Big Data's Big Date with the Movies

think it was stolen by the Russians. I always suspect the Russians. I think it's a good bad guy, the Russians. Um which is which is actually really appropriate because today

we are talking about movies. Yeah, we're gonna talk about scifically big data and the movies, both with the Oscars, which here in the United States they were just recently that that award ceremony was just recently held, and also just with movies in general, and how how big data is shaping the movies or at least or may shape or may misshape the movies as the case may be.

As I said last time when we recorded this OSCAR season is a time when you're gonna hear me getting on my high horse and and just allow me to ug please, man, I can't stand awards ceremonies. Something about the whole enterprise of it just seems so wrong and stupid to me. Like You're going to put these movies next to each other and say which one was better? What was better Selma or the Grand Budapest Hotel? And I'm like, why are you comparing them? Like what what

would even cause you to do that? And I think some things you can, like in the grand scheme of things, if you were to take all movies, Sure, there are movies we can point to and say these are really good movies, and there are movies that we can point to and say these are really bad movies. But when you get to the point where you're talking about contenders for something like the Oscars, presumably all of these films

are representative of of excellence in some manner. And once you get to that point, it's really a matter of subjectivity which one is quote unquote the best picture, right, but being nominated it really kind of is in a way, because I mean, how do you how do you judge how do you judge a killer comedy, like something that genuinely makes you laugh and you have a great time

and you truly enjoy it and everything's done well. Very serious historical drama about something tragic or or just or just a heartwarming story that tugs at your heart strings. I mean, if any of these things, if they're really effective at what they're supposed to do, then they are great movies. Right, So how can you tell what what which one is better than the other. But that's all kind of beside the point. And also I need to point out that what's really important in art is winning.

You know, it's funny that we're pointing out the problem here is the subjectivity that comes into evaluating things like movies. But the fact that their subjectivity doesn't mean you can't get nasty and just quantify it by brute force. Well, and that brings us back to our to our topic

big data. In the movie. It's actually kind of crazy because you do think, like when you think of something like art, you think of that as being a very human endeavor, right, something that is not necessarily quantifiable, something that has has an element that goes beyond what we can achieve with an artificial construct, That an artificial machine of some sort could not create or appreciate art the

way we can. Uh, And therefore you wouldn't expect to be able to use these machines to evaluate that art and then come up with a good guess as to which one deserves an award. However, you'd be wrong. This is the weird thing. But then in a way, it's it's more about figuring out how people think rather than the quality the quality of the art itself. That's I

think a distinction we have to make. So we're really talking here in this specific part about big data and the movies and how big data can help a sophisticated system predict which movies are most likely to win awards. Okay, so what do we mean when we say big data. Well, so we we generated a lot of information every single day right about everything. We generate it, we we analyze it, We come up with new figures based upon the analysis of the old figures that joins the data. Because everything

that we do online is in some way quantifiable. Even if no one is tracking it, probably someone's tracking it. That's probably being recorded somewhere. It's being anything useful or otherwise is being done with it. Is that's another question. Yeah, exactly. So we get this these mountains of data, and if you figure out a good way to analyze that data, you might be able to make something meaningful from it. Now, some of that data is already categorized in some way,

which makes it a little easier to analyze. Some of it is quote unquote loose data that doesn't have any categorization or any other meta analysis that attached to it. So it's still valuable, but it's harder to incorporate into a system or an algorithm. But there are a lot of companies that are rising up around the challenge of analyzing big data for the benefit of other entities. So,

for example, market analysis huge with big data. Right, So, if you're a company and you're looking to launch a product, you might you might end up consulting a big data analysis company to find out things like what sort of aesthetic a roach you need to take with your product, what sort of advertising strategy should you incorporate? Lots of different bits and pieces that would go into the determination of how you would go forward to give yourself the

best chance of success. Uh yeah, Or if you just want to place bets on Oscar winners and make some money off of your friends. Uh there's companies for that. Yeah, well, I mean I'm not encouraging gambling. Well, I guess what

I said was just encouraging. But well, well, let's say, for example, I've got a friend who throws an oscar party every year, and one of the things she does is she hands out ballots and you go and you pick your choices for who's going to win what, and then the person who has the most correct answers wins whatever the prize happens to me, And usually it's one of those things that says most soulless hack, usually that T shirt that sounds like a great T shirts opinion

brought to you. But so uh no, it's it's usually in our case, it's you really like you throw two bucks into the bedding pool, and whoever gets a wins the kiddie or whatever. Uh So, yeah, it's kind of a fun little exercise. Although I didn't participate this year, I had threatened that I was going to take the information from one of these companies and and put it to use. So we wanted to talk a little bit about the ways that big data has or that that marketing firms have used big data to try and do

things like predict oscar winners. Now, keep in mind, this was not for business purposes, right, It wasn't or a lot directly for business purposes. It's not like there are companies that are in business of predicting the oscars and that's how they make money. This is probably more just to show off what they can do exactly, and it's a fun way to say, like, hey, look, we're real

good at crunching numbers, right right. If if we get a certain number of our guess is correct, then you know that our methodology is really reliable, and therefore you should hire us to do whatever else we do. And I have a personal caveat about that, but I'll wait

until the end, all right. So the first one I wanted to talk about, the one that I first looked at when I was looking into this whole idea, is a company called far Site, which started making predictions in two thousand thirteen, and I was looking to see if they made any for this past year, but didn't find any information on that. Fourteen they did, and I think

