Creativity is a P = NP problem

[jimmylegss]

If I tell you to write a brilliant piece of music, even if you are a good musician, you cannot do that on command. You can only kind of do it by accident, if you have special talents (and then maybe only once!).

But the thing is, the moment you have created it, we can almost all judge wether it is brilliant or not. That is really easy. But to actually create it in the first place is incredibly hard.

So if we ever solve this, that also means we should be able to create brilliant art on command?

 

[jedishrfu]

Actually I think you mean the famous millennial problem of P=NP:

https://en.wikipedia.org/wiki/P_versus_NP_problem

There have been attempts to create music by computer

http://en.wikipedia.org/wiki/Computer_music

But you may have a point in that solving P=NP means that a revolution in computing may occur and that computers would then be able to match our creativity somehow although this is pure speculation at this point.

Another interesting development was the program that could extract the equations of motion from the data of a swinging compound pendulum. It worked really well but they couldn't explain the "logic" it used . It was based on a genetic algorithm that tried various combinations of functions looking for the optimal fit. These types of algorithms are used in P=NP type problems to get an optimal but not the ultimate solution.

http://en.wikipedia.org/wiki/Genetic_algorithm

For me, I always felt that creativity came from a faulty memory where you tried to reconstruct what you forgot and in the process created something new and unique.

 

[billy_joule]

If I tell you to write a brilliant piece of music, even if you are a good musician, you cannot do that on command. You can only kind of do it by accident, if you have special talents (and then maybe only once!).

That is not true at all. Writing brilliant music on command is the job description of professional composers & jazz musicians.
Great Song writing & improvisation are skills learned from practice, certainly not once in a life time accidents!

 

[JorisL]

That is not true at all. Writing brilliant music on command is the job description of professional composers & jazz musicians.
Great Song writing & improvisation are skills learned from practice, certainly not once in a life time accidents!

That's great song writing, not brilliant song writing imo.
If the entire industry writes brilliant music, it is no longer brilliant but average.

Brilliant goes above and (way) beyond average music in my opinion. Although this might just be a case of differing definitions, I do however believe that's the brilliance alluded to in the OP.

 

[Vanadium 50]

Billy_joule is right. The whole premise is flawed.

 

[Vanadium 50]

Brilliant goes above and (way) beyond average music in my opinion.

And the fraction of people who are able to work at this level is not large. How many garage bands are there to one Gioachino Rossini?

 

[JorisL]

And the fraction of people who are able to work at this level is not large. How many garage bands are there to one Gioachino Rossini?

No idea, but different people look at genius differently. I voiced my views.

Furthermore, the exact statement of P=NP and this subjectiveness mean we cannot say much about the premise, I believe. (I might not be creative at the moment)

 

[certainly]

Perhaps this Vsauce video is relevant. (He talks around 7.5 minutes in of how humans tend to like certain patterns of music more than others.) :-

 

[DennisN]

I agree with billy_joule above.

If I tell you to write a brilliant piece of music, even if you are a good musician, you cannot do that on command. You can only kind of do it by accident, if you have special talents (and then maybe only once!.

There are many examples in history which speaks against this assertion. E.g. there are countless famous soundtracks (for e.g. movies and television) which have been commissioned. And here are four very famous and highly regarded works by classical composers, which were commissioned:

The Requiem Mass in D minor (K. 626) by Wolfgang Amadeus Mozart was composed in Vienna in 1791 and left unfinished at the composer's death on December 5. A completion dated 1792 by Franz Xaver Süssmayr was delivered to Count Franz von Walsegg, who had anonymously commissioned the piece for a Requiem Mass to commemorate the February 14 anniversary of his wife's death. (link)

It is a nice piece of music in my opinion.

The opera was commissioned as a result of the overwhelming success of Mozart's trip to Prague in January and February of 1787. (link)

The Philharmonic Society of London originally commissioned the symphony in 1817. The main composition work was done between autumn 1822 and the completion of the autograph in February 1824. (link)

After the success of The Sleeping Beauty in 1890, Ivan Vsevolozhsky, the director of the Imperial Theatres, commissioned Tchaikovsky to compose a double-bill program featuring both an opera and a ballet. The opera would be Iolanta. For the ballet, Tchaikovsky would again join forces with Marius Petipa, with whom he had collaborated on The Sleeping Beauty. (link)

But to actually create it in the first place is incredibly hard.

It may be, depending on the situation. But if the person who creates it has got talent and experience, it's a lot more easy to do.

 

[Ryan_m_b]

But the thing is, the moment you have created it, we can almost all judge wether it is brilliant or not. That is really easy. But to actually create it in the first place is incredibly hard.

