What this is
Close readings of real code, teardowns of systems and protocols, essays on engineering practice, and forecasts written down so they can be scored later.
Most technical writing about the future is unfalsifiable, because nobody returns to check. Every forecast here carries a resolution date, a stated confidence, and a review date — and the scoreboard shows which ones were wrong.
Who writes it
Michael Li — software and robotics engineer. Work spans AI agent systems, robotics, and the systems programming underneath both.
Longer pieces are also published as narrated walkthroughs; where a video exists, it is embedded at the top of the post.
Elsewhere
Posts are syndicated to other platforms with rel=canonical pointing back
here, so this is always the source of record. Links to any external copies
appear under each post.
Contact
Corrections and disagreements are welcome — particularly on the code reviews, where being wrong in public is the point.