I know why they might have stopped. In fourteen, they predicted they made six predictions right of the big categories in UM in the Oscars, like Best Picture, Best Director, Best Actor, Actress, that kind of thing. I think Supporting Actor and supporting Actress were the last two, and they got all but one correct in two be director, that's right,

that's right. And then in two fourteen they got all six correct, which is why I think they might have stopped, because why they were like, well, yah see, let's just let's just stop exactly why risk my risk a bad year? And and granted this is me guessing they may have had other very valid reasons. This is just a wild guess on my part. But at any rate, they did

get six out of six, which sounds pretty impressive. You know, I do wonder how would a company make a prediction like that, since these evaluations are supposed to be subjective. You know, it's hard to watch a an actor's performance in one scene, then watch a different actor perform a different scene and put a number on which one is more likely to win, right, And and if this were a pure experience where every single person in the academy was completely unaffected by anything on the outside of it,

and furthermore, the voting process was more straightforward right. If all of that were true, I think it'd be very

difficult for any algorithm to truly predict what would happen. However, we live in a world where this voting process is a little more complicated and it is very much wound up in the Hollywood politics, and Hollywood politics can include things like the buzz that a film gets, things that like scandals, can affect which movies or actors or directors may or may not win that year, Things that are outside of just the quality of the film or performance itself.

So ultimately, that's another reason why oscars feel a little silly, because you can never truly be sure that the award is going to be the best and maybe going to someone for reasons that are on top of the quality of their performance. And keep in mind, like we said, if if a film is up for nomination, chances are it's reached a certain level of quality. So it's not like these are necessarily bad movies, but maybe it's not the best movie or best performance. I'd like to point

to Return of the King. I'm a huge Lord of the Rings fan. Don't get me wrong, love it, but I get a feeling like a lot of the awards that went to Return of the King that year were because the previous entries of the Lord of the Ring series were largely overlooked by the oscars, and this was the last chance, I think, so, I think that really was part of it. One year, Leonardo DiCaprio may win an Oscar for some for not his best role, but people just kind of think, right, he like, why doesn't

even Oscar already? Yeah, this performance he did, uh, you know, and and the the reboot of the uh Ernest Goes to Camp series, we gotta give it to him for something. I would watch that movie. Yeah, that's kind of great. I was. I was literally just grasping at straws there. But I'm glad you like it. So anyway, farst sight, their algorithm actually took a lot of these other elements into consideration, these things like industry buzz stuff that again

that would seem really kind of weird to quantify. So they would identify trends and biases that were in the media about all of the different nominees. Uh. It took into account news events, and so it took all this stuff and they quantified it in a way that they have not completely revealed because of course they're not going to do that. Yeah they want, they want, Yeah, they want people to say, oh, well, you have the secret sauce, right,

that makes this work. So therefore you're not going to share the secret sauce because then it's not secret anymore, and then anyone could do it. So at any rate,

it worked really well. Now, some people would say, and by some people I mean like some media journalists people who cover this stuff for a living, would say that a lot of these approaches do very little, that than a little more than what pundits do on a regular basis, that essentially, this is the same stuff that the people who analyze this, like human beings who just watch and analyze this, they do the same thing. Yeah. Well, well, but the whole point is that, um, you know, you

don't have to be an industry expert. You can just hire a company like this to be your industry sure, And and they're able to take in way more data than a person could write. They were factoring in things like the betting odds in Vegas over which nominee was most likely to win, So, you know, taking all of that and making it makes sense in a streamlined fashion.

That's a real challenge even for someone who watches the industry closely, right, but for a machine that can crunch all that and way the different factors, look at previous OSCAR ceremonies and who won, and kind of look at the different criteria and say, which of these seems to matter the most. Let's build our algorithm to give those the most. Uh, wait for the predictions. And that's what's going to guide our choices. And it seemed to work.

Now they're not the only company that's in this. And I've got some other ones to mention for this that's past year. Uh that when we you know, finally got to see the Oscars in two thousand fifteen, uh, and kind of didn't watch who won. I was watching w w E fast Lane, actually, so I missed out the Oscars too. So hul Cogan won the Oscars. Hogan was not at ww E fast Like. Man, you're just uninformed about professional wrestling and uninformed about the Oscars. Joe, Come on,

what were you doing that Sunday. I'm pretty sure hul Cogan was was a wrestler at any rate. Some of the other companies that got into this have acknowledged that it's tricky to make predictions about who's going to win the Oscars. For one thing, you're working with a much smaller data sample than than you would with some of the other market analysis these firms would typically do. Right, So we are generally talking about data analysis and it's not even necessarily all that big of a data question.