Others have jumped on the premise that creativity is possible on command (many, many professions require it so if it wasn't reliable society wouldn't work) but I'd like to address this point instead. I'd argue that for art no we can't most all judge whether it's brilliant or not. I like music as much as the next guy but I don't really know anything about the theory, I can't play any instruments and don't religiously follow any musicians or movements but I do have several friends that do all of these. I'm always struck by conversations between them about the relative brilliance of certain songs that to me, hardly sound different. I'll find them brilliant if they make me feel good and occasionally if there's some musical performance that looks like it would be really complicated but that's all. My friends on the other hand immediately delve deep into the technique, the theory and everything else about the song to decide if it's brilliant or not.

So no, we can't all judge quickly because we don't even judge the same way. Some people judge how good music is by the skill of the performance, others by the musical theory, others simply by how it makes them feel and others still by how commercially successful the piece is.

Having said all that I do find the idea of this being an N=NP problem interesting, as I understand it (and I'm not a computer/maths expert by any means) the problem is whether or not answers that can be quickly checked could in theory be quickly created. When applied to music specifically this would seem to imply that there is a way of determining if a musical piece will be brilliant before it is made. That falls into the problem of people using different metrics, some of them subjective, to determine brilliance.

 

[jedishrfu]

I remember a song controversy a few years ago, where someone wrote a haunting piece of music "Swimming to the Other Side" by Pat Humphries:



and by Lui Collins:



and people wondered why it was so beautiful until Lui mentioned it had the same rhythmic structure as Pachebel's canon:



More discussion here at NPR:

http://sandbox.npr.org/programs/atc/features/2002/may/humphries/index.html

Recently, a survey of modern patents and inventions noticed a pattern of small refinement happening instead of great leaps of creativity to new inventions.

Anyway its a common pattern, a great idea is created or discovered and then small or not so small innovations on the idea to produce the present day invention, theory, artwork or song...

How to score randomly generated ideas from a computer using say genetic algorithms that cycle through all combinations is one problem. The other is adding a new thing to the set of combinations that is totally unlike the other things is where creativity may lie.

Sometimes we say these new "things" are paradigm shifts.

So the P=NP debate may really center on finding the right combination of things to create something new and not so much on the paradigm shifts that we see from time to time.

 

[DennisN]

So no, we can't all judge quickly because we don't even judge the same way. Some people judge how good music is by the skill of the performance, others by the musical theory, others simply by how it makes them feel and others still by how commercially successful the piece is.

Well said, and I agree.

After reading the post by jedishrfu above, I remembered a couple of songs which are said to have been partly inspired by classical pieces. It has been said that the beat and structure of "White Rabbit" (Jefferson Airplane) was partly inspired by the beat and the structure of "Boléro" (Ravel) (but I can't say I'm perfectly sure of it). Another example is "Every Breath You Take" (The Police) which is said to be partly inspired by Béla Bartók.

Here's another fun clip:
Axis of Awesome - 4 Chord Song(s)

 

[jimmylegss]

What about other creative pieces? Like a movie or visual art?

Reason I made this thread was this clip:

 

[billy_joule]

What about other creative pieces? Like a movie or visual art?

What about them?
Your original premise is flawed for all creative processes, not just music.

 

[zoobyshoe]

If the entire industry writes brilliant music, it is no longer brilliant but average.

Brilliant goes above and (way) beyond average music in my opinion. Although this might just be a case of differing definitions, I do however believe that's the brilliance alluded to in the OP.

I think this is true. We reserve the word "brilliant" for that which has exceeded the average 'luminosity'. To call something "brilliant" we require it demonstrate a leap forward, or up, from the status quo.

Some professionals can deliver a consistently high quality that is extremely impressive, but I think, by definition, their output is not "brilliant." That is: I think by "brilliant" the OP was referring to creative output that is outside the status quo, which is an unexpected and non-predictable leap above the status quo. We recognize it as such when we see it but couldn't have produced it ourselves.

 

[jedishrfu]

Even good movies can be constructed using known recipes as outlined in the Save The Cat book by Blake Snyder. He's developed ten movie categories and described the 15 critical beats that must happen in a film and has provided examples of movies that fit each category.

What's remarkable is that two wildly different movies may I fact be the same movie and you woudnt know it until someone points it out. One example was the Monsters Inc movie and the first Matrix movie. They follow the same basic plot structure and beat rhythm.

Following his method it might be possible to build a GA program that could construct a successful movie plot, the beats and maybe even the first draft script with characters and dialog. It would be a fun AI project and very difficult to do I suspect.

