“The life of the law has not been logic: it has been experience.”
— Oliver Wendell Holmes Jr., The Common Law (1881)
Legal realism is the view that jurisprudence should emulate the methods of natural science — that is, it should rely on empirical evidence.
Applied to AI, that means testing hype against actual results on real legal problems.
Data and methods are open source, inviting scrutiny, replication, and extension by the broader legal and technical community. The mission is to advance an evidence-based understanding of where AI genuinely serves the practice of law and where it does not.
The writing is my own. I use Claude and Codex as coding, drafting, and scraping tools. No one has offered to pay me to shill. No vendor relationships, no affiliate deals, no sponsored coverage.
Bio#
I’m a lawyer specializing in investigations and litigation. I studied cognitive science, symbolic systems, and philosophy of mind.
I started LegalRealist to test AI claims against real legal problems. I write the code and run the analysis myself, for better or worse.
My background came with statistics, programming, and a working knowledge of data analysis. I’ve built full-stack web apps that no one used. Thanks to vibe coding, I can now make them faster.
What to Expect#
New posts go up several times a month, covering legal AI from multiple angles: industry analysis, technical deep dives, practical playbooks, and the occasional philosophical detour. Most posts are part of a running series; all of them are meant to be useful six months after publication, not just the week they drop.
Contact#
Get in touch if you have questions or comments about my work, or want to collaborate.