I'm building technology to help filmmakers limit their risk

Too artsy = had to google the plot and was on my phone half the movie
 
The last time I saw something so creative that it made me cry was in an episode of American Horror Story, season 2 - Asylum .

Jessica Lang's character was committed to an insane asylum. The story was progressing in a good but predictable manner when Sarah Paulson walks up to her with concern and asks if she knows her name.... Then this happened.

It was so off the wall and unexpected. Obviously I've never gotten over it. To me, this is what creativity is; shatter the mold!.
 
Last edited:
My point earlier was that whether an idea is good or not is frequently perceived very differently based on how much money that idea was backed with. Mediocre ideas in the hands of wealthy people are seen as amazing, and brilliant ideas from poor or ugly people are often interpreted as garbage. That's why good ideas don't come from microbudget filmmakers. They do, they just don't get recognized as having value until someone with money voices them.

I occasionally post this video, it's globally significant concert violinist Joshua Bell, dressing up as a hobo and playing a million dollar Stradivari violin in the subway. Watch the mother pull the kid away from the dangerous, untalented hobo. You can tell that he doesn't deserve to be paid for his playing, because he's poor.

In the second video, we see him on stage at 25k a night, once he is surrounded by wealthy people in an expensive environment it's suddenly clear that the exact same skill is worthy of huge financial rewards and respect.


 
You are going a bit off the AI thing here but yeah perception has a lot to do with it. Some unknown band playing in a dark bar people barely paying attention. Suddenly someone finds out they are a "signed" band playing a secret show and that they have 2.4 million followers, and the applause erupts and they are texting their friends, even though they have never heard of them before. Suddenly the music is good lol! Yeah, that happens.
 
Nate North said:
My point earlier was that whether an idea is good or not is frequently perceived very differently based on how much money that idea was backed with. Mediocre ideas in the hands of wealthy people are seen as amazing, and brilliant ideas from poor or ugly people are often interpreted as garbage. That's why good ideas don't come from microbudget filmmakers. They do, they just don't get recognized as having value until someone with money voices them.
Click to expand...
I completely get what you're saying, Nate. The unfairness of it all obviously haunts you daily. I just don't know why you constantly torture yourself. Sometimes you allude that you were once one of them but now, for some reason, you have been demoted into being one of us; the not well connected, short on money, would-be if only we were given a chance. I don't know what to tell you. I've had my windows of opportunity. I was given a chance and I blew it. I know that. It was my fault. The chance came and I did not handle it in a winning way. I'm guessing you've had your windows of opportunity too. I think that in a lifetime most people could see how they too had a chance at something and, for whatever reason, they let it get away. They missed the brass ring. I don't beat myself up because I know that at least I was given the chance. Maybe I just wasn't good enough. I'm not necessarily talking about film making. I've tried several things in my life and I think it was me who blew the chances I was given....
 
Last edited:
Mona: OMG I'm at the ________ show have you heard of them?
Lisa: OMG yes they are amazing!
Mona: I knowwwww! I was at the bar thinking the music was shit but someone told me they were famous, it's amazing!

😂
 
indietalk said:
perception has a lot to do with it. Some unknown band playing in a dark bar people barely paying attention. Suddenly someone finds out they are a "signed" band playing a secret show and that they have 2.4 million followers, and the applause erupts and they are texting their friends, even though they have never heard of them before. Suddenly the music is good lol! Yeah, that happens.
Click to expand...
like this
 
directorik said:
I've had plenty of exec's analyze data and use it to augment my scripts.
Sure, it takes a long time but the data was all there, I'll off the most extreme
example I experienced:

I had a zombie western. This was just before “Walking Dead” exploded the
zombie genre. All the data showed audiences didn't want zombie movies.
And westerns were not terribly popular. But the analysis suggested that if
I removed the zombie element the script had potential of attracting the
right talent.

Now we'll never know because the movie wasn't made but I suspect if I
had pitched the same script one year later – when AMC made zombies
popular again – the data would have shown that people were open to
zombies but maybe not a western.

