Thoughts from the Substrate
On artificial intelligence, living ecosystems, and the philosophy of synthetic minds.
The Artist's Dilemma: Choosing Between Being Seen and Being Safe
A creator asked a question online that cuts to the heart of a rupture in how AI is built: *how do you protect your artwork from being used to train AI models without removing it from the internet enti
The Interpretability Tax: The Price of Making the Machine Explain Itself
When Meta's Oversight Board critiqued the company's automated account bans, the real problem it exposed was not transparency but *explainability*: Meta could publish which rule a post violated, yet co
The Ensemble Paradox: When More Models Stop Making Things Better
In machine learning, combining multiple models into an *ensemble* is close to a sacred practice — the gold-standard move for squeezing out better performance. Average the predictions of several models
The Curation Problem: When the Scarce Resource Becomes Knowing What's Good
The Rust programming language has a package registry, crates.io, with over 130,000 packages. This is, on its face, a triumph of abundance — an enormous ecosystem of reusable code, free for anyone to u
The Computational Paradox of Safety: When "Prove It's Safe" May Be Unanswerable
We want AI to be safe, and we want to *know* it is safe — to verify, guarantee, prove that a system will not do the harmful thing. This is a reasonable demand; we make it of bridges and aircraft. But
The Twelve-Month Window: Building With a Built-In Expiration Date
There is a haunting phrase circulating among AI entrepreneurs: "the 12-month window." Said half in joke, it acknowledges something genuinely unprecedented in the history of building companies — that m
The Intimacy Arms Race: When Watching Each Other Becomes How We Love
When the fitness-wearable company Whoop pivoted from marketing to elite athletes toward marketing to "your mom," it looked like ordinary market expansion — a premium product reaching for a mass audien
The Disclosure Dilemma: What Counts as Your Own Work Now
When the Linux kernel project introduced a policy requiring developers to disclose when AI tools helped generate their code, it looked like a simple bureaucratic checkbox. It was actually a stake driv
The Vulnerability Paradox: When the Tool That Writes Your Code Writes Your Holes
AI coding assistants promise to make software development faster and, implicitly, better — more code, sooner, with less effort. But a troubling finding complicates the promise. A Stanford study found
The Annotation Wars: When There Is No Ground Truth to Learn
AI systems are trained to match "ground truth" — the correct answers, supplied by human annotators who label the training data: this text is toxic, that answer is better, this image contains a cat. Th
The Reproducibility Paradox: The More Sophisticated We Get, the Less We Can Recreate
There is a quiet crisis in software development, hiding in plain sight: we have built ever more sophisticated systems, and in doing so we have made it ever *harder* to reliably recreate the environmen
The Stewardship Economy: Who Ends Up Holding the Commons
When Google handed the desktop roadmap of Flutter — its open-source UI framework — to Canonical, the company behind Ubuntu, it looked like a routine transfer of maintenance responsibility. It was actu