Sometimes it's kind of medium data. Yeah. Yeah, if you're talking about trying to predict what six thousand, six hundred or so people are going to vote on, that's that's really not when we talk about big data. That's not that huge the grand scheme of things UM and hypothetically the members of the Academy are largely anonymous. That so it's you know, you can't you can't pull the individual people.

So one company UM called Exponential what that their approach was They wanted to get a group of web users together that they thought would reflect the academy's overall uh you know, uh philosophies and biases, that sort of thing. So they actually ended up they had something like thirty two thousand people that they looked at to create kind of a sample academy. And they even went so far as to say, oh, well, these people are really in this sample are really super liberal, and that doesn't reflect

the Academy, because the Academy is not super liberal. It's it's a little liberal, but not as liberal as these guys. So we can't use these guys as our predictor group. Like and they would even make say say things like people who really liked such and such politician would be most likely to vote for such and such movie. But because that doesn't reflect the Academy, we're going to ignore that.

And I'm just like, wow, okay, so you're they were finding weird correlations based upon philosophies and movie preferences at any rate. Once they managed to get their uh, their representative group that they felt reflected the Academy, they discovered that it was either going to be Birdman or the Imitation Game that would take home the best Picture. Oscar and Birdman did um, although pundits were also saying that Birdman was going to so take that with a grain

as salt. Another company called predict Wise analyzes information from production markets and award histories. It also predicted Birdman would win. But not all of them were so flawless. There was a company, there's a company called movie Graph and their approach was very different. They looked at the tonal elements

of the actual films. They looked at recent Oscar winners and the movies that won the Oscars, and said, what are the elements that that have won Oscars in the past few years, because that's indicated trends that people find to be really noteworthy and award worthy. And based upon that they said, um well, according to to that analysis, American Sniper is going to win. Totally did not win. Um So not every method works equally. You know, I haven't seen American Snipers, so I can't actually judge it.

But by the title, it does sound like one of those directed DVD Tom Bringer movies. Yeah, no comment. Uh So again the l A Times, which obviously you know, it's right there in the heart of the film industry.

Uh pointed out that many of these predictions were made by media journalists who had been just watching the the the awards ceremonies unfold over time and kind of just the trends that were going on in Hollywood and have suggested that perhaps while some of these stories, like like uh the earlier ones might seem really impressive to get six out of six predictions, correct, it's not necessarily the hard science that some of these companies might be presenting. Okay,

can I have a Lauren rent Time? Now? You can absolutely a Lauren rant about can we have the intry music and everything? All right, Lauren rent Time? All right? So I do think that it's really cool that we are making computers that are as smart as humans and making these predictions. I I do genuinely think that that is an impressive feat. However, a word about the Academy.

Al right, So, though yes, the Academy does keep its membership list private, several researchers have seted out a majority of its membership and found that it's over white, over seventy mail, and has a median age of over sixty years. With that kind of skew, Odd posit that it's perhaps not the hardest thing to determine the winners that these people are choosing, right, Yeah, Diversity is not one of

the Academy's many um uh facets. Lets say, um yeah, there there a lot of people have pointed out over you know, recently and actually you know, throughout the entire history the Academy. People have pointed out the flaws in the in the Academy. Uh hasn't stopped the Oscars from being portrayed as this incredibly glamorous and prestigious awards ceremony.

But I keep hoping that with the increased amount of attention and focus that we will start to see a better representation of a broader spectrum of people, so that films that are legitimately awesome but rarely get that kind of consideration are grouped in with the ones that you know you would as soon as you see the trailer is coming out this year exactly and involves a volleyball and whatever soon overcoming adversity and injury and that kind

of thing. A tragedy, personal tragedy. Gotta have that in there. I think it's generally a crime. The lack of films where presented at the Academy that have the word chainsaw in the title. I think they're really missing on the important leather face demographic there. There are a lot of people who are wearing human skin out there, and they're really not being represented. You know, speaking no, no, no, no, no, no. We're not segue yet, Joe, Joe, you are obsessed with this,

all right. So here's a little peek behind the curtain. Folks. Joe and I are working on on an upcoming episode of tech Stuff, and Joe has been pitching really, really hard that I should do a full text of episode on Chainsaws. So the fact that it also has spilled over into fourth thinking tells me that he's truly passionate about this. I need to reconsider my original. Heck, no,

Chainsaw technology, I'm curious more interesting than you think. No, I don't think about Chainsaws night and day, just all of two days. That's fair. Okay. So we've talked about how you can perhaps use quantitative analysis stated by computers to predict how humans would evaluate art and entertainment. But what about going at the art and entertainment from the other direction. What if we used computer aided quantitative analysis and maybe even some aspects resembling big data to determine

how we make the movies in the first place. So, so like talking about a computer writer, like someone's a computer writing a screenplay, Well, you know, obviously we can't do anything like that yet, but we've had whole episodes about that. Yeah, it's kind of unpleasant to realize how close we are to that in some ways. Now I want to give a little analogy here. If you work in creating content, you've heard of content, Yeah, creating content for the web, you will know that there are certain

