-that-one-time-that-somebody-talked-about-at-that-one-person's-house...but that's not a scale way to fill out a roster.

With IMDb TEAMbuilder you begin with the IMDB link to your producer (reinforcing to the user the value of IMDb) and a link to one of their past productions that represents what you're looking for in the show. Then the ENGINE kicks in, and using the IMDb professional graph, we begin to build a potential team for the project. In different production categories (producer, casting, director, sound, music, costume, art) the list of millions if prioritized based (among other things) on how often they have worked with the producer, overall experience, and if they worked on the past project you've identified as representative.

You can build a list of 1 to 100 people per role, and the IMDb TEAMbuilder will kick back a prioritized list of talent, by role, with a list of past collaborations with the producer and direct links to their IMDb profile.

AWESOME

How We Built It

AWS does so much of the heavy lifting that devs can spend their time thinking of better solutions instead of grinding out dev environment management, oauth varieties, scalable storage -- so we went heavy on AWS. Python Lambda, AWS Data Exchange, API Gateway... After we got an idea, Step 1 was to explore the API to find out what data could help bring our dream alive. There was a lot of "guessing" as to the right way to call the API to get what we want. We experimented with different weighting schemes to create talent separation and support our story of great team building. Finally, to make it "playable" to the Devpost community we created a slightly-more-sophisticated-than-usual Google Sheet that facilitated input to the API and displayed the results.

Challenges We Ran Into

Two things you can take away from the video -- you can tell I have a cold in my voiceover and we spent a lot of time trying to lock down the right idea. We went pretty far down the road on a crypto-programmatic-analysis project, but in the end with only one coin and 30 days of data it was not very compelling, so we regrouped.

Accomplishments that I'm proud of

We wrestled with FUN ideas and decided to pursue impactful. Which ended up being more fun probably than any other project we might have pursued. You know you're on to something when the "future feature" whiteboard needs to be photographed and erased to make room for more.

When we hit documentation gaps, we're glad we were creative enough to guess the right solutions, and it was a great intro to AWS Data.

What We Learned

When I am in the "shopping mode" for a data source, I have a pretty short attention span. Some of the APIs we banged up against early in the process "killed us with friction".
I may need a couple of refresher courses on parts of AWS. It's been so long since I originated an API (instead of just building out the code) that I had to relearn a lot of the settings. Also, I know there is a better way to troubleshoot between Lambda and API Gateway, I just know it.
The IMDb graph is powerful. The biggest impact to productivity on this project was endless clicking on links and seeing all the great things so many people have worked on.
We'd never used App Script with Google Sheets before. It was pretty straightforward and easier to use to fetch data than a lot of big dev languages

What's next for Scenario Tester

I think our wish list for v2 is probably unmanageable at this point and needs to be prioritized.

1. ERROR MANAGEMENT We have very limited error testing at this point. Except for some cursory "is that IMDB link formatted correctly" (not even "is it a legit profile"), you can fire up the engine with bad inputs and get bad results.

2. PRETTIER REPORTING It's still a big "print it out" world. We got pretty far down the road with PyPDF and server-side report creation, but by the deadline it was still a little too jenky to be included in the submission.

3. EXPAND TALENT PRIORITIZATION Collaboration frequency, longevity and attachment to a single production is a great start, but there's a lot more gold to be mined. a. First on our list was "is this person available?", but it wasn't clear how to parse out if someone was working on a 2022 production if it had wrapped or was ongoing. b. Incorporation (or at the include in reporting) of awards c. More fully embrace working on successful productions. A junior person who got to see first hand the majesty of Aaron Sorkin and Tommy Schlamme on West Wing may have more to offer than a senior person on "I'm Dying Up Here".

THE CHALLENGE

Since this project was spawned from a challenge, it guided a lot of our dev. They turned out to be pretty good landmarks and so we'll probably keep that list on the wall: Technical Implementation: The Google > AWS > IMDb > AWS > Google flow has ended up being very solid. For the objective of "riding the professional graph" we did a great job of being broad (producer, casting, director, sound, music, costume, art) and working to push "separation", i.e., if the sorting pack is too tight, it's not meaningful. We tried and discarded data items (sometimes with tears) and kept a list of "I wish this API did ____".

Potential Value: This is what killed a lot of our earlier ideas. There was a lot of energy behind creating a "Double Feature Finder" but it failed the "if we make the best one ever, would it change anything?" In the end, we decided a scalable tool to support production team creation could create real value for the users, and ultimately, people sitting on their couches at home.

Quality of the Idea: Our most fun ideas and the ones that my 11 year old liked the best got knocked off by the Potential Value screen. The great details we included were requiring an IMDb link to kick off the work and then presenting IMDb links at the end -- it reinforces the brand and value of IMDb and we think more likely fits in with how people work. Obviously we are biased on this point.

Additionally: we also tried to put some effort into the quality of our submission. People tend to get a little heated when videos are more than 10 seconds too long, so we tried 7 seconds under the mark. Also, we had a much more serious version of the video that was very "traversing the graph" and "complex deductive" and "propensity weighting" and "Djikstra's Algorithm" and "cross platform data transport"....it just didn't "feel" as fun and interesting as our topic of making awesome TV shows and movies. So we turned our voiceover budget into pizzas, committed to two takes or less, picked the guy with an incredible cold, and tried to make a video that captured the spirit of the work but didn't drain your soul to watch it.

Feedback: oh, we left feedback.