Research distillation is the opposite of research debt. It can be incredibly satisfying, combining deep scientific understanding, empathy, and design to do justice to our research and lay bare beautiful insights. Distillation is also hard. It’s tempting to think of explaining an idea as just putting a layer of polish on it, but good explanations often involve transforming the idea. This kind of refinement of an idea can take just as much effort and deep understanding as the initial discovery. This leaves us with no easy way out. We can’t solve research debt by having one person write a textbook: their energy is spread too thin to polish every idea from scratch. We can’t outsource distillation to less skilled non-experts: refining and explaining ideas requires creativity and deep understanding, just as much as novel research. Research distillation doesn’t have to be you, but it does have to be us.

Research debt is the accumulation of missing interpretive labor. It’s extremely natural for young ideas to go through a stage of debt, like early prototypes in engineering. The problem is that we often stop at that point. Young ideas aren’t ending points for us to put in a paper and abandon. When we let things stop there the debt piles up. It becomes harder to understand and build on each other’s work and the field fragments.

Developing good abstractions, notations, visualizations, and so forth, is improving the user interfaces for ideas. This helps both with understanding ideas for the first time and with thinking clearly about them. Conversely, if we can’t explain an idea well, that’s often a sign that we don’t understand it as well as we could.

Emergent AI does not suggest that the computer be given rules to follow but tries to set up a system of independent elements within a computer from whose interactions intelligence is expected to emerge. Its sustaining images are drawn, not from the logical, but from the biological. Families of neuron-like entities, societies of anthropomorphized subminds and sub-subminds, are in a simultaneous interaction whose goal is the generation of a fragment of mind. We noted that these models are sometimes theorized in notions of "mind as society," where negotiational processes are placed at the heart of all thinking. Those who espouse and support such models are more inclined to find bricolage acceptable than are classical Piagetians. What concerns us here is not which of these trends in AI is "correct," just as we aren't advocating a choice between the use of icons and the use of textual instructions in computer operating systems. What does concern us is that the new trends -- icons, object-oriented programming, actor languages, society of mind, emergent AI -- all create an intellectual climate in the computational world that undermines the idea that formal methods are the only methods.

As it happens, the Macintosh's iconic style may be winning this argument. The designers of computer interfaces might interpret this as final proof of the technical superiority of icons. A psychologist might read it as putting in question the hard/soft split. Perhaps everyone is really "soft" after all, and "hard" is a construct that is dropped when it is not needed for acceptability or prestige or functionality. Others might simply say that icons are "easier." All of the above may be in part true. But from our perspective what is important is that the iconic victories are part of a larger cultural shift towards an acceptance of concrete, relational ways of thinking.

Music students live in a culture that, over time, has slowly grown a language and models for close relationships with music machines. The harpsichord, like the visual artists' pencils, brushes, and paints, is "just a tool." And yet we understand that artists' encounters with these can (and indeed, will most probably) be close, sensuous, and relational. And that artists will develop highly personal styles of working with them.

The computer presence has provoked a "romantic reaction" in our culture.24 As people take computers seriously as simulated mind, many are in conflict with the mechanistic image that is reflected back to them in the mirror of the machine. They define the specificity of people in terms of what computers cannot do. Simulated thinking may be thinking, but simulated love is never love. Women express this sentiment with particular urgency. We believe this is because a conflict fuels their conviction. A comfortable style of thinking would have them get close to the objects of thought. The computer offers them objects of thought. But the closer they get to this machine, the more anxious they feel. The more they become involved with the computer, the more they insist that it is only a neutral tool. A way out of the impasse would require profound change in the culture that surrounds the computer tool. If the computer is a tool, and of course it is, is it more like a hammer or more like a harpsichord?

The development of a new computer culture would require more than technological progress and more than environments where there is permission to work with highly personal approaches. It would require a new and softer construction of the technological, with a new set of intellectual and emotional values more like those we apply to harpsichords than hammers.25 If computers are really the tools we use to write, to design, to play with ideas and shapes and images, they should not be addressed with the language of desktop calculators. Moving out of the impasse also would require the reconstruction of our cultural assumptions about hard logic as the "law" of thought. Addressing this question brings us full circle to where we began, with the assertion that epistemological pluralism is a necessary condition for a more inclusive computer culture.

Our central thesis is that equal access to even the most basic elements of computation requires an epistemological pluralism, accepting the validity of multiple ways of knowing and thinking.

Like many people, I’ve thought 2016 was a surreal year; the Cubs won the World Series, Hillary Clinton went on television to warn people about white-supremacist memes, Elon Musk has landed rockets on ocean platforms and started an organization to develop Friendly AI.  Surreal, right? No. It’s real, not surreal. If reality looks weird, this means our stories about it are wrong. Did polls and newspapers and social media fail to see this election coming? Then those sources just took a hit in credibility. On a longer-term note, if you know there’s a replication crisis in scientific research, that should be shaking up your trust in published papers. There may be a crisis in politics. But before we can do anything sensible about that, we need to understand that there is a crisis in credence. If the world looks weird to you and me today, that is not a matter for rueful laughter, it is a sign that we are probably badly wrong about lots of things.

Now that you have your component hierarchy, it's time to implement your app. The easiest way is to build a version that takes your data model and renders the UI but has no interactivity. It's best to decouple these processes because building a static version requires a lot of typing and no thinking, and adding interactivity requires a lot of thinking and not a lot of typing.

You can build top-down or bottom-up. That is, you can either start with building the components higher up in the hierarchy (i.e. starting with FilterableProductTable) or with the ones lower in it (ProductRow). In simpler examples, it's usually easier to go top-down, and on larger projects, it's easier to go bottom-up and write tests as you build.

To build a static version of your app that renders your data model, you'll want to build components that reuse other components and pass data using props. props are a way of passing data from parent to child. If you're familiar with the concept of state, don't use state at all to build this static version. State is reserved only for interactivity, that is, data that changes over time.

Since you're often displaying a JSON data model to a user, you'll find that if your model was built correctly, your UI (and therefore your component structure) will map nicely. That's because UI and data models tend to adhere to the same information architecture, which means the work of separating your UI into components is often trivial. Just break it up into components that represent exactly one piece of your data model.

The Greater Good

The more I examined these efforts at sedentarization, the more I came to see them as a state’s attempt to make a society legible, to arrange the population in ways that simplified the classic state functions of taxation, conscription, and prevention of rebellion.  Having begun to think in these terms, I began to see legibility as a central problem in statecraft. The pre-modern state was, in many crucial respects, particularly blind; it knew precious little about its subjects, their wealth, their landholdings and yields, their location, their very identity. It lacked anything like a detailed “map” of its terrain and its people.

Pressing deeper still, there is a fundamental difference between Messenger and Instagram in terms of general mental models about audience and distribution. The Instagram core product was already a one-to-many, public broadcast model, just like Stories, whereas Messenger is primarily a private conversation model. Broadcasting to your entire graph in a space typically reserved for more intimate conversations represents a confusing combination of norm and privacy assumptions.