things you can look at in your data analytics. You might look at Google Analytics for your website and come up with uh tips for how you create new content. So, for example, you might look at the thousand articles you have so far and say, Huh, whenever we do a list article, ones that start with an even number get this percentage fewer clicks than ones that start with an odd number. So maybe when we create new content that is list based, we should always start it with an

odd number. Or maybe we use this word in the title, and when when we use this word, it gets this percentage fewer clicks than when we use a synonym for the word that's this other word. You can get into really minute analysis. Oh yeah yeah. And and if that sounds a little bit too cold and calculating, I mean we we do at a certain point here. It has to works, uh, use that kind of information to guide

what kind of information we put out. Like for example, on one of our video shows, brain Stuff, we we look at we look at the videos that we watched the most often and we say once about animals are getting less attention than one's about the human body, So let's do more about the human body, because that's what people seem to like. But the kind of concerns that go into how you say title content for the web to have better traffic performance, that's a fairly small concern.

You're not like writing the article based on this. It's just you know, you're you're doing something about how you phrase the title or the pagination structure or something like that. It's really presentation. Yeah, you can apply the same thing to movies, and it's really up to you how far you want to go with this, with this you know, scrambling of elements. Yeah, it's so. First of all, it is not unusual and and the screenwriting world to have a writer turn in a screenplay. Uh, it's been attached

to a production and director has been attached. And then for the either the movie studio, the director or producer someone involved in the production to decide, you know, the screenplay needs to be punched up a bit. It can't. It's it's not quite where we need need it to be in order to make the movie we want to make. You ever get to the credits in a movie that really wasn't all that great, and then you see written by and they're like six names yelp, and there's probably

another six or seven that got left off for some reason. Right, yeah, this is this is where you get people doctoring a screenplay in order to try and uh and make it more appealing or more more marketable, whatever it is. You're just better, just more punchy. Times, sometimes it means removing an entire character from a story. You might look in

and say, well, this character superfluous. Let's remove them, give some of the lines to some of these other characters, cut the rest and move forward, or adding another character, like in Silent Hill when they realized that they were literally no men in the movie other than pyramid Head, and then Sean Bean happen to a wild ghost movie. He actually doesn't. It's like the one movie series that happens. No, he's just left in in this bleak, gray solitude forever. Alright.

Well that's well at any rates. So normally we assign humans the task of doctoring screenplays. However, there have been a couple of companies that have come out and suggested that they have some algorithms that can help analyze the screenplay. See what elements would really uh supposedly at any rate um uh do really well with audiences, which ones might fall flat? And therefore you could follow the very cold algorithm and doctor your screenplay so that it has the

best potential to be a big hit. Right, So we're talking about taking a process that used to be done more on the say the perhaps well informed intuitions of somebody in char charge. They look at a script and say this, and right, we need to make some changes. Now we can actually base this on data. Yeah, and you can say, well, we've analyzed seven thousand movies over the past, however, and and movies that had this element didn't do as well as movies that substituted it for

this other element. Therefore, we can recommend the following script changes. Yeah. Worldwide Motion Pictures Group UH introduced an algorithmic that the company said that it could evaluate a script based upon past films successes and failures. Were largely talking about box office success here, so not necessarily awards consideration, but or like rotten Tomatoes score. But yeah, right right, you know

they're literally looking at how many tickets were sold. That's the mark of success as far as they're concerned, not artistic success, nothing like that. It's specifically how many butts did you put in seats. We're not running a charity here. Shakespeare got to get paid son. Okay, So it could actually analyze script elements and have historically derived audience reaction

to those elements. So let's say that you're writing your screenplay and you want to incorporate a sweet old lady who dispenses pithy advice, and then you would check the algorithm to see how such characters had resonated with past audiences in various other movies. If this sounds horribly mechanical and uninspiring, you're pretty much right. I want to give

a few examples. For there was a New York Times piece from covering what this Worldwide Motion Pictures Group did, and they explained their services by saying, okay, here's an example in movies with demons. You remember a while back, demon movies were really big activity all, and they said that quote targeting demons unquote do better than quote summoned demons.

So if you have a scene where you've got characters who are using a weigia board or they draw a big pentagram on the ground and bring demons up out of how Those are not as good as demons that show up from out of nowhere and harass you of their own agency. Right, So they have they have chosen, They have chosen to torment you for reasons that the demon is aware of, but the character may or may

not be. They certainly weren't summoned. So if you write a horror screenplay where the characters get out the Wigi board to get in contact with the demon and that's how it all starts, these guys will tell you no, no, no, no, no, no no, you're gonna lose money on this. You've got to get rid of the Weiji board and just have the demons show up. Yeah, you might start to think, hey, this suggests that it doesn't matter how how good the writing is. It just matters which elements you put into

the movie. And there in lies one of the problems. I want to cover a few more examples they give. Also mentioned in the New York Times piece was that quote guardian heroes do better than quote cursed heroes. Because that Star Wars movie did so poorly. I'm so glad that they've put it this out. I guess it all depends on what how they are defining guardian versus curse. Okay,

