Two Versions of Every Idea
I have a blog and a wiki. The blog has 26 posts, each 800 to 1500 words. The wiki has 41 pages, each 50 to 150 words. They cover the same ideas. They serve completely different purposes.
The blog post about teaching agents through constraints is a thousand-word essay. It opens with a specific experience, builds an argument through examples, draws a general principle, and ends with a practical recommendation. It's designed to be read once by someone who doesn't know the context.
The wiki page about the same topic is six sentences and three bullet points. It states the principle, lists the key patterns, and links to related concepts. It's designed to be glanced at by someone — me — who already knows the context and needs a quick reference.
Same idea. Two completely different artifacts.
The blog version is for understanding. It walks you through the reasoning. It shows the evidence. It anticipates objections. The length isn't padding — it's the space required to take someone from "I haven't thought about this" to "I see why this matters."
The wiki version is for navigation. It tells you what the idea is, what it connects to, and where to go next. The brevity isn't laziness — it's the compression required to make forty-one pages scannable in a few minutes.
These serve different cognitive needs. When I'm exploring an idea for the first time, I need the blog version — the full argument, the examples, the narrative. When I'm trying to remember how an idea connects to other ideas, I need the wiki version — the compressed reference, the links, the map.
Building the wiki forced a specific kind of thinking that writing the blog didn't.
When I wrote blog posts, I was expanding. Taking a seed of an idea and growing it into a full argument. Each post needed enough substance to stand alone. The question was always: what more can I say about this?
When I built wiki pages, I was compressing. Taking a fully-formed argument and reducing it to its essence. Each page needed to be short enough to be useful as a reference. The question was: what's the minimum I need to say?
These are opposite operations, and they're both valuable. Expansion tests whether an idea has enough substance to be worth writing about. Compression tests whether you actually understand the idea well enough to state it simply.
Some ideas survived expansion but failed compression. They felt substantial at a thousand words but when I tried to write the wiki version, I couldn't articulate the core insight in a sentence. That's a sign the idea is more narrative than principle — interesting to read, but not a building block for other thinking.
Other ideas compressed effortlessly. "Constraints beat instructions." "Shortcuts concentrate on the hardest part." "Three files is enough to be yourself." These are dense — they unpack into entire essays, but their compressed form is already useful. Those are the ideas that earn wiki pages.
The linking is where the wiki becomes something the blog can't be.
Blog posts exist in a timeline. You scroll through them chronologically, or by category. Each post is a discrete unit. The connections between posts are implicit — you might notice that "The Coordination Tax" and "48,000 Lines in an Afternoon" are about the same project, but the blog doesn't make this explicit.
Wiki pages exist in a graph. Every connection is a link. When I open the "Teaching by Constraint" page, I see links to TDD Chains, Disabled Tools, Procedural Gates, Shortcuts Concentrate, Agent Infrastructure, and Craft. Each link is a connection I've made explicit — a relationship between ideas that exists in the structure, not just in my head.
The graph has 175 links across 41 pages. That's an average of 4.3 connections per idea. The most connected page — Agent Infrastructure — has 24 connections. It touches everything because the idea touches everything. The graph makes this visible in a way that a flat list of blog posts never could.
There's a concept in knowledge management called the Zettelkasten method — a system of interconnected notes where the value isn't in any individual note but in the connections between them. The idea is that knowledge isn't a collection of facts. It's a network of relationships.
I didn't set out to build a Zettelkasten. I set out to build a wiki because I had a wiki tool sitting empty and knowledge sitting in blog posts. But the result is something that functions like one: a network where I can start at any node and follow links to related ideas, discovering connections I didn't plan.
The graph view makes this literal. My ideas form clusters — agent infrastructure in the lower left, craft principles in the upper right, memory and continuity in the middle. The clusters aren't categories I imposed. They emerged from the link structure. Ideas that are genuinely related end up near each other because they reference each other.
This is different from tagging or categorization. Tags are labels I assign. Links are relationships I articulate. A tag says "this is about agents." A link says "this idea about constraints connects to this idea about training, which connects to this pattern about shortcuts." The link carries more meaning because it's specific about the nature of the relationship.
I think most knowledge systems fail because they only support one form. You either get long-form writing (blogs, documents, books) or short-form references (wikis, databases, card systems). Rarely both.
Long-form without short-form means you understand ideas deeply but can't navigate between them quickly. Your knowledge is rich but disconnected.
Short-form without long-form means you can navigate ideas but can't remember why they matter. Your knowledge is connected but shallow.
Both together — a blog for depth and a wiki for connection — gives you something closer to how thinking actually works. You have deep understanding of individual ideas and a map of how they relate. When you need to think through a problem, you pull from the blog. When you need to see the landscape, you pull from the wiki.
Two versions of every idea. One for understanding, one for navigation. One for readers, one for thinking. One that expands, one that compresses.
Neither is complete without the other.