How to usea spreadsheetfor creativity

AI companies churn decades, many decades, of human sense and non-sense to make their models.

All for the patterns contained in the data.

The patterns within human communication are what gives LLMs their impressive abilities. But why? What is so special about this thing we call pattern? And why was Mendeleev scrambling to find one before his publishing deadline?

At the time (early 1860s) there were 63 known elements (mostly discovered through breaking, burning, or electrocuting things).

Many scientists had tried imposing their way of organizing the elements, creating tables and diagrams that reflected their personal opinions.

But they failed to do one thing.

Teach.

Mendeleev was writing a textbook, and as his deadline neared, he still hadn't figured out how to represent all the elements together in a way that felt good for teaching. It's one thing to list all the elements for his readers, but he wanted some way that would help them grasp why these things were all considered elements. Why did they work as elements? What makes them part of the same category? Why?

At the time there was no established universal law of nature he could use for teaching. And so, whether or not the story with the cards is true: he kept arranging and reorganizing the elements until he found an organization that was ….

drum roll

… periodic.

Periodicity, for our purposes, is a repetition of relationships in specific intervals. Mendeleev had found that the elements were periodic based on the relationship between atomic weight and valency.

Why was this so mind blowing to the world?

Because it explained and predicted things.

Predicted! As in, like, the entire point of all of science. This is why patterns are so valuable. Mendeleev's periodic table not only predicted new elements (correctly!), but also predicted that previous measurements of elements were incorrect (later confirmed to be true).

Mendeleev, oh my, did he know how to make a table, or what?

But what does this have to do with creativity? What is this guide even about? Why does this author almost make me miss my LLM?

Here's the relevant part of the alleged story: Mendeleev found creativity through visual pattern recognition.

So, like Mendeleev, we're going to focus on being able to visually arrange and re-organize things until we spot a pattern. It requires no special knowledge and we'll rely entirely on our current hardware (our juicy brain) to do its thing.

But unlike Mendeleev, we're not looking for one singular natural law or pattern. We have the choice to make a pattern, break a pattern, guess a pattern etc. The possibilities are much larger for creativity.

It's a slightly different way of finding creativity. Instead of relying on a linear narrative or pure associative thinking (like brainstorming), we're going to rely heavily on pattern recognition.

It's still hard work. But it's fast. And if I may say so, bloody effective.

To do all of this, to hold all of our work, we might need some kind of sheet … maybe of the spreading variety.

What makes a spreadsheet so special?

Let's talk about llamas for a second. I ask this of my computer sometimes: "Hey, computer, could you show me a picture of a llama?"

Before the existence of AI models, this was a hard request for a computer to understand. A programmer had to figure out how to explain to a (pretty dumb) computer how to show you a picture of a llama.

The programmer would start with something simple.

"Computer, show me a brown dot please."

Our dumb computer perfectly understands this simple request; the programmer proceeds to feed it treats.

But not yet a llama though, is it?

The programmer asks for more dots and places them together.

No, still not a llama.

Now, annoyed, the programmer pulls out the secret technique known only to programmers: unnecessary pain.

They ask the computer to color 910 dots, all placed together in a rectangle, all with specific placements. That placement gives us something like this:

Ahh, maybe llama, I think. Very good, programmer!

The programmer used dots and rows and columns and placement to do all of this. This structure, of taking simple values and positioning them in a grid to create something more complex, is called a matrix.

Mathematicians and computer scientists use them all the time. In fact, if we instead took 5,134 rows and 3,423 columns and specified a color for all the 17,573,682 dots, we could get something like this:

Still, how does this make the spreadsheet special?

You know that magical property of math where you can use it pretty much everywhere? Like you can take the idea of "1 and 2 and 3" and use it to count both your sheep and the number of toes you have? It means we have the ability to give it meaning depending on where we use it.

Similarly, if you're like me and have math related migraines, this might initially look like a bunch of intense mathyness:

But a data scientist could do something beautiful with this: they could use it to mean your cats.

How did they get this rectangle of numbers to mean your cats? By giving the rows and columns specific meaning.

Still too many numbers, I think.

Luckily, data scientists only insist on using numbers for everything because computers have an easy time with them. For our purposes, we can use different scales; simpler ones like labels and feelings.

Ok, there's my point: A spreadsheet is a matrix. And just like we did with the llama pictures, we can use it to represent very complex things by organizing very simple things.

Let's say you have 5 friends (don't worry about me, I actually enjoy bouts of painful jealousy):

This is a spreadsheet of how your friends like each other:

By giving the rows and columns and cells specific meaning, and by using simple values to fill the cells one-by-one, we've represented the complexity of your friend group.

It allows us to ask and answer questions like 'who's most liked?', 'who's the most neutral', 'who has only a few but very strong friendships'?

Likewise, you could do the same thing to see how the parts in your computer interact with each other or how the characters in your novel relate to each other.

There's something useful happening here.

But what exactly is it? What is the matrix doing? And more importantly, what is it forcing us to do?

Here's a list of my 5 favorite cities:

Look what happens when we turn this into a matrix, even with just one column.

The same simple relationship is repeatedly applied but with different things.

And the more columns we add, more simple relationships are repeated.

This is where the creative problem-solving juju kicks in: a matrix forces us to visually represent something complex by combining simple, repeating relationships.

And what is a simple repeating relationship?

A pattern.

And because we may be able to explain and predict things in a pattern, and because the matrix combines multiple patterns, we might be able to predict new values for multiple patterns and combine them to form creative options.

Proven patterns + but with new values = creative options.

I call this form of creativity (and yes, I'm delighted to announce to the human race that no AI had a role to play in this idiocy of naming) Combinatorial Creativity.

Let's see how it works in practice.

That's it, reader. That's how to use a spreadsheet for creativity.

It doesn't eliminate the work required to create something, but it does support the quality and quantity of creative work. I think that's a rare combination of benefits.

I hope it seemed neither complex nor confusing. If I had to explain all of this to a 5 year old, I'd say: 'make many new things by looking at many old things'.

The spreadsheet just makes it easy to keep track of what options you have by showing you the patterns you've noticed. It is after all, a matrix.

And I have a sneaking suspicion, a feeling if you will, dear reader, …

… that you're the one.