Writer's Block, and Repetition

Over the past few months, writing has been a struggle. Ideas come, I make note of them, I step away from them, come back to them, and the idea morphs from this nugget of opportunity to his festering mess.

I look at what everyone else is writing, and everyone else’s pieces look, feel, read, and are received so well. Even the ones that downplay logic and employ the ol’ tautological kayfabe (”I’m right bc I said I’m right, I mean look at me! You wouldn’t want to look dumb challenging someone who’s right, right?”) get a lot of play.

It’s up and it’s stuck: truly languishing. In the past two years since I last made my writing public, I’ve accomplished a lot. I got a new job, I’ve travelled, learned and unlearned a lot, and wrote over 20,000 words’ worth of [redacted] with anon besties.

I’ve written some long-ish tweet threads and posts on Lens, but haven’t published anything that’s designed to stand on its own as a blog post. I want to blame it on the first year of the pandemic, where I spent my spare time on social apps, and slowly optimized my thinking for conversation instead of long-form thought. And now that I’m trying to reintroduce long-form thought, it’s not coming as naturally as it once felt. After seeing how people’s words often get misconstrued or taken in bad faith, it’s easy and rational to adopt an adversarial way of thinking, even with friends or colleagues. And in that sense, the novelty of getting an idea out there has worn off. You never know if someone with more capital than you will casually reject it, call undue attention to you, and make a moment out of you.

Even so, look at the world! With everything that’s happening, there’s so much to write about:

How does one build a consumer-facing app when infrastructure plays are being invested in way more often?

• Despite economic conditions, there’s a lot of funding in good ideas. We’ve seen capital flow from apps, to infra, to climate and biotech, to web3, to AI in ways that commands conviction or good branding to come out as the most winningest. And because of the asymmetric outcomes we see in VC investments, we see a sort of power law distribution with returns. Some small % of companies get the lion’s share of the returns.
• In web3, we see a reaction to that, and regulatory uncertainty, in the form of capital allocation to infrastructure companies. The logic there is fairly solid: “If I can take a bet on a platform, then the winners will build on this platform. Even if they don’t someone will want this tech to compete with the winning platform[s].” Who could argue with that?
• At scale, we run into the opposite problem: so many early stage capital allocators are either waiting or investing mostly in infrastructure, to the detriment of the companies that would buy and build on that infrastructure. When everyone’s selling shovels, who breaks the dam?
• Said differently, if we view it from the perspective of the “Man in the Arena” poem, it’s as though a group of people, compelled by vim, vigor, and spirit, have jumped into the arena to fight. But they take benches, create a barricade blocking them from the blows of the fighters, and sell armor and weapons to the fighters. form the stands, built a barricade around themselves, and sold weapons to the people who ran through the barricades before realizing what they were doing

What does it mean when the Overton Window on ways to build Web3 apps is still at “pay for actions that mean something?”

Leela Vs The World has everything that makes for a great web game: good art direction, a banging track on loop, and a subway-legible UI . Or does it? If you look at the board on the left, it’s fairly certain it’s a game of chess. But this is a game against an AI agent, whose moves are verified on-chain to be outputs of a specific ML model.

• That’s cool and all, but the fact that you have to purchase the ability to make a move blocks out people who aren’t bought in with making a move. And how do you get people to buy in? You sell them a good narrative, or tell them the reward is worth it.
• So then you look at the prize pool, and see that it’s less than 200USD at time of writing. That’s some money, but arguably not worth the opportunity cost or cognitive load of actively nudging this game to completion for many people.
• And so, you end up with this great concept that either needs a ton of money plowed into it to get a bunch of people or a few specialized coders push this forward. And for proof of concepts like putting auditable AI models onchain, the wrong kind of juice may erupt from the fruit if you squeeze it the wrong way.

What are the intersections between composability, agency, and technology?

Composability is like a gestalt cake, where the whole is different from the sum of its parts

1. Composability refers to the extent to which behaviors, actions, or objects can be separated and recombined. In other words, high composability means that a given activity can be divided into smaller parts and reassembled, while low composability refers to actions that are more singular or indivisible.

Agency is the ability to FAFO (ability to make choices of your own free will)l

1. Agency refers to the capacity of an individual to act independently and make their own free choices. High agency means the individual has a significant amount of control over the outcome, while low agency means they have little control.

