Creative kernels

July 2024

Have you ever heard musicians work on their songs, right when they are getting started?

You can often hear the melody or beat that got them excited. You can hear the initial creative kernel the rest of the song is written around.

Here is what I find fascinating and why I want a label for "creative kernels":

  • To an untrained ear like me, the creative kernel by itself doesn't stand out. Compared to the final song, it is often incomplete or meh.
  • The artist makes up everything else around that creative kernel with little regard for its quality. If the creative kernel is a melody, the lyrics can be gibberish. If creative kernel is a beat, the singing can be monotone.

Some conclusions from that:

  • Following that creative kernel helps you deal with the otherwise intimidating blank canvas.
  • The first version of what you make is always bad compared to the final version. Focusing on the creative kernel can help you decide if you are sufficiently excited about it to actually make the final version. And drop it if you aren't.
  • Knowing how creative kernels look like in the beginning (weak) can give you the confidence to work on your own. If it is ever going to be good or polished, it will be so after production.

None of these are novel observations but there is something very visceral about seeing it in action. I invite you to watch these well-documented examples and get the same chills I did:

  • The Beatles. The Get Back documentary is full of sections where you can see them get excited by a little piece of melody that ends up being The Best Song of All Time. They often sing with no lyrics or laughably bad ones (i.e. "She attracts me like a cauliflower").
  • Billie Eillish. There are hours of interviews of Billie Eillish and Finneas talking about their creative process. In most of them, they are semi-explicit about the creative kernel for the song
    • Therefore I am was made around a "super catchy hook".
    • everything i wanted was written to express an awful dream Billie had.
    • bad guy starts with a beat Billie made in Logic one of the first times she used it.

And I think these extend to everything creative, not just music. From the game design world:

  • Jonathan Blow. He explicitly talks about prototypes as a way to explore "an idea" and decide if it is worth working on. And in the Braid Anniversary commentary, he keeps referring to the idea that guides the entire game.
  • Marc Ten Bosch's Miegakure "explores the fourth dimension to make puzzles". It comes from the same school of game design as Braid but with a very different kernel.

Why "creative kernel"?

Most artists casually refer to "the idea behind" or "the thing that" but there are a few reason I insist on having a distinct label for this:

  • There a lot of ideas in any creative project. But the creative kernel is special.
  • The creative kernel helps judge the other ideas.
  • If the creator is not excited about the creative kernel, the project might as well be dead. For other ideas in the project, you can just cut them from the project.

"Creative kernel" is the best label I've got but sounds pretentious. Others that could work "seed", "anchor", "keystone". If you come up with something better, do send it to me.

The rest of the post looks at each example I gave above and tries to justify what I saw in the artists process or language to lead me to think they had a well-defined creative kernel.

The Beatles: Get Back

The documentary The Beatles: Get Back shows this non-stop. Here are some scenes that were uploaded to YouTube that showcase what I mean:

Paul "discovers" the song Get Back. Notice that, initially, he is not focused on any given melody. He tries a few (0:39 - 1:00) that never make it into the song. What he is really excited about is that energetic strumming in A → D. As the melodies change, he keeps strumming the same chords. And at 1:05 or so "he finds" a melody for the voice he likes. We can recognize it as the melody of the final song but it is far from complete or even good. At 2m0s he tries a little variation on that melody and stops.

When working on Something, John shows he also doesn't always care about lyrics:

Just say whatever comes in your head each time, "Attracts me like a cauliflower", until you get the word. "Attracts me like a pomegranate"

This works because the creative kernel they are working around is a melody and lyrics don't matter to that kernel. Or at least not right now. Right now, the lyrics might as well be scaffolding. You can see that George doesn't care that much either: he sings the rest of the song without lyrics. Paul doesn't care either when working on The Long and Winding Road.

Billie Eillish and Finneas

If you listen to them talk about their creative process, you'll notice they'll often refer to "the thing we started with" or "the thing we had".