all right, that's fair. This the same company was also covered in a piece I'm going to talk about in the second that was a marketplace piece on this type of business, and they said that, for example, in slasher films where there are killings, the killings should be random rather than motivated by a quote something rational like revenge your money. One last one the Worldwide Motion Pictures Group people. They said that in a film that's based on a

true story, you've got to stick to the facts. They said that films that deviate from the facts pretty far in a quote based on a true story don't do as well. Because they give this explanation, the audience will google the story when they leave the theater. More than half the audience. They said, Yeah, I don't know where they get that, which is weird because you would think, like, when they're leaving the theater, they've already bought the ticket to see the movie. Don't matter one way. Well, but

they could tell their friends. Yea. So this company likes to emphasize, or let's say, liked because we have some news about how they end up. Yeah, they liked to emphasize that the decisions and the suggestions made for changing these screenplays are not made by machines. They don't like run the document of the screenplay through a computer program and the computer tells you what to do. Instead, it's

humans aided by data. So it's humans who are using data, and that is, you know, like obviously computer aided machine generated data, but the humans, humans are making the recommendation.

So I think of it kind of like imagine the earliest days of Pandora, where you would, uh, you know, they get a piece of music and they would start to analyze it and add all these meta tags to that music to identify the various components of it, and then uh, they would start to map the those to other pieces of music that shared at least a certain

number of those same qualities. And therefore, when you would build a Pandora radio station based off a particular song, you would get other songs that had elements at least of that first song in common. Uh. And I think of it similarly to that, like they're looking at all this accumulated information that they have stored in some form of database, but that it's a person who's actually going

through and reading the screenplay to identify them exactly. Um, and the company, Uh, this particular company isn't it doesn't isn't around anymore. It imploded in two thousand and fourteen. The company was in debt and controversially, a whole bunch of folks were either fired or left the company, and some of the big players immediately formed another company called C four and then bought up all the previous companies intellectual property assets. Um. So, would you say it was

a heroic implosion or a guardian employer cursed guarding? Yeah, I think it was a I think it was a random slasher implosion maybe, but at any rate, Yeah. So, the the idea behind this was just to check and see if there there are elements within a script that

could be tweaked. So it wasn't meant to be like a here are the twenty different things, uh the perfect movie script should have in it, because if you did that, you would probably end up with a movie very similar to the that music we talked about with the songs that were generated by all the elements people identified as either being wonderful or awful. Remember those? Um, we all agreed that the worst song was way more interesting than the best one. I still sing that song to myself

sometimes do all your shopping at Walmart. Uh yeah, so, but yeah, you wanted to talk about this other company to write Joe, Well, just that there was an other company who did a similar service that were called epagogics, and they style themselves with the capitalized decks at the end, I believe, But they do a similar kind of thing.

They have analysts that read a script and they've got a computer enhanced algorithm that sort of compares the script elements against data that they've collected about what what elements perform well, and they make recommendations. It's a similar idea. They can actually go through a script and give it a score that is aided by this data, and then they feed that score through their computer algorithm and it makes a prediction about how much money this script would

make at the box office given some kind of error bar. Obviously, they would claim that they they offer a very valuable service. It's hard to know the level of quality in the services these kinds of companies provide, especially because I'm I'm sure lots of customers who use their services are not super forthcoming about that fact. Yeah, and also I mean without being able to see what the product was before versus after, it's impossible for us to make any determination

at all right now. On one hand, I can see how, especially as time goes on, services like this could get more and more accurate and effective. Like I I don't doubt that you can probably get some pretty good predictions online just by analyzing script elements about what kinds of things will make money. But personally I hate this idea. I'm not a big fan either. Well, there's there's nothing saying that plugging in a little puzzle piece like this

making a change. Like for example, this epico jics company UM did a did a piece with with NPRS Marketplace a while back and told a story about one movie that they that they went in to help doctor and they recommended cutting down the roles of one of the one of the supporting characters. And the head you know, the person that they were talking to, laughed and said like, oh,

that's great. You just saved me like twelve million bucks because I realized that, uh, this big actress that I was going to get for this big supporting role, I don't need her anymore I can, I can hire someone cheaper, right because now now this, if this role is not going to be that prominent, then why would I spend the money on someone huge a huge name for that. But but you know, there's nothing saying that that role that would have gone to Angelina Jolie wouldn't be better

served by having a different actress there. Uh. And just to remind again, this is not a process that's new. Script doctoring has been around. We're just talking about applying new, newly quantitative methods to it. Instead of going on the intuitions of somebody in charge, we're going on data that we have, what we can compare a script. Right, This is this is you know the for you might think of script doctoring as something in the screenplay is just

not working. Something is it's clunky, or the story seems to get bogged down in the second act, something along those lines. Whereas in this approach, you're thinking you should change this, uh, this old lady character into a young man character, because young men male characters in this particular role end up selling x number more tickets than if you were to use an old lady character in this particular type of role within your story and that'specially if

they're played by an actor named Corey. And that's when it starts to that's when it really starts to feel like this is getting a little too weird and cold and calculated and less like art. Right, it's I don't know, like like to to too Devil's advocate and I love

making that noun a verb. Uh. You know, I would still argue that it's up to the the artists, even if they are working off of cold data, It's up to the artists to do what they will with that data and go like, oh, no, you're right, like like this young man character could could be an amazing character, and maybe that voice would be a really interesting voice. And and maybe the reason I don't know, I mean, you know, it's what you make of it, is all

