Architecture

I Built a 1977 Computer From Scratch. It Changed How I See Every Modern Device.

The SB mini II isn’t a replicaβ€”it’s a lesson in computational archaeology. Building a clone of the 1977 Apple II from scratch forces you to understand hardware-software co-design at a fundamental level, revealing timeless truths that modern abstraction layers obscure. This isn’t nostalgia; it’s the deepest possible education in how computers truly work.

Your CAD Software Is a Bug Factory. Functional Programming Fixes That.

Traditional CAD software runs on mutable state β€” the same paradigm that gave us null pointer exceptions. Every broken reference, every collapsed feature tree, every unreproducible model traces back to that root cause. A new Haskell-based CAD environment compiled to WASM and running in the browser proves that functional purity can eliminate entire classes of design bugs, making geometry logical rather than just visual.

Your iPhone Is Less Secure Than a 1970s Military Computer β€” Here’s Why

We’ve spent 50 years making computers faster, but we forgot to make them safer. The 1970s KSOS operating system proved that security can be built into the core, not patched on later. Modern OS design abandoned that approach β€” and we’re still paying the price with endless zero-days and emergency updates.

Stop Fearing AI Will Kill Your UI. It’s the Opposite.

The hype says AI will kill GUI. But the truth is the opposite: the more autonomous agents become, the more humans need a trustworthy interface to verify what changed. Skill and GUI are complementary layers β€” one for doing, one for seeing. Product managers must now design the trust boundary between autonomous action and human accountability.

Markdown Is a Trap: Why Everyone’s Wrong About the β€˜Universal’ Format for AI Agents

Everyone from Karpathy to Google to Tan says Markdown is the answer for AI agents. But they’re solving completely different problems. This article cuts through the false consensus and shows why Markdown’s flexibility is actually its biggest weakness β€” and how to avoid cargo-culting a solution that doesn’t fit your architecture.

Stop Making Chips Faster. You’re Chasing the Wrong Bottleneck.

Data movement, not computation, is the dominant bottleneck in modern computing. A fourth-root complexity scaling law reveals that as hardware density grows, the cost of moving data outpaces compute speedups β€” meaning the industry’s obsession with faster chips is optimizing the wrong variable. The real leverage lies in radical memory hierarchy redesign and non-von Neumann architectures.