TENTPLE

About Tentpole

Why did I build Tentpole?

I love movies. I love talking about movies. I love movie podcasts. I love movie posters. I love casting “What ifs.” I’ve spent entirely too much time thinking about Quentin Tarantino’s unmade Star Trek picture or the version of Michael Mann’s Collateral where Adam Sandler doesn’t drop out due to scheduling conflicts. (He was supposed to play Jamie Foxx’s role, and, although I love Collateral regardless, I feel like the Sandler version would have been better. Jamie Foxx is too cool. When he goes to the bar to trick Javier Bardem, Foxx is too convincing. No cab driver can flip a switch to that level).

I think about Tentpole as fantasy sports for movies. Fantasy sports is building a fictional reality where Patrick Mahomes gets to throw to Ja’Marr Chase. Tentpole is building a fictional reality where Meryl Streep and Roddy Piper are in a James Wan film. Playing the game and watching the reel land on “Heat” or “One Battle After Another” gives me the same “YES!” feeling as Jahmyr Gibbs scoring a touchdown for my fake football team.

How do the box office and critic scores work?

I’m a data scientist by trade, so I did regression analysis to give weights for the various inputs and interactions in the game. These scores are deterministic, meaning the same inputs will provide the same scores (i.e. not LLMs). I would not bet my life on the accuracy of these scores, but I think they are directionally accurate enough to feel real.

I don’t want to publish the scoring rubric because I think part of the fun of the game is discovery. I will give a few hints to help:

Directors and Leads get the most weight, so choose them wisely.
An actor is scored the same regardless if they are lead or supporting. That score is given different weight depending on their place on the call sheet.
Directors turn in a consistent performance, regardless of the cast. Actors turn in a different performance depending on who they are surrounded by (the data supports this).
Think about actors in terms of horizontal range (genres) and vertical range (quality of performance).
- Nicolas Cage has a wide horizontal range (lots of genres) and tall vertical range (varying quality).
- Daniel Day-Lewis has a narrow horizontal range (mostly dramas), and a tight vertical range (always high quality).
- Should you pick Cage or DDL to lead your movie? It depends on your movie’s genre, director, rating and supporting cast. There are scenarios where Cage outscores DDL.

Where are the movies I’ve made? Is there a leaderboard?

Your fake movie generates a unique, persistent URL that you can return to forever (or at least until something breaks). A login where you can save the movies you’ve made is coming.

A leaderboard is intentionally absent. As soon as there is a leaderboard, there is an incentive to optimize. The goal of the game is not to refresh the page until you get Christopher Nolan (i.e. Nolanmaxxing), which I think people would do if there was a leaderboard.

Take a game like Immaculate Grid, which I adore, but became stale to me because once you figure out the handful of obscure players who you can pick every day, you’ve won the game. Tentpole isn’t a game to win, it is to make the best with what you’ve been given, which I think is true to the spirit of filmmaking.

Is AI used to write the movie title, descriptions, and quotes?

Yes. The first version of this game did not have the fake movie titles or descriptions. It was just the box office and critic scores. It was fine. However, as soon as I started playing around with AI generated titles and descriptions, the game went to a new level.

The game is truly at its best when you are building toward a Scorsese-directed crime movie and you have DiCaprio, Gandolfini and Pesci, but then you run out of reshoots and it ends up a space gangster picture. Or you have built out a cast of Oscar winners and just need a lead actor to tie it all together, and you are forced to cast Michael Jordan. The descriptions are now my favorite part of the game, and the critic quotes in particular are often hilarious to me. If that turns you off from the game, I totally understand, there is a switch at the bottom of the cast list that says "Score only mode." Flip that and you get the same game, just without the AI summary.

But isn’t AI destroying resources, warming the planet, and wasting water?

Probably, but so are the lights in my house, the TV I watch, and the car I drive. Blaming consumers for climate change is a psy-op by the fossil fuel industry. We can’t live like Doug Forcett. The movie you make on this game has the same number of LLM calls as a Google search.

Maybe I’m an optimist, but I think AI creating a higher demand for electricity will be the catalyst for green energy solutions and might even be the thing that saves us from climate change. Also, In terms of water, let’s get rid of golf courses first (or at least tax them appropriately).

Aren’t these LLMs built off data that they don’t own?

I fully advocate for AI companies providing a dividend that represents their underlying training data. If AI is trained on the sum of human knowledge, then I think they have a responsibility to compensate humanity for revenue generated from it. Let’s think about culture as a natural resource that nobody owns, but we all benefit from. Under that framework, I say we do the Alaska oil dividend but for knowledge. Some fraction of every dollar an AI company makes should go into a fund that gets distributed to all Americans. UBI – let’s go.

Aren’t these LLMs projecting a caricature of these actors and directors?

Part of what makes the game fun is LLM’s perception of what type of movie Wes Anderson makes or what kind of actor Mr. T represents.

However, I have run across a couple of situations where the LLM was acting in a way that I didn’t like. For example, when Christopher Reeve was selected for the fake movie, his character was often described as disabled or paralyzed. Similarly, Val Kilmer’s fake movies often had his character whispering. I see why the LLM made those associations (in particular because those actors appeared in films with those ailments), but it doesn’t mean I need real world sadness to spill into fake movies. I put in safeguards for those cases and a few others, but if you come across any other examples, I encourage you to reach out so that I can make manual adjustments to the engine: jesse@tentpolegame.com.

Why don’t you remove noted sex creeps and/or cannibals from the game?

I considered it, but the fact is, horrible people are a part of movie history and it’s impossible to scrub them from it. To remove Woody Allen movies would be to remove Diane Keaton in Annie Hall. Ultimately, if you don’t want to cast Kevin Spacey or Jeffrey Jones in your fake movie, then just don’t do it.

This is an affront to art and life itself!

Respectfully disagree. This isn’t replacing movies, it is engaging with movies. My perspective is, this game gives me a new way to interact with an art form that I love. And if it nudges one person out there to go watch a movie they wouldn’t have otherwise, then it’s a win. If you like the game and hate AI, flip the "Score only mode" switch at the bottom.

This game won't be for everyone, but I love it. In fact, if I can be vulnerable for a second, I’ve had a few emotional moments while playing the game. I loved John Candy and thought he was an underrated actor. In the game, I was building a Coen Brothers movie and the reel landed on “Planes, Trains and Automobiles.” I got to pair Candy with legendary directors that he never got a chance to work with, but would have gotten an amazing performance out of him. This fake movie brought a real tear to my eye, because for one brief moment I got to picture that movie in my head.