Therefore I Am

Here is Finneas talking about writing Therefore I Am:

I wrote and recorded the chorus on my own voice probably a couple of keys lower than where we landed for Billie and just had it as a hook ... the chorus was super catchy

I just sat on it. I just had this "I'm not your friend" thing.

She thought it was infections and cool.

In this case, the creative kernel was a catchy hook that sat in his computers for a long time before becoming a song. You can see how for this particular song, the meaning behind the lyrics was not particularly important to the creative kernel:

everything i wanted

From the Song Exploder episode on everything i wanted, the initial creative kernel for that song is a nightmare1:

It’s a crazy dream.

In the actual dream when I jumped off there were fans, standing there at the bottom, and they just filmed me jump off and nobody did anything.

It is a serious matter and they take it seriously. From how they talk about it, you can tell that they cared about those lyrics in a way they don't care about other lyrics:

Finneas got mad because he didn’t wanna write about it, you know, he didn’t like to think of me, as in this headspace.

But in the previous example, Therefore I am, Finneas is very happy to change the lyrics right away if they sound better even if they change the meaning of the song completely:

The original line I'd written was "You think you need a man, I think therefore I am" but Billie said that it should be "You think that you're a man, I think therefore I am" which is way cooler and actually the internal rhyme structure of that is "you're" and "fore" which is way better anyway.

And then, we set to write the verses for that.

In other words: for one song they take the lyrics very seriously, for the other they don't.

But everything i wanted gets a second creative kernel. From 2:50 in this other video:

Finneas came up with "As long as I am here nobody can't hurt you". We decided to make it about other relationship and how we help each other. It just completely paved the way for the rest of the song.

And you can feel it in the song itself. The transition goes from 1:12 - 1:20 and the chorus (1:20 - 1:52) comes from that second creative kernel.

bad guy

The creative kernel for bad guy seems to be a beat pattern Billie made when learning Logic

When I got that pattern down, I was so pumped

We had that and then we sat around for.. almost a year.

From their language, you can tell the creative kernel was not "the chorus":

The chorus has no hook... and then we just did this (sings "chorus" synth melody)

And neither were the lyrics:

People want to hear that "duh", this is why I did it, and this is what it means, but I just thought it was fucking funny.

This song should not be taken seriously so I had to add a little stupiosity.

Jonathan Blow on prototyping videogames

Jonathan Blow has a talk on prototyping games. If you haven't watched it yet, go do that. It is much better than this post.

In the talk, he shows two prototypes. As he discusses them, he refers to two different thing:

  • The prototype. The artifact that he made that he can play with.
  • The "idea" behind the prototype. That idea is what I am referring to as creative kernel.

The prototypes succeeded as prototypes: they showed that the creative kernels they were exploring were not worth more time. The creative kernels failed.

In the Braid: Anniversary Edition commentary2, Jonathan talks in length about the original creative process that took him to what Braid became. He identifies the following initial creative kernel:

What if a game showed how quantum mechanics worked as part of the game mechanics?

But then, after implementing one of the ideas that comes from that, he finds that bidirectional time (aka rewind) is enough3 to have an entire game around it. The creative kernel thus becomes:

What puzzles can you solve with rewind that you can't solve in other ways?

And from there he makes everything else4. The level design, the mechanics, the art, the music, the story, all of it ties back to reinforce that creative kernel5.

The prototype helped (a) refine the creative kernel and (b) show that it was fruitful.


  1. From 2:45 or so in the other video: "We called the song Nightmare for a while before it was Everything I Wanted"
  2. You can listen to much of the Braid commentary as a podcast.
  3. The physics origin of the game doesn't quite end when Jonathan discovers rewind: World 5 is inspired by entanglement and World 6 is inspired by gravitational time dilation.
  4. There is one puzzle in the game that is explicitly not about rewind6. And guess what, that puzzle is the creative kernel for what became Jonathan's second game, The Witness).
  5. La intencionalidad que maneja es de locos. I should write an entire essay on that.