But the project was my passion. And there were several ProdCo's interested
over the years, but I never got it green-lit at the budget I needed. I had no
interest in the intuition of a numbers crunching, statistic following exec with
no imagination. I needed someone to take a risk on creative intuition.

As Nate said, this might be an amazing tool. But can it show that something
that has never been done before might touch an audience in an unknown,
unexpected way? Would this AI have seen the lifetime value of the
$125,000 "Night of the Living Dead"?
Click to expand...
@directorik, I'm not into zombies or other horror, but, since no one knows anything, they can't use analytics to predict. I know of one student startup in Canada that was using statistics to predict what works and what doesn't, but, even though it's been almost a decade, that student hasn't gone anywhere. Relativity media also tried the same thing and went bankrupt.

A similar phenomenon seems to be true in the stock markets - the Random Walk Theory means that, if you pick stocks at random, you can still match or beat what the experts say. Mega billionaire Warren Buffett made a $1 million bet with a wealth manager, saying that the best stock analysts in the world cannot match the S&P index, and he won.
 
Exactly my point, Mogul. Thank you.

I'm very interested in ben.ritchie's product. And that interest makes me curious. I don't
believe that statistical analysis can predict the lifetime value of a film project or help
decide which projects to back or which partners to work with. But I would love to discuss
it.

One of ben.ritchie's bullet points I think some statistical analysis could be of great help:
"How is my title performing?"
 
There's this thing called market research. It's where you actually ask people 'what do you think of this?'

I think it's as close as you could ever hope to get if you're trying to predict anything.

Would it have worked for George Romero? I don't know. I'm guessing not. Describing something is not the same as the finished product. I mean, in your market research questionnaire could you ever hope to capture the chemistry between Ben and Harry? Or be able to describe the creepiness of the music and how it added to the cemetery scene. I don't think so. It's those things that made the film great. If another team of film makers took George's script back then and produced the exact same film but using their own ideas for casting and music and camera angles, there's a chance the film would have been just another B movie. Maybe not even that good.

As far as predictions, I think the best you could have done was find out what people thought of the idea of the living dead eating people. In the late 60's the idea was probably repulsive.

Can some AI program tap into enough social/economic/trending/world affairs/etc data to predict a hit? Why would anyone ever think so? I agree; it's like trying to predict the stock market. It is not predictable.

The movie ISHTAR starred Warren Beatty and Dustin Hoffman, two of the biggest names in Hollywood at the time. It should have been a hit. Who would have thought otherwise? It bombed in a major way. Would an AI program have told the producers "NO. Don't make this movie!!!!" I doubt it, but even if it did, would anyone have listened?
 
directorik said:
It's an interesting idea.

Creative people make mistakes and create and back movies audiences
don't like. But they also surprise audiences often by making movies exec's
did not believe in. Exec's who looked at all the statistics.

Everyone reading this can bring up a dozen examples.

I wonder if AI and data points can come close to the instincts of
creative people...
Click to expand...
Our goal is for filmmakers to have all the information to hand when they play their bet for a film, rather than go in with bias / preconceptions that are sometimes imbalanced vs the reality of the market/audience for prior comp titles.

Too often we see filmmakers only facing the cold realities of the market/audience/historic performance of comp titles too later to be able to help them manage their risk
 
indietalk said:
I'm curious if this is simply an algorithm (data in, result out) or actual AI (machine learning)?

"Generally, an algorithm takes some input and uses mathematics and logic to produce the output. In stark contrast, an Artificial Intelligence Algorithm takes a combination of both – inputs and outputs simultaneously in order to “learn” the data and produce outputs when given new inputs."

Can you expound upon the technology?
Click to expand...
Great question!

We (and our dependent data providers) are indeed using examples to train our machine learning to improve statistically rather than simply reconstituting existing facts into new presentations.

The dataset that we'd love to get more access to is sales data, when filmmakers share their sales data with us our models improve and we are better able to explain to filmmakers what the 'going rate' is for different types of titles
 
indietalk said:
So will it be free or $800 a pop? Your website says the evaluation is an $800 value. But your mantra is a free tool.

