If it makes you feel the same as before, does it matter if the details are different?

"They won't remember what you did, but they'll remember how you made them feel."

I learned about data hashing in high school. It's ultimately a time-saver in the task of making sure that an arbitrarily large chunk of data is 100% what you expect it to be. A hash (it has other names; not important here) is a short, meaningless summary of the whole. A hash might look like 49A4DB2776A0. If the whole changes in any way, big or small, the hash changes too. Pieces of software are often distributed with corresponding hashes, each marked with the rules used for a particular hash; this allows you to generate your own hash for that chunk of software, to then compare with the hash that it came with. If your manually-generated hash matches the one that the software provider gave you, then you can be assured that what you downloaded was actually what they intended to give you.

I learned about data compression earlier than that, maybe by watching images load on the early internet. No one had a fast connection to anything, so I got to watch different opinions on data compression and storage play out in front of me, as an image would slowly load. Maybe blurry at first, and then clearer; maybe one row of pixels at a time, scanning downward. The game with image compression is always about making sure you still see what the image creator intended you to see, while minimizing the amount of data that actually has to be moved around. Loss of detail in a blue sky might be an acceptable way to trim the file size; loss of detail in a human face might not. I later learned about lossless data compression, which is to say, basically all compression for data that isn't destined for sight or sound. We can lose some detail and still get the gist of the content (which is why televisions and radios were able to get off the ground at all), but things like software and spreadsheets (usually) have to be kept in full fidelity. Zero loss. Can't just remove every other row from the spreadsheet to make it smaller.

There's a little piece of my brain that was there when I took in these definitions. It absorbed the ideas, considered them for a moment, and then immediately hopped up on a soapbox in the village square to start ranting about achieving absolute compression by reducing data to a hash and then preserving both the hash and the exact size of the original dataset. Because, given a hash and the exact size of the original dataset, a sufficiently able computing device would be able to generate the original dataset.

Almost. This idea almost works.

Hash collisions occur when two chunks of data result in the same hash. A Good Hashing Function rarely produces collisions -- and I mean "randomly picking a single atom out of multiple universes worth of atoms" rare.

But still. We're dealing with an unbounded reality, far as anyone can tell. Probability is a wildcard. Across infinity, there are infinite collisions.

When I hand this notion to my brain-villager, up on its soapbox, it gets a faraway look in its eyes and just grins.

How would you store a tree?

Would you catalog its leaves and branches and roots? Count them, measure them, plot their shapes in 3D space?

How would you get it back, once you stored it? A tree is shaped to fit the earth exactly where it stands: how would you go about arranging a new receptacle for it, once pulled from storage? After clearing a space for the canopy, would you hollow out a space for each root?

And anyway, would it be alive?

Nature does not deal in bit-perfect data storage. The intricacies of this world exist in the continuous tracings of fractal pattern. True random access does not appear to be a thing for this universe.

Still, boots on the digital ground, data storage is obviously useful. There's a good argument for making sure that the spreadsheet you put in is the spreadsheet you take out.

But no matter how pristine the backup, we're always going to experience the data differently -- because we are always different. Everything is changing, all of the time. The only thing that we take with us is us. (Though even that is debatable, and hotly debated.)

If our experience cannot be guaranteed by perfect data transfer, then it's worth reconsidering the connection between the data and the experience to see if we can do better. (I'm not just talking about data here; I'm talking about any version of "push this button and feel x" determinism.) I must then look at all the other ways to move information around, and I find myself -- as ever -- looking at trees.

A seed is not a backup. It's not even necessarily a blueprint. But, in the environment it was made for, its patterns will spiral out into something with a recognizable lineage. It may even be interchangeable with others of its kind, depending on what you care about. It will be singular, necessarily (because nothing ever happens the same way twice), but it will still be the offspring of its progenitor.

Crucially, that-which-comes-from-the-seed will fit in the world. A crystalline backup of a tree that once was, summoned back into existance, would not. (It might survive the reintroduction to existence, but it would be traumatic.)

So, if you want a tree, what do you do? (I'm tempted to end this section there for the drama of it, but I respect you too much to theatrically risk data loss, so here: you plant a fuckin' seed.)

It's not that remembering doesn't matter, but I think it's useful to consider how much it matters, and how that varies based on what's actually important.

Across the measurable capabilities of my brain, my working memory dips lower than the rest. This inclines me to think about these things, because they are relevant to me. (In this, I am the part of society's collective brain that just saw something important, and is taking to the soapbox in the village square.) Because my memory doesn't do for me what it does for you (the average "you"), I find other ways to conjure experience from data.

Mm. That's actually a very good way to put this.

As I said, if the data is retrieved perfectly, with no differences from the original, we'll still experience it differently. Often (always?), the purpose of data storage (and hashing, and compression) is to create a specific experience via data retrieval. But pristine data retrieval is not enough to guarantee an experience. It's ... actually almost irrelevant, I think. For me, anyway.

But regardless of how large or small a role it plays, it is less than 100%. Other things matter too. And if the priority is to create a particular experience, one must consider a seed. Analogous to a fixed-length hash, when paired with the inputs of time and energy and material, a seed produces the experience that created it. Except it's better, because it is relevant to the local context in a way that its predecessor could not ever be. That which is alive fits.

Society has collectively spent a bunch of time trying to package, distribute, and then unpackage predefined experiences, with the intent to give the recipient a push-button feeling. The plan is that a high-fidelity replica of the experience will reliably create the intended feeling, over and over again, for all recipients.

I don't mean this critically. It's nice that all western toilets do basically the same thing. I value that.

We have more tools for the creation of experience, though.

I sort of see myself as standing back from time, collecting seeds as I explore the whole, and arranging and re-arranging those seeds on the timeline -- seeing how this seed here and this other one there changes everything for every moment that follows. It's play, is what it is. I am a delighted child, babbling in the water, wholly absorbed by the way my movement moves the water, my eyes dazzled by the light's refraction and reflection.

What will happen next?

It's not that I have no idea. But I certainly don't know for sure. I am learning the language of this seed-play -- and I am evolving the language as I use it, because that is how language works. ... Huh, didn't expect that sentence to come out. But it's true: as a user wriggles in the constraints of a barely-understood language, their motion looses the language. As I find my way, I create a way.

Which makes sense. I come from a seed, too.

Last updated