predicting hit movies, boon or bane? at DVinfo.net

Dennis Khaye · October 17th, 2006, 12:02 PM

October 16th's New Yorker has an article in it called "The Formula" by Malcolm Gladwell. The log line is "What if you built a machine to predict hit movies?". Later in the article it becomes clear that all they need is a copy of the script in order to do it.

I can't/won't give you the whole article here but let me say it is very well written and I urge you to grab a copy.

They already predict hits in the music business. Check out a little consulting company in New York city named Platinum Blue. They've written a program that utilizes "spectral deconvolution software" that predicts if a song will be a hit or not before they release it. This is how Norah Jones got discovered.

The article tells the story of a group of men who do the same thing for movies. The two front men for the project are Dick Copaken and Nick Meaney. Their company is called Epagogix.

Take an arm full of scripts that passed the read test and pay to have them analyzed by Epagogix then throw away the trash and make the hits. They also claim that the software will tell you where a story is lacking and how to fix it. You'll still need a good writer to make it good but at least you know what direction to go in. All this from just a script.

My question to you is, as a writer, producer, director or studio exec, boon or bane?

Roger Rosales · October 17th, 2006, 03:26 PM

Bane.

I think it's complete BS. It's basicly relying on a few individuals certain taste and a program with "AI" to tell us what's good and what isn't. This is probably boon for hollywood as most of the movies in recent years (slowly rising) have been no more than just mindless drivel. Either bad plot, mindless action, bad drama, over done drama, trying to hard to be "artsy", etc...and this program will continue to cater to those movies and audience. American moviegoers have little to no mentality (at least the younger generation). I say this because I'm part of it. Growing up, movies that truly are inspiring, my contemporaries found confusing or just "gay".

Sure they can probably predict a hit movie, but will it be any good?

My 2 cents.

Mathieu Ghekiere · October 17th, 2006, 04:40 PM

Movies shouldn't be made on such a mechanical way, it's still art!

I think there would have been many scripts that such a program would have put in the 'trash' can, that later have been real film classics.
The director, actors, composer, director of photography, editor,... all make the movie, not only the script!
Film is an art and it should stay organic, and not become mechanical fabric work like that.

Steven Fokkinga · October 17th, 2006, 05:00 PM

This reminds me of an excellent Roald Dahl story about the invention of a writing machine that would always produce bestsellers. It was in "someone like you".

Glenn Chan · October 17th, 2006, 10:25 PM

Bane.

In my opinion, this kind of thing would be difficult for a machine to do if it doesn't accept external input.

Consider this:
Tastes and liking for movies will fluctuate depending on time and context. In Shakespeare's time, he was not considered as good as Ben Johnson. Nowadays, people don't really know who Ben Johnson is, while Shakespeare is considered a master (in terms of artistic merit at least).

As well, star power plays a huge role in commercial success- this cannot be measured by reading a script. 80% of a Hollywood budget may go to the stars (i.e. directors, actors, actresses).

2- If you look at playing chess, a computer may eventually supercede human ability. However, evaluating a movie for potential success may be a different matter. You have to somehow break down the variables into algorithms and heuristics. Some of these variables are hard to measure and elusive. As well, the variables are also constantly changing (so you need to re-design if new variables are identify) and data input may be difficult (the program may need human assistance, which may be biased).

Some values are hard to quantify... i.e. star power, concepts and ideas presented in scripts, size of plot holes, novelty, etc. However, you could take a random sample of the target market and ask them whether the script is novel, contrived, etc. This lets you quantify certain things. This could be expensive though (although maybe not for a $200 million movie).

3- Computers are good at factor analysis... if you collect enough data, you can figure out the strongest correlations in movie success. As well, you could figure out your deviation and error and such... which will let you know how useful your results are.

Marcus Marchesseault · October 18th, 2006, 03:49 PM

general script fault at line 2384

writer's block is futile

all your script are belong to us

the policeman's beard is half constructed

end of line

Nick Hiltgen · October 23rd, 2006, 01:19 PM

"all your script are belong to us"

genius.

Marcus Marchesseault · October 23rd, 2006, 04:24 PM

I am not a genius. I only play one in independant shorts! :)

"All your script are belong to us" was easy. I like the following references better:

"the policeman's beard is half constructed"

Paste and search that for a laugh. I think the wikipedia explanation is the best.

"end of line" Is of course how the great MCP ends all of his dictates. You all know about the Master Control Program, don't you? It is by his will that we converse in this manner. MCP also has a non-fictional history.

