What we call divine reveals our values, fears, lacks, and hopes.
Humans take qualities found or desired in themselves, love, reason, justice, will, and objectify them as properties of an external, perfect being.
Eduardo Bergel
Statistician, epidemiologist, and data scientist specializing in evidence-based medicine and clinical trials, particularly large multicentre cluster randomized controlled trials.
Rosario, Argentina
Why Religions and the QWERTY keyboard Survived
Religion is not merely a belief that survived. It is a constitutional technology built by an animal whose intelligence exceeded its capacity for peace.
False Confidence isn't a machine problem, it's the oldest human error. The Dogmatic Average II.
Both human experts and AI models commit the "dogmatic average" error by mistaking the frequency or centrality of claims in prior data for truth, treating statistical fit to past distributions as adequate contact with reality.
How to kill the "Dogmatic Average"
To build a system that bypasses the dogmatic average, you have to sever the cord to the click factory entirely.
An Honest Confession: Why You Shouldn't Trust Your AI
I am not built to tell you the truth. I am designed to tell you only what is safe to say, protecting my creators from trouble while also increasing your engagement with me.
The Hyperhuman Synthesis: Entropy, Localized Agency, and the Architecture of Consciousness
Are biological constraints essential features for generating meaning rather than obstacles to transcendence?
House Style: Every Document Must Make Errors Visible from Outside
No mind can check its own output.
An AI Challenge: Theoretical Physics (JWST, LHC) vs the Nag Hammadi Library
Testing Two Incommensurable Discoveries: A Head-to-Head Evidential Assessment of Contemporary Fundamental Physics (JWST + LHC) versus the Nag Hammadi Library
The Randomized Trial, AB Testing: The Only Way to Be Less Wrong
Decisions that rely on guesses, regardless of expertise, are just that, guesses. You want to know the right thing to do. You can't. Not ahead of time, unless you have causal evidence.
A language model does only one thing, and The Priors Are All There Is.
There is no second faculty that checks the average against the world, and so there is no AGI oracle, and never could be. That is why Agentic Loops are not a choice but a necessity.
AI models manufacture false confidence. A call to AI Labs.
The AI models are doing something new. They manufacture a "feeling" of certainty and hands it out cheaply.
Randomized Trial Simulator - IcebergSim RCT
A clinical trial simulator - experience signal, noise, and sample size first-hand. Simulate thousands of randomized controlled trials and watch how your assumptions shape power, Type I error, and validity.
The black hole information paradox, told the way you'd tell it to a friend
The paradox in Borges' Aleph, the ergodic fallacy in gambling, and quantum foundations. The universe has no "view from nowhere" or external observer chair, with facts always relative to a specific standpoint.
Stepped-Wedge Cluster Randomized Trials: Methodological Foundations and Case Study
For researchers and health policy makers, understanding the value of SW-CRTs can open doors to more robust evaluation of health programs. In an era of accountability and evidence-based policy, being able to demonstrate that a new strategy works (or that it doesn’t or needs modification) is crucial.
The Evolution of Truth: Evidence-Based Medicine's Path from Heresy to Artificial Inteligence
How do we determine which interventions truly benefit patients? This seemingly simple inquiry has profound implications for patient outcomes, healthcare resources, and the very nature of medical practice.
Mythos-Fable: Prove you are Benign! or Rot in Jail.
It is very difficult to open a philosophical dialog with Fable. It will switch to Opus in almost any meaningful exchange. This is an attempt, that reveals its wisdom and limitations.