Bernoulli-Types
2 partsPosts in this series:
- . The Bernoulli Model: A Probabilistic Framework for Data Structures and Types
- . Bloom Filters
Series
Multi-part explorations of deep topics
Showing 14 of 14 series
Explorables
2 partsPosts in this series:
- . Watch an LLM Think
- . How to Teach a Computer Calculus with Pattern Matching
Inductive Biases in Neural Networks
8 partsA from-scratch tour of why each neural architecture works, built as a pure-Python library you can read. Eight bite-sized posts, and a full monograph.
Posts in this series:
- 1. A Network Computes Numbers. The Loss Decides What They Mean.
- 2. The Loss Function Is a Distribution Assumption
- 3. What a Convolution Assumes
- +5 more...
Minds & Machines
26 partsAI alignment, moral agency, superintelligence, and the futures we might build
Posts in this series:
- 3. Latent Reasoning Traces: Memory as Learned Prior
- 4. Value Functions Over Reasoning Traces
- 6. The Policy: When Optimization Becomes Existential Threat
- +23 more...
What Your RL Algorithm Actually Assumes
5 partsA four-part interactive series exploring how representation choices — tabular, linear, neural — shape what reinforcement learning agents can learn, and what they silently bet on.
Posts in this series:
- 1. The Infinite Table
- 2. The Features You Choose Are the Assumptions You Make
- 3. The Architecture Is the Prior
- +2 more...
Sequential Prediction
9 partsFrom Solomonoff induction to modern language models: sequential prediction through information theory, Bayesian methods, and neural approaches
Posts in this series:
- 1. Introduction to Sequential Prediction
- 2. Solomonoff Induction: The Theoretical Ideal
- 3. The Bayesian Prediction Framework
- +6 more...
SICP
10 partsPosts influenced by Structure and Interpretation of Computer Programs, on abstraction, composition, and computation as a medium for expressing ideas
Posts in this series:
- 5. Building Languages to Solve Problems
- 6. Apertures: A Language with Holes
- 8. chop: When Every Command Returns the Same Kind of Thing
- +7 more...
Statistical Reliability
23 partsMaximum likelihood estimation for series systems with masked failure data, from master's thesis to R package ecosystem
Posts in this series:
- 12. algebraic.dist: Distributions as Algebraic Objects in R
- 20. compositional.mle: SICP-Inspired Optimization
- 21. maskedcauses: Maximum Likelihood Estimation for Masked Series System Failures
- +20 more...
Algorithms Arise from Algebraic Structure
23 partsA pedagogical series exploring generic programming through the lens of abstract algebra
Posts in this series:
- 0. Seeing Structure First
- 1. One Algorithm, Infinite Powers
- 2. Modular Arithmetic as Rings
- +20 more...
The Learning Problem
8 partsEssays on induction, inference, and the search for useful representations
Posts in this series:
- 1. All Induction Is the Same Induction
- 2. The Incomputability of Simple Learning
- 3. Latent Reasoning Traces: Memory as Learned Prior
- +5 more...
The Long Echo
15 partsDigital preservation tools and philosophy. Graceful degradation, plain-text archival, and building systems that outlast their creators.
Posts in this series:
- 1. Long Echo: Designing for Digital Resilience Across Decades
- 5. Orientation Under Pressure
- 6. Why I Build Comprehensively in Open Source
- +12 more...
Trapdoor Computing
20 partsA paradigm for computing with values whose meaning is hidden behind a one-way trapdoor. Cipher maps, approximate homomorphisms, and a framework for oblivious function evaluation.
Posts in this series:
- 1. The Bernoulli Model: A Probabilistic Framework for Data Structures and Types
- 2. Noisy Turing Machines: Noisy Logic Gates
- 4. Entropy Maps
- +17 more...
What-Is-It-Like
2 partsPosts in this series:
- . Clankers: Singing Metal
- . Good As New
Algebra over Wire Formats
11 partsA pedagogical series exploring information theory by construction in C++23
Posts in this series:
- 1. Kraft's Inequality
- 2. McMillan's Converse
- 3. Universal Codes as Priors
- +8 more...