October 17th, 2006, 12:02 PM	#1
Dennis Khaye Major Player Join Date: Oct 2005 Location: Grand Rapids, MI Posts: 255	predicting hit movies, boon or bane? October 16th's New Yorker has an article in it called "The Formula" by Malcolm Gladwell. The log line is "What if you built a machine to predict hit movies?". Later in the article it becomes clear that all they need is a copy of the script in order to do it. I can't/won't give you the whole article here but let me say it is very well written and I urge you to grab a copy. They already predict hits in the music business. Check out a little consulting company in New York city named Platinum Blue. They've written a program that utilizes "spectral deconvolution software" that predicts if a song will be a hit or not before they release it. This is how Norah Jones got discovered. The article tells the story of a group of men who do the same thing for movies. The two front men for the project are Dick Copaken and Nick Meaney. Their company is called Epagogix. Take an arm full of scripts that passed the read test and pay to have them analyzed by Epagogix then throw away the trash and make the hits. They also claim that the software will tell you where a story is lacking and how to fix it. You'll still need a good writer to make it good but at least you know what direction to go in. All this from just a script. My question to you is, as a writer, producer, director or studio exec, boon or bane?

October 17th, 2006, 03:26 PM	#2
Roger Rosales Major Player Join Date: Aug 2003 Location: Los Angeles, CA Posts: 254	Bane. I think it's complete BS. It's basicly relying on a few individuals certain taste and a program with "AI" to tell us what's good and what isn't. This is probably boon for hollywood as most of the movies in recent years (slowly rising) have been no more than just mindless drivel. Either bad plot, mindless action, bad drama, over done drama, trying to hard to be "artsy", etc...and this program will continue to cater to those movies and audience. American moviegoers have little to no mentality (at least the younger generation). I say this because I'm part of it. Growing up, movies that truly are inspiring, my contemporaries found confusing or just "gay". Sure they can probably predict a hit movie, but will it be any good? My 2 cents. __________________ -Roger Rosales www.rogerrosales.com

October 17th, 2006, 04:40 PM	#3
Mathieu Ghekiere Inner Circle Join Date: Jun 2004 Location: Belgium Posts: 2,195	Movies shouldn't be made on such a mechanical way, it's still art! I think there would have been many scripts that such a program would have put in the 'trash' can, that later have been real film classics. The director, actors, composer, director of photography, editor,... all make the movie, not only the script! Film is an art and it should stay organic, and not become mechanical fabric work like that. __________________ http://productionsofthe3rdkind.com http://pr3k.com

October 18th, 2006, 03:49 PM	#6
Marcus Marchesseault Trustee Join Date: Nov 2005 Location: Honolulu, HI Posts: 1,961	general script fault at line 2384 writer's block is futile all your script are belong to us the policeman's beard is half constructed end of line Last edited by Marcus Marchesseault; October 18th, 2006 at 04:36 PM.

October 23rd, 2006, 01:19 PM	#7
Nick Hiltgen Trustee Join Date: May 2003 Location: Atlanta GA Posts: 1,427	"all your script are belong to us" genius. __________________ I have a dream that one day canon will release a 35mm ef to xl adapter and I'll have iris control and a 35mm dof of all my ef lenses, and it will be awesome...

October 17th, 2006, 05:00 PM	#4
Steven Fokkinga Major Player Join Date: May 2004 Location: The Netherlands Posts: 204	This reminds me of an excellent Roald Dahl story about the invention of a writing machine that would always produce bestsellers. It was in "someone like you".

October 17th, 2006, 10:25 PM	#5
Glenn Chan Inner Circle Join Date: Jun 2003 Location: Toronto, Canada Posts: 4,750	Bane. In my opinion, this kind of thing would be difficult for a machine to do if it doesn't accept external input. Consider this: Tastes and liking for movies will fluctuate depending on time and context. In Shakespeare's time, he was not considered as good as Ben Johnson. Nowadays, people don't really know who Ben Johnson is, while Shakespeare is considered a master (in terms of artistic merit at least). As well, star power plays a huge role in commercial success- this cannot be measured by reading a script. 80% of a Hollywood budget may go to the stars (i.e. directors, actors, actresses). 2- If you look at playing chess, a computer may eventually supercede human ability. However, evaluating a movie for potential success may be a different matter. You have to somehow break down the variables into algorithms and heuristics. Some of these variables are hard to measure and elusive. As well, the variables are also constantly changing (so you need to re-design if new variables are identify) and data input may be difficult (the program may need human assistance, which may be biased). Some values are hard to quantify... i.e. star power, concepts and ideas presented in scripts, size of plot holes, novelty, etc. However, you could take a random sample of the target market and ask them whether the script is novel, contrived, etc. This lets you quantify certain things. This could be expensive though (although maybe not for a $200 million movie). 3- Computers are good at factor analysis... if you collect enough data, you can figure out the strongest correlations in movie success. As well, you could figure out your deviation and error and such... which will let you know how useful your results are.

October 23rd, 2006, 04:24 PM	#8
Marcus Marchesseault Trustee Join Date: Nov 2005 Location: Honolulu, HI Posts: 1,961	I am not a genius. I only play one in independant shorts! :) "All your script are belong to us" was easy. I like the following references better: "the policeman's beard is half constructed" Paste and search that for a laugh. I think the wikipedia explanation is the best. "end of line" Is of course how the great MCP ends all of his dictates. You all know about the Master Control Program, don't you? It is by his will that we converse in this manner. MCP also has a non-fictional history.