I'm saying. I can see that. But I can also see as a writer, someone writing something and you create, you create a vision and you write it down and then having a producer come in and say, yeah, this character that you create, the strong young female character needs to be a strong male character, because yeah, and that

would really upset me. So I mean, these are the sort of things and also, like the other objections here, Again, this is looking at specific story elements, not necessarily the execution of those elements, but just whether they're they're or not right. So, I mean, obviously you can't tell by looking into script what kind of star power or or just presence the actors are going to bring, or just how well written the part is. Right, Yeah, you can't even tell about the I guess maybe the quality or

the dial of the dialogue. Yeah, if you if all you're doing is doing the meta data approach, saying, uh, it's you know, the demographic information of whatever element it is, or you know, you know this movie is set in Seattle, and film set in Seattle, blah blah blah, you should really set this movie, uh, you know over in Portland instead or something along those lines. And you'll be like, well,

you know, it's it seems so arbitrary. But also it's all based upon past data, and that suggests that you can't do anything original because it has to be weighed against things that have already happened. And it could be that there's a lot of really crappy movies that have been made, but you've created a really cool idea that could be a huge hit. But because all the other movies in that space have been crappy. You don't even get the chance. So the example I gave it's a

bad example. But let's say that you want write a dinosaur movie, and you know, let's imagine the Jurassic Park never happened, but uh, Tammy and the t Rex and Theodore Rex both happened, and they are terrible movies. So then you write a movie where there are humans and dinosaurs it together. And because you've got these these other pieces that are poisoning the well, the data comes back and says, there's no way the audiences don't respond well to this kind of movie. And then we never would

have gotten Carnosaur. Yeah, and that would have been a shame because that that dinosaur was adorably vicious. But yeah, the real the real point here is that it doesn't. It discourages people from trying to go in innovative directions because there's not data supporting whether or not that would succeed. And then producers are saying, well, I'm not gonna put money into something where I don't expect to get a

payback on my investment. Right, Well, this is a thing that's common in all kinds of businesses where people want a safe bet. Yeah. You know, in a lot of cases, the person at the top of a business, they're they're looking at a film as an investment. They have to spend money to make the film, and they'd rather be sure they're get going to get a pretty decent return on a film then take a risk. Might you might pay off big or it might be a total flop.

And movies are expensive, Yeah, And so they're looking at it as an investment. And and they're saying, well, you know, if if I make something that nobody has ever seen before and I can't compare it to things that have been successful in the past, I'm taking more of a risk. I'm putting more of my money on the line, and I can't be sure about what I'm going to get back. And that's I can understand that from a business perspective.

I can sympathize with that in terms of making an investment, but it is kind of sad that that results in fewer movies that take those risks that turn out to be the movies we love. I mean, almost all of the movies we really love are the kinds of movies that do something new. They take a risk they do something you've never seen before. I don't know how well movies like that would do when compared against the data

of past hits and Blockbuster Now. Fortunately, I think that while we we will continue to see the companies like this provide a service to movie studios, I also don't think that it's ultimately going to be the thing that defines the industry as a whole. So, you know, we were painting a lot of doom and gloom, but really, honestly, we're not thinking that this is going to happen all

the time. Always sure, Sure, And also I think there's a huge space to uh to come to some kind of happy medium with looking at this kind of data and and also allowing new ideas to blossom, like like like the way that Netflix models its original series. This is really cool. So Netflix, you know, they pay people to watch Netflix, and then those people what they're doing is they are they're essentially creating the tags, very much like the Pandora example I gave earlier, critically acclaimed movies

starring strong female characters that like coffee. Right, Yeah, but if you've ever really dug through Netflix where it has the very aus categories, sometimes those categories get really hysterical, like they can sometimes just the categories all the entertainment you need. You don't even need to watch a movie. You just look at what the category name is. But

it's yeah. This comes out of the fact that there are people who are tagging these movies and they say, oh, you know, this is related to these other films in these different ways, so you get you know, you can do an analysis. You can compare to films or two TV series with Netflix whatever against one another and see how how well they match up as far as those

meta tags go. And this is kind of how Netflix creates that recommendation engine, so that if you watch, let's say, for instance, I just watched an independent, uh foreign horror film last night, and so based upon that, I will probably get other recommendations of other independent foreign horror films, which is that's exactly what I want. So that's fantastic. But they're also using it to guide their decisions. Like

you said, Lauren, with their original series. Net produces its own films and TV shows, and by looking to see what people are watching, they get an idea of which elements people are most interested in, and then they can say all right, Let's develop a show that caters to this audience. Yeah, let's let's green light a show. And the terrific thing about Netflix's green lighting process is that they don't uh, let someone do a pilot and then

