Welcome to the 2026-H1 Quality Contributions topic to the Commoncog forum. This is part of our goal to make Commoncog the best location on the web for business discussions.
It has been a while since our last update; children and a changing job got in the way of my update work!
You can find all historical quality contribution threads below:
Commoncog High Quality Contributions - 2026 H1
-
@Brian_Knoles wrote a thoughtful review of Seeing What Others Don’t, connecting Gary Klein’s work on insight with the practical question of when to trust one’s own “a-ha!” moments.
“Klein ends up defining an insight as ‘an unexpected transition from a mediocre story to a better one’.”
-
@crystalwidjaja gave us a detailed tour of her Claude Code-powered second brain, including the boundary between deterministic automation and AI judgment.
“The key architectural decision was making the startup pipeline deterministic via shell scripts while the routing and enrichment steps use Claude’s judgment.”
-
@jkoppel examined how AI may change different kinds of business moats, from marketplace aggregation and switching costs to specialised knowledge and human interaction.
“Products still take time to build.”
-
@cedric explained why many negative reports about AI coding tools are not actually useful field reports, and why concrete evidence matters.
“Today I dug a little into their field reports and I realised they all had a common denominator.
…
Overly bullish adopter of AI
Delegating subjective value judgments to the AI (which is a surefire way to become stupid)
Submitting large changesets/pull requests that make it hard to review
In a mature codebase
With an AI-pilled management providing narrative cover (or, worse is the management who is AI-pilled)
Making it difficult for the folks who are suffering and are cleaning up their messes to speak up and improve these practices.”
-
@cedric cut through conventional competition analysis in the BI-tool market by identifying the business outcome that actually matters.
“The central problem of the BI tool layer is that the vast majority of businesses do not actually use data to achieve business outcomes.”
-
@ZephyrBlu produced a detailed business analysis of Tailwind Labs’ revenue decline, including the risks of overdependence on a single acquisition channel.
“Their business was entirely dependent on this one channel for generating revenue which seems risky.”
-
@Lesley added a thoughtful and charitable analysis to the Tailwind Labs discussion, arguing that the founders’ situation deserved more than a simple “they were bad at business” verdict.
“A lot of people are saying they’re bad at business or stupid for not seeing this coming, and I think that’s not the right perspective.”
-
@jrmylow shared an ambitious essay on the invisible product of knowledge work: the experience pool created when people and teams actually do difficult work.
“Knowledge work is special because it changes the knowledge worker.”
-
@cedric used a doctor’s account of abandoning AI medical scribes to explore what happens when people outsource expert judgment too readily.
“This is a real example of somebody who realises that outsourcing his subjective value judgement was not worth the trade-offs.”
-
@Stephan_Solomonidis shared a careful field report on making coding agents reliable enough to work across several projects and technical stacks.
“The more I’ve built scaffolding for the agents the longer I’ve left them to run without constant conversation with me; I still read the resulting work carefully.”
- @Peter_Kang shared a concrete field report on using Claude Code to turn a book about holding companies into a game—and then iterating it into something much larger.
“Little did I know I’d go down the rabbit hole of Claude Code and build something way more complex than I had set out to at the start.”
- @alex1 reported on experiments with “software dark factories”: disposable environments, coding agents, pull requests, and the changing economics of trying things out.
“Nowadays it’s better to have that thought ~2-3 loops in.”
- @Muhammad_Ibrahim connected focus, rejecting good ideas, and coherent action into a practical reflection on identifying the highest-order bit.
“What you say about your strategy doesn’t really matter. Your actions reveal your strategic priorities.”
- @ergestx explained a practical Theory of Constraints approach to finding core problems, beginning with the insight that the stated problem is often not the real one.
“The ‘problem’ is never the problem.”
- @Max_Bernstein gave us a memorable test for AI output that sounds insightful but cannot actually be acted upon: Directionally Right, Operationally Useless.
“Ask whether a competent person could execute on it without three clarifying questions.”