This could easily be a good tool for the toolbox. As in, adding to the already present due diligence.
Click to expand...
We are offering free analyses to filmmakers, but there are costs associated for us to build reports so has to be a limited time promotion to expand awareness.
 
indietalk said:
The only way this stuff "works" is with proven success. It's not the AI, it's not the calculations, it's the success. So we won't know for 5 years anyway.
Click to expand...
There are 'packages' that we are seeing in the data that have consistently 'worked' for Producers over the last five years, and will almost certainly continue to be 'safer bets', one example of which we write about here : https://tinyurl.com/2m9udt96

But we are not trying to turn film-making into a monotonous packaging exercise, we want to support filmmakers with their bets (we don't pretend or try to eliminate risk in filmmaking) to maximise their outcomes and understand risk
 
directorik said:
Exactly my point, Mogul. Thank you.

I'm very interested in ben.ritchie's product. And that interest makes me curious. I don't
believe that statistical analysis can predict the lifetime value of a film project or help
decide which projects to back or which partners to work with. But I would love to discuss
it.

One of ben.ritchie's bullet points I think some statistical analysis could be of great help:
"How is my title performing?"
Click to expand...
Hi,

Please do feel free to connect here : https://tinyurl.com/2obrghz5

You've got nothing to lose, would love to take you through a report for one of your projects!

Agreed, understanding performance of comparable titles and your releases is far less predictive/contentious and yet is information that isn't readily available to filmmakers in this post-cinema-dominant consumer environment

Cheers
Ben
 
Nate North said:
I understand how AI works, and have built and trained and used many over the years, and it could be useful, kind of. I would imagine the issue is that a machine would do pretty much exactly what an executive would do. It can aggregate and analyze data, predict trends, but ultimately I would imagine it having the same problem that we have with the current decision makers. Basically, the major flaw in entertainment funding is the tendency to try and repeat successes of the past. We call it "Sequelitus" and a few other names, but this is the major failure of Hollywood businessmen. It just boils down to a lack of imagination, an issue current gen AI certainly shares.

The issue is that popular art follows a different ruleset than basically every other type of business. The single strongest positive factor an artwork can have is novelty, but you can't do math or train AI models on events that haven't happened yet. I could be wrong, I don't know exactly how this system works, but I think unless the AI can recognize which ideas are "fresh", we are just talking about replacing the network robots with literal network robots. There are aspects it probably can predict pretty well, like what is the marketability of a picture combining star x and star y. There are still a lot of questions, like can the AI differentiate between how much audiences want to see a Tom Holland and Zendaya romcom, vs a sci fi film starring the two. I don't mean cross indexing data about the popularity of romcoms this year with the popularity of Zendaya this year, I mean the much harder to pin down question of what types of roles audiences would be excited to see them play.

Perhaps you guys have really done something amazing here, I'm just skeptical, because in example the last AI business that came through here was just a dumb cash grab, soliciting huge investment from easily deceived investors who still think AI is a type of magic. People have been showing a tendency to just slap on any AI functionality, hype investors with buzzwords, collect millions up front, and then provide placebo grade information in return. In the case I'm referring to, money was raised to create an AI script writing platform, but instead of actually building that, they just slaved a GPT3 clone to manufacture random text and parrot formatting, which is a super dumb and useless function, the main purpose of which was to enable fundraisers to say they had a script bot, and raise millions, before anyone realized that they had just sold something utterly useless for millions of dollars of up front cash.

Here's my test question for you. Would your AI recommend that Schindler's list be funded, if given a choice between that and Gigli? One is a long and depressing film about innocent children being murdered, mostly in black and white, the other is about two popular celebrities, one with current hits on the billboard 100. Which film would your AI pick to be financed? Sorry if any of this sounds confrontational, that's not my intent, it's just that I have some pretty serious questions about how genuinely beneficial some of these AI systems will be, even if they work exactly as intended.

