At the most basic level, a digital media file (document, song, video, etc) is just a rather large pattern of 1's and 0's.

We can say that for any given piece of digital media, there corresponds a binary "fingerprint" unique to said media which defines it.

For example, an MP3 of a song will have a very specific binary pattern, that if altered, results in the file no longer being the same song.

Armed with this knowledge, it seems feasible to generate a binary pattern which results in a new (or copy of existing) media.

This leads one to conclude that all digital media already exists, but just has not been stored in format yet. All possible binary patterns must be discoverable or are already known.

Taking the thought experiment a little farther, it would seem possible to generate a pattern for content not yet produced, say an upcoming movie that's still in production.

Assuming every piece of media has a unique binary "fingerprint", how realistic is it to produce a binary pattern which is consumable in some media form?


A favicon (the small 16x16 icon that your browser displays next to the address of a website) is 6144 bits long. (Assuming 24-bit color.) Encryption algorithms that are unbreakable by today's standards use only 512 bits, because it is considered impossible to enumerate all combinations of a sequence of 512 bits.

What this should tell you is that absolutely any random bit pattern that you can think of is virtually guaranteed to be nothing but noise. The bit patterns that are meaningful to human beings are so few, and so very far apart from each other, that they are virtually non-existent, they are literally lost in the noise.

So, the chances of arbitrarily flipping bits and coming up with something even remotely meaningful are astronomically small.

And we have only touched 16-by-16-pixel favicons yet.

| improve this answer | |
  • So... we basically need more computational power. In a theoretical world with infinite computational power, we could generate media by discovering usable binary patterns... no? – SnakeDoc Jun 3 '15 at 21:13
  • 3
    Almost, but not really, because of the "discovering" part. We can already produce random bit patterns much faster than we can view them, so the production is not the bottleneck. Someone still has to view each generated media in order to determine whether it is meaningful, and it would take human beings a huge amount of time to simply do that. Imagine sifting through nothing but noise for centuries until one day you see something that resembles a face, maybe. – Mike Nakis Jun 3 '15 at 21:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.