Technology represents the tools that enable, facilitate, and sometimes restrict agency or composability for other reasons (business reasons, yada yada)

Here are examples for each category:

Low Composability, Low Agency

1. Attending a concert: You don't have much control over the event itself, and the experience can't be easily broken down and reassembled.
2. Riding a bus: You have little control over the route, schedule, or the actions of the other passengers, and the experience itself is fairly singular.
3. Watching a movie at a theater: The plot and the viewing experience are fixed and can't be broken down or controlled.

Low Composability, High Agency

1. Going for a run: You can choose when, where, and how fast you run, but the action  (running) is relatively indivisible.
2. Cooking a specific dish from a recipe: You can decide when to cook, what ingredients to use, and how strictly you follow the recipe, but cooking the dish itself cannot be broken down into smaller, standalone actions (ie there's only so many ways to blanche something).
3. Painting a picture: You have control over the process, but the final product, a completed painting, is relatively singular and can't be divided into standalone parts.

High Composability, Low Agency

1. Working in an assembly line: The work is divided into discrete, repetitive tasks, but the worker often has little control over the larger process.
2. Attending a structured class or lecture: The content is broken down into discrete topics, but the attendee has little control over the content or pace of the presentation.
3. Playing in an orchestra: The music is composed of individual parts, but the player has little control over the overall composition or performance.

High Composability, High Agency

1. Building a Lego model: The model can be built and rebuilt in different ways, and you have full control over how you do it.
2. Writing a novel or story: The narrative can be broken down into chapters, sections, and individual scenes, which can be rearranged or rewritten, and the writer has complete control over this process.
3. Writing code: The code can be structured in a number of ways depending on what the writer intends to optimize (speed, performance, verbosity, elegance). The code can also be called or reference by other programs, if it’s written in a language and method that allows that.

• There’s an emergent element in the interplay between composability, agency, and technology. Over the course of human civilization, technological tools from the hammer to the airplane have increased agency, and by extension composability of actions. No longer does someone have to spend years of their lives sailing from one country to the next - they spend a few hours in a metal bird! The ability to do more with a set of tools increases one’s agency, and increasing one’s agency in this sense allows them to think with more composability.
• Historically, the best products have given people the ability to learn and do more. But in the past few years, we’ve seen a lot of investment in products that downshift agency and composability for business reasons.

The preceding bullet points are a collection of questions and thoughts that have persisted over the past few weeks. Through writing little bits of an argument over time, returning to the doc, adding links and tags for links, and reviewing what I write, I’m hoping to get back to where I was with my writing skills.

To return to the epistolographical dysfunction I was describing, I want to call it something like writer’s block, but the issue with that is that I know how I personally get through: doing more repetitions. Writing more, every day. Reviewing old notes before bed and when I wake up to increase familiarity. Taking mindful time to write something down with an end goal in mind: summary, persuasion, cartography.

I want to call it something like coder’s block, but the issue with that is that the solution is the same: ship and iterate. Shipping for writing is obvious: you post the article online. Iterating on writing seems simple, but there’s a cognitive barrier to iterations with writing. Over time, we’ve seen iterations come to video, music, and gaming, so arguably collective acceptance of it has shifted.

But how do you iterate on writing? Writing is seen as static. The way a brain interprets words is stochastic, with a finite amount of patience that’s dwindling the more short-form CDNs masquerading as a social network pop up . Iterating on code is more dynamic, interactive, exploratory: the coder can add logic that changes the color, the behavior, and the values in ways that can make new experiences. I’d posit that the same thing happens with writing. Writers add sentences that change the flow. They change words to increase or decrease the chances of one interpretation over the other. They rearrange, add, and remove whole sections of an argument if it seems out of place.

When you consider that coding is also referred to as “writing code,” and that programs like Microsoft Word and Notion can track changes, you start to see some of the flexibility for changelogs in writing. But I haven’t seen it as much as I’ve seen other forms of digital content updated on-the-fly (yet).

Over the next couple of weeks, I’ll be writing to get ideas out there, work on my reps, and find or figure out if there’s a solution to the following problem:

What do modern forms of text-based media look like?

If the solution doesn’t exist, maybe we can make it together.