The Weave

February 3rd, 2012

Nearly everything else that has been discussed here over the years comes down to a single issue or its ripple effects.

The recent iOS-should-have-a-filing-system discussion brought this to the fore. Lukas Mathis, generally smart, talented and discerning (modulo Waffle readership), noted a simple thought: “Organizing documents based on their app is akin to organizing notes based on the pencil you used to write them”.

All of a sudden, the digital era is distilled to its absolute essence. Just like one of the great movements of the past hundred years was to make everything that was actually something out of synthetic materials, like plastic, one of the other great movements underway now, that will probably be going for another few decades at least, is the urge to make everything into a simulation of something. In this way, the screen, and not the countless advances in semiconductors, biology or chemistry, is the most important concept of the past hundred years running. We are inexorably headed down this road, it has a bunch of consequences and there’s nothing anyone can do about it.

Part of defining yourself in relation to this is that nearly everything that is electronic and practical in everyday use offers a simulation of something. No person can alone make a pencil, but what’s more important is that a canvas knows nothing of what it’ll be subjected to. You can paint it a thousand ways, you can fray its fabric, you can burn it, you can make a new kind of canvas out of it by interlacing it with another material or you could dismantle it, sew together a bunch of them and sail a ship with it.

Nothing digital is yet a fabric canvas. We are barely staggering into pencil territory.

The core of the digital problem and the entire expulsion behind the digital opportunity is that there are no ground rules. Actual materials aren’t being simulated, they are subjected to laws of nature. We have to conform to these laws, but as long as we do so, we can make anything we’d like. Now take any computer program. There is not one behavior that’s not explicitly programmed. Those behaviors may not always be intentional, but they are programmed.

Although doing so while hawking his own goods, Joel Spolsky observed that the applications that people really love and take to are the ones that arrange for data structures. You can draw anything you’d like in a notebook but it gives you an interesting format to relate to and to exploit. As far as this can be provided within today’s computers, something like Excel fits this bill.

The flip side of this is that you will have to do a lot of work to make Excel work conform strictly to a process. This is where the other end of the spectrum comes in. Other successful applications learn to exploit the opportunity of being digital. You have to simulate something; it’s suddenly simple to simulate an application that forces you to do things in a certain way. Viewed in isolation, it’s definitely not as pleasant as something free-form, but in context, when you have to do something in a specific way or go through a bunch of steps in order (especially carrying data along), what you end up with can be miles better at doing that particular task.

Platform choice, positions of software freedom, even some basic political attitude comes down to how comfortable you are on this scale. And you don’t pick one camp, you pick a million camps. Mostly, you don’t pick at all, you gravitate back and forth. I don’t have to reconcile my love for some things with my hate for other things unless I’m invested in the idea of having one position, holding it everywhere and declaring myself better than others or enlightened on the back of this decision.

Douglas Adams visited this idea in the late 80’s and some engineers banded together later on to make OpenDoc, which is basically what he wanted. It turns out that it didn’t work that well. It was a way of making the canvas end of the spectrum through consistently playing nice with the ordered end of the spectrum. The order didn’t allow for the flexibility and infinite reuse to happen. There’s no synthesis for it; there are no atoms, no chemical reactions, no material binding, no layer of paint, no individual threads making the sheet of fabric. There’s only a simulation of it.

The simulation always works, but the simulation always works the same. There’s no way to make it not always work the same without making it less stable.

I may be wrong, and there aren’t really two ends of this spectrum, just one with two characteristics that are very hard to combine or provide at the same time because of their tension. But if I’m right, the most extraordinary invention would have to be the digital canvas. A weave not of components, but of some sort of primordial ooze. Something that would have digital natural laws, on top of which you could build new things and reuse it for completely unintended purposes.

Composability is the word for this. We have it in the small; UNIX filters, services, plugins. We have it slightly above that: If you want human reasoning, build yourself a vast neural network and teach it to trust and to train and optimize itself. We don’t have it yet in the large. Probably we will have to go smaller to go larger. Atoms are eminently composable.

Simulations are far from bad. They are dependable and they have done more to advance the progress of humanity than almost anything else. I just hope that we don’t stop here.