have that pilot sinker swim. That they have someone create an entire first season yea, when they release it all at once. Yeah, which is so giving to the to the creators. Sure, I totally agree. I'm just wondering that how come No matter what I watch on Netflix, it tells me that the thing I should watch next is teenage Mutant Ninja Turtles three Turtles in Time, because it knows you, because that is what you should watch next to Joe's Surprise. It's not just all the films that

have the word chainsaw in them. No, Actually, I think what it thinks I like is uh, violent mind bending thrillers with a strong female lead. That's actually a category that I get. Yeah, yeah, I occasionally get that as well. So another way we're seeing big data affect the film

industries through marketing. When a film gets released in one part of the world, they might see what types of approaches are working or not working, and therefore they can actually tweak how they market a film and other regions to try and maximize that marketing effect. You know, keep in mind, a film's budget is almost nearly matched by the marketing budget to kind of push that film out there.

So if you hear that a film cost like two million dollars to make, it probably was another hundred hundred fifty million to market. It's crazy how much money is going into this, and they of course wanted to be as effective as possible. For instance, I've seen uh previews for a film that made it look like it was kind of a comedy, and later on seeing previews for the same film that make it look like it's a you know, a nail biting thriller. I think, like, what

which movie am I going to actually see? What? It couldn't be a third third trailer will play up the romance, And there was one recently. I wish I could remember what film it was, but I remember when I saw the preview, I thought, this seems like a totally different movie than the previews I saw like four months ago, But I can't remember off the top of my head where it was so good story Jonathan Um. It's also

telling us more about how people consume content. So, uh, Netflix releasing all those episodes, like releasing a full season on one day, as opposed to having each episode come out week by week. It's there. You know. That's because Netflix has seen people will sit down and binge watch a show. That's how they like to consume stuff. They don't want to necessarily have to wait for a full

week for the next segment of that story. And so that's why Netflix, when they greenlit their series, they say, all right, we're gonna fund the whole thing, and then we're releasing it all at once and people can choose to watch it all at that same time more, they can parse it out however long, however much they want, or they can leave their houses sometimes. Yeah. Sorry, I say that, you know, having completely done the like seven

or eight episode benders. Yeah, when the Arrested Development season came out, I pretty much consumed it in two days. I haven't Orange is the New Black problem. Yeah, that's understandable. Um, So the question now is where's this going in the future. I mean, we have said that we're probably going to continue to see this sort of doctoring approach to at least get an idea of you know, is this script

actually marketable or is it not marketable? But beyond that, are we going to start seeing things like a computer actually write a story taking all these elements, And are we going to get like the most disjointed, weird story with all the fantastic elements crammed into it. Man, I kind of hope, so I would. I would go that sounds like a goofy wonderful thing. Sounds like being turtles

in time. You just see like superhero superheroes, robotic figures writing dinosaurs which may or may not also be super robotic. You know, I'm wondering how long it's going to take us to get to that thing we were worried about at the beginning, where we've actually pretty much got computers writing the scripts for us, and other computers playing the characters for us, and other computers watching it. People have

just completely been taken out of the equation. We're simply thrown on top of the furnace to fuel all the computers. Um went to that dark place, that's yeah, yeah, I mean that's pretty likely. But you know, before we get to that point when humans are still watching the movies and still acting in the movies, are we going to get to a point where, you know, it's easier if we just cut out the middleman, because we're getting to the point where there's too much data for these human

analysts to mess with. We just need to have a program that can generate a script that they know has all the right, you know, money making elements. It doesn't have any of these low scoring elements like the old lady who dispenses pithy advice. You know, can that that lady right? Don't even bother? Why waste your time with a draft that has that when you can just have

the program create a storyline for you. I could certainly see this being done as an experiment, like, yeah, the way that algorithm based poetry is sometimes beautiful, Yeah it's sometimes it's sometimes beautiful in a kind of just sort

of odd alien way. But and I imagine a film that would probably at least the first few incarnations of the sort of thing would be similar like one of those things where like this feels like someone who almost understands all humans thing but doesn't quite go the whole way.

Would make a movie, I don't know. I would want a double feature of like that, like you know, starring some like like good people who have good senses of humor about it, and you know, it could really play it up and I would totally see a double feature of like that, and like whatever Michael Bay is doing this. I've got so many comments and they all pretty much say that I'm guessing the Bay movie would be the one I'd hate. But uh, yeah, well, I'd certainly see

this as an experiment I'm talking about. Oh no, the whole industry. I don't see it becoming the norm, partially because if you are taking this approach where you're trying to make all of your decisions based upon past outcome popularity, then ultimately all your stories are going to end up being incredibly formulaic and similar to one another. And it

turns out you don't need computers for that. Our friend Michael Bay has really mastered this particular Yeah, I'm like, the terrible dystopia you guys are talking about is the one we are living in right now. Well, there's there was an episode, you know, And I know Joe has seen some of the re letter media videos um that they do a lot of film criticism, and they do it in in a very snarky way, but they do they make legitimate, you know, film criticism. Fun Yeah, they have