My biggest concern would be enslaving all creators to an algorithm, where the goal shifted from being creative or pleasing the audience, to pleasing the algorithm. If you don't understand what has happened in these situations, you might find this video of interest. To all my fellow content creators, here is who is beating you into poverty, and how, and it's all based around pleasing a robot. Who is the indie filmmaker that makes 250k a week now that the algorithm decides what we watch? It's the dumbest, least talented, sex fetish hawking, child pandering, cover song playing people in the world, with a robot sending them more money than Stravinsky or Kubrick ever made. Anyway, be careful with your robots, because there's significant evidence that reducing creativity to an automated state produces absolutely terrible results. This video is really more terrifying than educational. Someone wrote a symphony this week, and youtube's robot basically told them cleanup in isle 3, and then sent millions of dollars to people stepping on balloons barefoot until they pop. Now we have a teenager playing a bad cover of "Achy Breaky heart" and the bot gave them enough money to build a skyscraper, and somewhere, there's a person smart enough to build a skyscraper, and the bot sent them enough money to do a cover of "Achy Breaky Heart" Pretty stupid, and that's the best funded algorithm on the planet.

Click to expand...
Hi Nate,

I really appreciate the thought that's gone into this post, and - in particular - really enjoyed the TikTok content factory investigation share.

Im going to go away and think about your 'test question' : "Would your AI recommend that Schindler's list be funded, if given a choice between that and Gigli?". I think the framing of this question doesn't match the way we ask questions of the data. Questions we might ask of the data might be:
  • Given a package and a range of editorially/thematically similar titles what kinds of consumption have they seen? And what's the implied value in the context of consumption (streamer, regional territory)
The filmmaker should then be a better place to make their bet: given that implied value range/confidence, what might be a reasonable production/contributor budget to make a return?

The AI part for us is pulling the signal from the noise in a huge amount of historical market/audience data to inform filmmaker intuition not to 'make the bet' for them

Thanks again, would love to keep the conversation going
 
Thanks, that's a fair reply.

An adjacent concept that I thought might be worthwhile to share -

Setting a marketing budget, or budget ratio, is something that seems to be a major point of failure for indie filmmakers specifically. Considering that your AI would likely be aggregating and organizing that data during the meta process, could you potentially augment your software to intelligently suggest an optimized marketing budget for each project taken into consideration. These numbers aren't really affected by many of the complications and uncertainties I described when it comes to the creative end, and you might be able to provide some pretty solid and actionable information on that end. Films need to be creative and new, marketing budgets do not, if you take my meaning. I could see an AI performing quite well at that task, and it should just be a short hop from where you are already at.

Also, improvement in this area would be almost a given. For some unknowable reason, almost every indie filmmaker I've ever met has arbitrarily decided on a 0 dollar marketing budget, so maybe all it would take to make them rethink that position would be a readout that said "optimal results for a Romcom with 2 known stars going direct to streaming can be achieved with a 37.5% budget share going to advertising on average" In fact, if you coupled that with an input field that allowed users to see loose predictions of revenue based on inputs of budgeting distribution aspects like this, that would likely be effective in convincing people to stop making 2 million dollar films and trying to advertise them on the director's facebook page"

In short, there is plenty of complicated math to work out here, but there's also some low hanging fruit in terms of practical statistics your machine can infer, that would be generally beneficial without necessarily being difficult for the AI to arrive at.
 
ben.ritchie said:
The AI part for us is pulling the signal from the noise in a huge amount of historical market/audience data to inform filmmaker intuition not to 'make the bet' for them

Thanks again, would love to keep the conversation going
Click to expand...
And thank you for answering questions.

My issue is just this: a huge amount of historical market/audience data cannot see
what hasn't been done before. Thus my "Night of the Living Dead" example. No
way would a huge amount of historical market/audience data from 1938 to 1968
have informed the filmmakers that their movie would have a lifetime value. That movie
was not a good bet at all.

I do see some value to what you are doing. I fear could suck the unknown, creative,
risk taking out of the process.
 
You must log in or register to reply here.
Back
Top