On Art, Information Theory, Compression, and Strategies for Investing in Artworks


How I think about art, and a proposal for how we might one day measure and quantify art, from the POV of computer science. Also: a novel strategy for making winning investments in the art world

Art and Use-value

I highly doubt this view is original, but I’ve never seen anyone else present it, and I feel like it’s the most practically useful way for me to view:

  1.  The purpose of art
  2.  What makes some artworks more successful (or better) than others

I’m less interested in giving a philosophical background for why someone might be interested these questions (there’s already enough written on that), so I’m going to jump right in.

I’ll present this as a very simple (and very informal) logical argument. I don’t think it’s ‘right’ in the sense that things are absolutely right or wrong (and honestly, I don’t care), but I think it’s an interesting and constructive way to view art (a subject I’ve always loved and been interested in). And moreover, as a student of computer science, I also believe this view offers a novel way to measure things we generally think of as intangible and subjective, such as ‘the quality of of an artwork’.

DISCLAIMER:  Except maybe in advertising agencies, the value of art has always been subjective, and it will hopefully remain that way–but I still wonder, maybe just as an intellectual exercise: is there a way that we might quantify art in terms of a specific use-value? I think there is. And details are below.


The Purpose of art
1) The purpose of art is to transmit to emotional information

How to judge the quality of art
2) If the purpose of art is to transmit emotional information, then artwork that transfers more emotional information can be said to be of ‘higher quality’
3) Since large artworks (think: a novel of 3,000 pages, or wall-sized painting) have more opportunity to transfer emotional information than small artworks (think: a 10 page short story, or 5x5inch painting) —> THEN: the best way to measure the quality of an artwork is not just the *amount* of emotional information it transfers, but how much emotional information it transfers given its size.

  • That is, what we really want to measure is emotional information density, emotional entropy.
  • And specifically: the more emotion that gets packed into an artwork, the better.
  • Here’s a graphic I made using Balsamiq. It’s crude and kind of sarcastic, but you get the point:

    Click for larger version

 Art as compression
4) In computer science packaging information into a smaller file format so that it can be more easily transferred is called compression. There are many compression algorithms, and some are better than others at different tasks. The goal of compression, very simply, is this: Package as much information as you can into the smallest file size possible. Then transfer it to another person. That person de-compresses the compressed file and gets ALL of the information, even though it was transferred in a much smaller package.

  • My suggestion is this: We can think of an artwork as the compressed file. We can think of the emotional content the artist is trying to convey as the compressed data. And we can think of the an artist’s technique or creative strategies as his compression algorithm. The better the algorithm, the more the data gets compressed, and the more information we can transfer in less space.

The role of artists
5) Following this analogy, I suggest that: the role of artists is to put as much emotion as possible into their package (the artwork) so that these emotions can be experienced across time and space. And the more emotional information an artist is able to transfer, the better the artwork. The goal of art, really is information transfer, a very specific kind of communication. We want to make others feel a certain way: We want to make others know that someone is there for them. That we all feel the same emotions sometimes. That there can be transcendence, that there can be an end to metaphysical sadness—this and so much more. All of it can be possible, and there’s a role for all of it.

What it means practically
6) For artists: Think very carefully about the emotional content you intend to transfer to others. If your goal is to make others feel specific emotions, then what emotions do you want to share, is there a sort of responsibility that comes with being a tremendously gifted artist? (I mean, I wouldn’t really want to paint things that made everyone feel desolation and sadness, rather than profundity, transcendence and stillness.)
7) For computer scientists: Using this heuristic, maybe it’s possible to evaluate a body of work from different historical artists and determine the ‘most artistic’ of them all. For instance: take all the paintings in a given museum, and ask visitors to rate the amount of emotion they experience while looking at them (say on a 1-100 scale). Then, normalize this data based on the size of the artwork, and see which painting has the highest ‘emotional entropy’: The most emotional content given its size. Is this the greatest, or ‘most artistic’ artwork in museum? Either way, it’s hard to argue that it’s not the most successful.

How to make money from it =)

(see, I wasn’t just BSing about use-value)

This gets even more interesting once we think: Can we tie emotional entropy to historic auction prices? If there’s a relationship, and we can establish that, couldn’t we use this relationship to identify the contemporary artworks (even from relatively young or unknown artists) that will one day sell for the most money? And if so, shouldn’t we start buying them all immediately? There’s a business idea there that’s free for the taking… I’m being somewhat sarcastic, but in all seriousness, I think it would be very interesting to examine this data and try, even as an experiment. So if you’re similarly interested, feel free to contact me.