Binary churn flags tell you that a user left but say nothing about the timing. This post covers discrete time-to-event modeling: why fixed intervals fit WordPress data, how right censoring skews a model, and a small PHP function for hazard rates.
Notes from moving a Connect Four agent off tabular Q-learning and onto a DQN: why correlated updates destabilize the network, how action masking handles full columns, where the GIL caps throughput, and why the agent learned attack before defense.
An LLM can retrieve the right document and still quote the wrong price. This is the detection and healing layer I built to catch that: five failure patterns, a faithfulness score, deterministic patches, and a SQLite drift monitor that caught a real bug in staging.
LangChain is fine for a prototype. Once there is real traffic, several agents coordinating and a pager attached, the abstraction costs more than it saves. This is the case for writing the orchestration layer yourself.
WordPress wants to build Block Editor Handbook pages straight from block.json at build time. That would kill doc drift and give core blocks a reference that matches the code you are actually running.
Utility software gets designed to be invisible, which is why people resent it. Linear, Vercel and MacPaw treat the maintenance layer as something a person actually sits through, and it shows.
Ridge, Lasso and ElasticNet land within 0.3% of each other on prediction accuracy, but Ridge fits eight times faster and Lasso’s feature selection falls apart once features are correlated. Three numbers tell you which one to fit.
Saturation is a correction tool, not a volume knob. How CSS saturate() actually paints, why the order you chain filters in changes the result, and the stacking context backdrop-filter quietly creates.
Manual regression testing was my bottleneck, so I automated it. This covers wp-env setup, Playwright with the WordPress test utils, and testing block variations and meta bindings without clicking through the editor.
Hard-coded constants assume a perfect environment. This covers four ways to model uncertainty, from robust optimization to two-stage recourse, and refactors a fixed cron batch size into one that reacts to load.