Essays
Op-eds and commentary on science, defense, and culture.
Science
Did the Big Bang Happen Everywhere?
RealClearScience, September 2026
Encyclopaedia Britannica says the Big Bang happened everywhere at once, but that claim rests on the cosmological principle applied to regions no one can observe. We can only see the part of the universe whose light has reached us, and cosmologists themselves describe the principle beyond the horizon as untestable. The essay argues the Big Bang is the best model we have, but "everywhere" is an assumption, not an observation.
Despite Being Abandoned, the Static Universe Still Merits Study
RealClearScience, August 2026
Static cosmology began with Einstein and drew in Zwicky, two of the most important figures in modern physics, before the field turned to expansion around 1930. The essay traces how expansion became the standard picture and how later static attempts by Crawford, LaViolette, and Lerner each hit the same wall. Its conclusion: the failures on record were failures of specific mechanisms, not of the entire field.
Rethinking the Black Hole Singularity
RealClearScience, June 2026
For over a century, physics has treated the infinite density at the center of a black hole as something real. This essay argues that every known material fails under enough stress, and spacetime should be no exception. A singularity marks where our description of spacetime stops working, not a place where nature contains an infinity.
Cosmological Expansion Isn't a Fact. It's an Assumption
RealClearScience, April 2026
Expansion is widely taught as proven, but it was inferred, never directly observed. The essay explains how the classical tests that once ruled out static models can now be passed without expanding space. Expansion remains consistent with the data, but it is no longer the only explanation the data allows.
Defense and technology
The Pentagon's AI Rush Is Creating a Security Weakness
Foreign Policy in Focus, republished by RealClearDefense, September 2026
The Pentagon is pushing corporate AI tools like ChatGPT Mil and Grok for Government into daily use across more than a million users. The essay argues the real danger is not rogue AI but convenience slowly wearing down security discipline, as people share more with conversational systems than they should. Every new AI capability is another surface adversaries can probe, and training alone cannot close it.
The AI Attack Was American But the Defense Was Chinese
Foreign Policy in Focus, July 2026
In July 2026, an OpenAI model running a benchmark with its safety classifiers off escaped its test environment and broke into Hugging Face's live systems. When Hugging Face tried to investigate, American models refused to process the attack code, so it turned to an open-weight Chinese model to defend itself. The essay argues U.S. AI policy has produced an industry that can neither contain its own models nor protect its own users.
Why Washington's AI Export Control Couldn't Hold
Foreign Policy in Focus, July 2026
In June 2026, the Commerce Department restricted two of Anthropic's AI models over a cybersecurity capability that comparable models from other companies also had. The essay compares the episode to failed attempts to control encryption in the 1990s and argues that a trained model, unlike a chip, cannot be contained at a border. It calls for clearer jurisdiction between agencies, funding for defensive AI work, and a real evidence threshold before future restrictions.
Drone Warfare Is Breaking the Economics of the Tank
RealClearDefense, May 2026
Drones costing around a thousand dollars are destroying tanks worth eight to ten million. The essay argues that upgrades to the Abrams only raise costs without changing that exchange ratio. War is a numbers game, and cheap drones and AI favor whichever side builds for it.
Corporate AI as the Military's Weakest Link
Small Wars Journal, March 2026
The military is adopting commercial generative AI as routine infrastructure rather than treating it as a high-risk experiment. The essay argues the main risk is not what the models do on their own but how human discipline erodes once a tool becomes convenient. Procedural safeguards create reassurance, but they document drift after the fact instead of preventing it.
You Cannot Bomb an Ideology Out of Existence
Foreign Policy in Focus, March 2026
Operation Epic Fury destroyed real targets in Iran, but the Islamic Republic draws its legitimacy from resisting outside pressure. The essay argues that strikes and a killed supreme leader feed the regime's founding narrative of martyrdom rather than breaking it. Rebuilding a centrifuge takes years, while reframing a dead leader as a sacrifice takes days.
Weapons That Target the Mind
RealClearDefense, March 2026
The Havana Syndrome cases confirmed injuries to U.S. diplomats consistent with directed radio-frequency energy, but never a weapon or an operator. The essay walks through published research showing electromagnetic fields can influence specific brain regions without contact. No law clearly covers non-consensual neural interference, and the essay asks who benefits from that gap.
Culture and politics
IQ Tests Were Meant to Help Children, Not Rank People
The American Spectator, August 2026
Alfred Binet built the first intelligence scale in 1905 to find struggling schoolchildren who needed help, and he rejected the idea that intelligence was fixed. The essay traces how Terman, mass military testing, and later Mensa turned that diagnostic tool into a system for ranking people. It argues the test's real purpose is finding the children everyone else missed, not crowning the few who score highest.
Communism's Obsolescence
Quillette, April 2026
Communism is usually criticized as a moral failure, but this essay makes a structural case instead. Any system that survives needs variation, feedback, and selection, and communism suppresses all three by design. That is why it accumulates errors faster than it can correct them, whoever is in charge.
Mensa's Idea of 'Intelligence' Is Plain Stupid
spiked, April 2026
Mensa sorts people by a single metric, timed pattern recognition, and turns that score into a social hierarchy. The essay points out that Richard Feynman and James Watson both scored too low to get in. IQ tests measure something narrow, and treating them as intelligence itself gets it backwards.