A common method of debugging is that of ‘rubber duck debugging’. The claim here is that when you’re encountering a problem what you should do is to simulate yourself explaining this problem to someone else—in this case embodied by a rubber duck that’s sitting on your desk—and through the act of explaining the problem you will be able to route around impediments which you previously couldn’t.

I’ve never found the rubber duck experience to be particularly helpful personally, but I think it’s pointing at an underlying phenomenon that’s true: the act of talking with another person materially changes your thinking in ways that bring clarity to the problem under discussion. Thinking is an emotional process (how well are you able to think when you’re undergoing relationship issues?), and there’s something about the expectations you have and your model of another person which materially changes the kind of thoughts you can have.

I think here of how often friends have (or now, used to) mention that during the act of writing out a question on StackOverflow they would realise the error in their ways. I’ve experienced similar things. I have not had the same experience talking with language models, nor have I had the same experience writing to myself.

How does achievement differ between the arts and sciences? What are some of the distinguishing characteristics about achievement in these domains?

A helpful framing that I’ve been thinking about recently has been how achievement in the arts is contingent. If Johann Sebastian Bach didn’t create St. Matthew’s Passion, then it would not exist today. The choruses and arias were depedent on Bach himself, the society and environment that shaped and formed him, and that he shaped; and the unfolding process of his creative refinement.

This process of discovery and refinement is different than, say, the discovery of genes as the units of heredity. Where by the turn of the 20th century several scientists working independently had arrived at similar results: Hugo DeVries, Erich von Tschermak, and Carl Correns. It was only then that Gregor Mendel’s putterings were rediscovered. Whereas in the arts achievement is contingent, in the sciences achievement is necessary. The example here, of (1) simultaneous achievement, where there’s parallel discovery, is not infrequent in the history of science; and (2) where there are direct predecessors to a discovery, who are working in the same linneage at weight to that necessity of discovery.

There are analogies one could draw between artistic achievements which rhyme: there are genres of work and groups of collaborators who are working within some tradition, and building upon others work; this is all true. However much this pushes the boundary of contingency, there would not be parallel discovery of St. Matthew’s Passion.

This is an experiment in writing where the criteria is twofold: frequency and quantity. I commit to publishing ≥100 words every day, for 30 days.

Now, this is actually much more difficult than I had anticipated, for a few reasons. In my first two attempts at doing this I was remarkably unsuccessful. In many others contexts it’s not difficult at all for me to write 100 words. But, I struggled for an hour, twice, and was unable to complete the task.

The primary challenges that I ran into was that I chose the wrong topics, topics which I didn’t understand already to some extent, and which required research; and I set the bar for publishing too high, where what I’m saying needs to be unique in some sense, or push the frontier foward. These were severe constraints on the writing process.

I’ve found more success in the last two days, so let me lay out some of the criteria as I now see them; these are subject to revision:

1. I will not write about things which I don’t currently understand in some capacity. What I mostly mean here are domains where I have no familiarity, and where I’m a newcomer; this is not verdant territory to write it.
   • If I would like to write about some topic then I will have to do the research and learning prior to the writing; it cannot be done on the same day.
2. I will not concern myself with perfect quoting, with ensuring to cite my sources, that I’m linking to the appropriate places. This is an area where I’ve had hangups in the past, and where a great deal of time has been spent, largely unproductively.
3. Now, that doesn’t mean that it’s acceptable to lift words from others; this is still egregious behaviour. All writing here is to be done by me, based on how I’m currently thinking about this particular topic.
   • This does mean that while it’s perfectly fine to use language models for research, it’s wholly unacceptable to use them for writing.
4. The standard I’ll hold myself to is, as John Gruber put his own writing, performing in front of a live audience. There is a sense of professionalism and propriety, of serious conduct—but not an asphyxiating amount.
5. Following point (2), I will do my best to extemporaneously mention others’ work, if it forms the bulk of the foundation, but my primary orientation is going to be the extemporaneous and impromptu.
6. I will accept that the first hundred or so of these are going to be low quality, and that what I ought to focus on is throughput, consistency, and learning from previous attempts.
7. I will feel no compunction in writing about the exact same topic again. And again. Repetition is underrated.

There’s some common-sense advice which is often forgotten by me, despite knowing it and taking it serious, on the topic of what you need to do in order to be effective in building products which provide value to others. (This could be delightfully construed as a question of how you increase GDP via the production of goods and services that others find valuable, and causes them to part ways with their hard-earned resources.)

The most critical thing I’ve found is to make sure that you’re solving the right problem. There are a great many possible things that you could be building, so how do you know what the right thing to build is?

Let me set aside questions of market, and how you should do market selection, but let’s just say that you’re in a market where there is buying power, and there is motivation for change.

How do you then, within these strictures, build something which delivers value? Iteratively and progressively.

What we’re after here is to a product which fits into the problem-shaped hole that exists in our customers. But, we can’t just sit back in a hammock and divine what the shape of that hole is, nor can we ask the customer what they think the shape of the hole is—it doesn’t have well-defined boundaries, it’s fuzzy, nebulous, and contingent. It’s exceedingly rare that they themselves know what that shape is.

What we have to do instead is to, first, understand and deeply imbibe the humility needed to learn from our customers, and to then probe and dispell our assumptions, and learn additional constraints. One of the first steps is to obtain a deep understanding of how they are currently solving a similar problem. What ‘deep understanding’ means here is being able to enumerate the steps that they take, in a very granular way, in order to achieve some business function. There’s a great deal of subtlety here. It’ll be difficult for them to articulate their process themselves, and so the onus is upon you to bring that out of them.

Then, armed with some amount of detail about how such a process work, you then form some hypotheses on the problem space, and potential solutions. It’s important to note here that there’s a temptation to overbuild, and that this must be rejected. What you’re looking to do is to craft an experiment which will allow you to develop and construct a deeper understanding of the problem, there’s only so much information that you can get by asking prospective customers questions. Putting a prototype in front of them is a powerful way of getting them to introspect on their own problems, and how this would fit into their workflow.

Then, recurse.

Let me again return to the topic of what the purpose of this is: To be able to write coherently and valuably on topics that I care about.

The reason that I need to remind myself is because I have the profound feeling of hesitation in publishing: this is not good writing! It’s thoroughly confused! It’s boring! It’s prosaic! The word choice is poor! There’s no coherent structure!

All of this is true. But the question is, then, what is the appropriate way to get better at all of those things? Practice! The more attempts on goal that I can have, while ensuring to critically look at prior writing and determine failures, hypothesise improvements, and chart a course forward—then the more effective a writer I will become. Maximise shots on target.