Where brilliance might come in is when someone develops a great movie in a totally new category beyond the ability to put things together like most movies today.

 

[jimmylegss]

What about them?
Your original premise is flawed for all creative processes, not just music.

Your premise is that brilliant movies or pieces of music can be produced on demand, but that is not true. Just solid pieces can be produced on command like other posters said. You need to know certain rules of movie making, and hten you can produce a solid movie that is watchable. Or just a catchy piece of music that is enjoyable to listen to because it follows certain mathematical rules.

But I am talking about mindblowing classics here. The godfathers, pulp fiction, first matrix movie, Goodfellas of this world. Often these classics also follow certain rules, but they also add something new, or break current rules in certain ways/ order. Even the masters like Stanly Kubrick did not produce only classics.

If this was purely formulaic, why are we not consistently producing mindblowing art? We know the techniques, yet few can really produce great works. Often following exactly the formula is watchable, but then afterwards we forget them quickly and complain that they are just that, formulaic.

So the formula alone is important, but it is also very important that this formula is broken somewhere in very specific ways, to produce a masterpiece.

 

[Evo]

But I am talking about mindblowing classics here. The godfathers, pulp fiction, first matrix movie, Goodfellas of this world. Often these classics also follow certain rules, but they also add something new, or break current rules in certain ways/ order. Even the masters like Stanly Kubrick did not produce only classics.

And many people can't stand those movies. I am not a fan. It's all subjective. I also agree with billy_joule.

 

[DennisN]

"Brilliant" is a subjective and imprecise term. And I also think it's good to remember that there are trends in arts like there are trends in fashion.

Who can ever start to create anything, and knowing for sure that the work will be brilliant, meaning widely regarded as brilliant by others? A person can only try to do his best, and trust his/her talents and mood to lead the way.

The way I think it works (speaking from my own previous experiences, and watching how others work) is that you start creating something when you are inspired and think that you can create something out of whatever it is you are creating, regardless of if it is a creation for your own pleasure or something bigger. You evaluate your work and progress during the creation process (and maybe ask for input from others), you modify, delete and rewrite things etc. If you're not satisified how it is going, you maybe scrap it, recycle it or archive it for later.

If the creation gets finished and is complete, you can personally be more or less satisified with what you've done, but it's not really up to you to say if it is brilliant or not. That's for others to judge after your creation has been "released", i.e. displayed, read, heard by others.

That is how I see it.

 

[Ryan_m_b]

I think a key point here is that the "brilliance" of anything, aside from being subjective, is contextual. N=NP requires that checking a solution is trivial. But checking to see if a work of art is brilliant is not trivial unless you use very stringent metrics.

Many films were widely regarded as terrible when the came out only to gather huge respect and following decades later. Musicians who only never knew obscurity are later described as ahead of their time. Van Gogh purportedly died rejected and unappreciated, only to become widely regarded as one of the most influential artists of the modern era.

So brilliance itself is in many cases an external attribute and can't itself be derived from just the work itself.

 

[Intrastellar]

Isn't the OP's argument in this (below) sentence

But the thing is, the moment you have created it, we can almost all judge wether it is brilliant or not. That is really easy. But to actually create it in the first place is incredibly hard.

rather than this ?

If I tell you to write a brilliant piece of music, even if you are a good musician, you cannot do that on command. You can only kind of do it by accident, if you have special talents (and then maybe only once!).

Does the same not apply in creative proofs or solutions ? Understanding a brilliant solution is easier than coming up with it. Is that not the definition of a brilliant solution ? Does that not tell us that creativity is a P = NP problem ?

 

[Ryan_m_b]

See my post before (and many others in the thread). For many things checking if I is brilliant is not simple. It's a case of putting it out there in the world and potentially waiting decades before it's widely regarded as brilliant. And from that it becomes obvious that if the consensus of on time is no and the other is yes then brilliance is subjective.

 

[Intrastellar]

See my post before (and many others in the thread). For many things checking if I is brilliant is not simple. It's a case of putting it out there in the world and potentially waiting decades before it's widely regarded as brilliant. And from that it becomes obvious that if the consensus of on time is no and the other is yes then brilliance is subjective.

The above replies were talking about art or music or movies, which I do not have much experience about. I wanted to see whether others thought the same idea applies to mathematics or sciences/engineering.

 

[Evo]

The above replies were talking about art or music or movies, which I do not have much experience about. I wanted to see whether others thought the same idea applies to mathematics or sciences/engineering.

That's getting off topic from the OP, he wanted to discuss art and music.

 

[Intrastellar]

That's getting off topic from the OP, he wanted to discuss art and music.