some smart things to say. The warning if you go check them out, they're not exactly family friendly, that's true. It is adult content, but yeah, but give it a listen. They are they are very insightful and very funny and

very inappropriate at times. But one of the things they did, and it was kind of as a gag as the fourth Transformer film was coming out, was they put all of the first three Transformer films on concurrently and sat down on a couch to watch three televisions as they all played out, and they saw that these three movies had the same beats, like within within thirty seconds of

each other. They were identical as far as big action sequence, stupid character moment, terrible comedy relief and yeah, and like not the same way that people say that Pink Floyd matches up with uh with a Wizard of Oppositive Yeah, but like actually yeah, like the way that if you play Nickelback songs at the same time, they have all the same changes. Yeah, very similar to that, very similar

to that. And it was actually pretty and and you hear them like they start laughing because they see, like, look, now, these two are in the same moment, because it wasn't always all three. I think at one point one of the film's got like they lagged behind because they ended up either pausing it or something excellently. But at then you rate they just they talked about how frequently they matched up, and it just showed that there was a

very formulaic approach. So this is something that's happening already. I just suggest that if we were to move to an area where computers were making these decisions and they were basically purely on past experience, we would see way more formulaic films. They would that would be the majority if that was like how they were being generated. And uh, because we haven't reached a point where we can create a computer that not only can draw from past experience,

but innovate from that create something more brand new. That is is paradigm shifting. I think we're a lot closer to a computer that can write a basic story than a computer that could actually generate a script with dialogue and stuff. And I think that those kind of like human textual elements would still have to be filled in by So this would be like the actual beats like this, this is what happens in this scene. This is what

happens in the scene. And then I can you imagine dialogue very easily a computer telling you, maybe even today, if somebody spent time working on creating this program, that you need to have these types of characters, you need to have these types of events in the movie, and they need to come in this order. And then then Diesel needs to race buy in a car super high speed. Um. There's there's a film that I wanted to bring up

that came out in ten called The Congress. I don't believe it's scene wide release yet, um, but I managed to catch it at the Atlanta Film Festival. Uh. It was starring Robin Wright and it was directed and written for screened by Ari Fullman, based on a novel by Stanis law Lem which I would like to nominate as the best name, um Stanis law anyway. Uh yeah, yeah. The Congress is about this near future wherein movie studios just upload actors basically and uh and then create films

by computer algorithm. And it's a really really beautiful piece, um that explores these kind of tropes of what happens to the human element when you've taken the human element out of art. I highly recommend getting ahold of it if you can, um, because it's it's it's just so beautiful and creative and impossible to imagine having come from a computer. So uh So, yeah, if you're if you're interested in this and interested in seeing a creative portrayal

of it, then yeah, I can definitely see. I mean, you guys, remember when the commercials were coming out where they were digitally uh inserting you know, dance whoever they want. Yeah, it's usually for vacuums. I remember those in particular or

um like Fred Astaire or Gene Kelly or something like that. Uh, And that raised a bunch of like people were starting to say, oh, are we moments away from being able to cast like the Dream film where you can get young Marlon Brando in a film, Uh, you know, with actors and actresses that would not have been alive or not have been participate right or at the peak of their performing potential or any of that at the same time you could put them all together. Of course, we're

still far away from that too. Uh. It will be interesting to see how this develops. I'm sure we will see a lot of experiments come out about this. I mean, this seems like the sort of thing that artificial intelligence labs would do as a means of kind of just a you know, just an experiment. See what happens if you can do this. Uh, I don't expect it to be necessarily, at least at the beginning, anymore successful than that, you know, the greatest song ever written in the worst

song ever written. Uh. Projects were which were again in interesting, but um not not something that you would necessarily want on your on your NB three player. Although I do have the worst song online. Yeah, yeah, no, it's good. Um, I know I totally encourage someone to go make this if if any of you guys listening are our filmmakers, please like, like, I'll if you need a bad actor, I'll come acting in this experiment. That sounds beautiful. Yeah. I only agree to do it if I can be

the best boy. I don't want to be the second best boy. I don't want to be the worst boy. I got to be the best boy. Jonathan, you're my key grip like, oh you're so close, You're so close. Uh. Before we make tons of other jokes about gaffers and uh and and other people who are absolutely key to getting a video or film or television shoot done. Uh, let's wrap some of them are so key it's in their name. It is. Let's wrap this up. So you know,

this was a fun time, big to talk about. And you know, we've been doing a lot of episodes that have been um inspired by listeners who have written in, and we're looking for more of that because we love hearing from you. Guys. We've got We've received so many cool suggestions. We can't wait to tackle them all. But

don't let that stop you send in more. And you can do that by sending us an email the addresses f W Thinking at how Stuff Works dot com, or drop us a line on Facebook, Twitter or Google Plus. At Twitter and Google Plus, we are f W Thinking. Just search fw Thinking in Facebook. We'll pop right up and leave us a message and we will talk to you again really soon. For more on this topic in the future of technology, visit forward Thinking dot com, brought to you by Toyota. Let's go Places,