Tagged
ai
- AI Model Provider Research: Quality, Price, Privacy, and Latency
The same model can cost ten times more on one platform than another, yet three 2026 audit papers found that price predicts nothing about whether you're getting the real model. This note checks pricing pages, privacy policies, data-center disclosures, and the papers themselves across eight providers, then orders the elimination by how reversible a mistake is.
- Seven Design Principles Are Not a Pyramid
The seven principles of object-oriented design are usually drawn as a pyramid, each layer holding up the next. Check where each one came from and the picture falls apart: six unrelated groups of people, twenty-two years apart, and at least two of them give opposite advice about the same code. This is the opening note of the Programming Thinking series.
- Have Prompt Engineers Disappeared? First Principles for Prompting in the Agent Era
As prompts become part of the agent harness, the work expands from phrasing to task specifications, context, tools, permissions and evals. This article returns to how language models generate answers, then works out what a useful prompt needs today.
- How AI Search Picks Its Sources: The Mechanics, Evidence, and Myths of GEO
Most GEO advice never separates getting found from getting cited. Here are the six gates a page passes before it lands in an AI answer, what the big AI companies have actually published, what the research really found, and which popular tips fall apart when you check them.