To the best of my understanding, he wanted to discuss the relation of P = NP to creativity in general, and brought art and music as examples.
@jimmylegss: Can you please clarify that ?

 

[Evo]

To the best of my understanding, he wanted to discuss the relation of P = NP to creativity in general, and brought art and music as examples.
@jimmylegss: Can you please clarify that ?

I agree that he misused the problem applying it to art, music, films, etc... and your version is the correct one. But it's his thread and he chose to change what he wanted to discuss to forms of art.

To quote V50

The whole premise is flawed.

To discuss it correctly, a new thread should be created in the appropriate forum.

 

[Ryan_m_b]

It's also worth noting the discussion on the definition of brilliant. In science and engineering you can, in some cases but not all, easily test to see if the new solution is better than current ones. Whether or not it will be considered brilliant is again subjective and external. To put it another way it's impossible to tell if a new discovery/invention will win the Nobel prize looking just at that solution itself, to do that you have to examine the wider context (and ultimately just wait and see).

 

[zoobyshoe]

I agree that he misused the problem applying it to art, music, films, etc... and your version is the correct one. But it's his thread and he chose to change what he wanted to discuss to forms of art.

To quote V50

The whole premise is flawed.

To discuss it correctly, a new thread should be created in the appropriate forum.

The premise billy_joule and Vanadium 50 claim is flawed is the premise that brilliant art cannot be created on demand:

That is not true at all. Writing brilliant music on command is the job description of professional composers & jazz musicians.
Great Song writing & improvisation are skills learned from practice, certainly not once in a life time accidents!

Billy_joule is right. The whole premise is flawed.

They are not claiming that P=NP cannot be applied to creativity. They are claiming the premise that brilliant art cannot be created on demand is wrong.

 

[Evo]

The premise billy_joule and Vanadium 50 claim is flawed is the premise that brilliant art cannot be created on demand:They are not claiming that P=NP cannot be applied to creativity. They are claiming the premise that brilliant art cannot be created on demand is wrong.

Well, I will still go by what the problem is supposed to be. http://mathworld.wolfram.com/PVersusNPProblem.html

http://en.wikipedia.org/wiki/P_versus_NP_problem

 

[zoobyshoe]

Well, I will still go by what the problem is supposed to be. http://mathworld.wolfram.com/PVersusNPProblem.html

http://en.wikipedia.org/wiki/P_versus_NP_problem

I don't think anyone else in this thread has a problem with this. I assume that is because, in the background, everyone's thinking in terms of this being a potential AI problem, one that could be given to an "intelligent" computer program to try and solve someday: write a brilliant piece of music, or write a brilliant short story. Regardless, it doesn't seem to have bothered anyone else.

 

[Evo]

I don't think anyone else in this thread has a problem with this. I assume that is because, in the background, everyone's thinking in terms of this being a potential AI problem, one that could be given to an "intelligent" computer program to try and solve someday: write a brilliant piece of music, or write a brilliant short story. Regardless, it doesn't seem to have bothered anyone else.

Hmmm, I seem to be reading a different thread.

From the OP explaining this thread. https://www.physicsforums.com/threads/creativity-is-a-p-np-problem.813205/#post-5105787

 

[BobG]

I agree with billy_joule above.
There are many examples in history which speaks against this assertion. E.g. there are countless famous soundtracks (for e.g. movies and television) which have been commissioned. And here are four very famous and highly regarded works by classical composers, which were commissioned:

...

It may be, depending on the situation. But if the person who creates it has got talent and experience, it's a lot more easy to do.

The fact that famous works were commissioned doesn't prove the original premise was flawed.

Creating a masterpiece on demand is like Babe Ruth promising the kid in the hospital that he'd hit a home run for him in his next game. Babe Ruth only hit 60 home runs in his best season. There's no way he has a 100% chance of hitting a home run for that kid, even if he happened to know the pitcher he'd be facing the next day stunk.

But if you're going to pay someone to hit a home run for you, paying a person that hits a lot of home runs would be a good choice.

You're commissioning someone to create a piece of work for you because paying that particular person gives you a decent chance of getting a masterpiece in return. You're also taking your chances the work will be a dud.

Creating a masterpiece may not be quite as rare as the original poster suggests when you're talking about a great artist, but his premise that you can't guarantee ahead of time that a great artist's work will be a masterpiece is still valid.

 

[DennisN]

The fact that famous works were commissioned doesn't prove the original premise was flawed.

I misunderstood and/or underestimated what the OP meant by "brilliant" - the OP then started talking about "classics" and "masterpieces" in post #17, which made me write my next post #19 in which I wrote about "brilliant, meaning widely regarded as brilliant by others" etc...