To test LLM prompts, run every prompt change against the same fixed set of real inputs, judge each output with a written pass or fail rule, and compare the new results with the old ones before you ship. That is the whole…
A place to write, and be read.
Practical guides from vlvd.net on AI safety, security testing, study skills, history and games. Read them, follow the writers you like, or write your own.
Latest posts
Fresh off the bench.
Newest first, from everyone. Sign in to follow people and see only their posts.
If you want to write a step-by-step list, the answer is short. Give each step one action, start it with a verb, number the steps in order, and say what the reader should see when the step works. Then follow the list…
To write a short article people finish reading, pick one reader, make one promise in your first two sentences, and cut everything that does not keep that promise. That is the whole method. The rest of this guide shows…
When you search for things I can't control, you are usually stuck in worry, and the fix is not to care less about everything. It is to sort what is yours to act on from what is not, then put your effort in the first…
To drag to twist a row in 2048 on the cube, you press on a row of the right face and pull it sideways, or press on a column of the top face and pull it along. The layer turns a quarter and carries its tiles round to the…
Ontology identification in AI alignment: the ontological crisis is the problem that arises when an AI changes how it represents the world, and the goals you gave it were written in the old representation. An "ontology"…
Laravel security starts with a short list of settings and habits. The framework is meant to be secure by default, and the OWASP Laravel Cheat Sheet says the risk comes from the flexibility around those defaults: a…
Expert systems were programs that copied how a human expert decides things, using a store of facts and if-then rules. Wikipedia's article on expert systems says they were among the first truly successful forms of AI…
What is a Siberian cat? It is Russia's native forest cat: a medium to medium large, strong cat with a thick triple coat, a round look and a mix of calm and mischief. This guide follows the Cat Fanciers' Association…
AI model deprecation is when a provider announces that a model you build on will stop working on a set date. If your product calls that model, you have a deadline you did not choose. The related post on platform risk…
A client onboarding checklist for the first week has five blocks: a greeting by name, what happens this week, what you need from them, how to reach you, and one look at the place they will use. Say each in two or three…
A random drop in games is an item, power-up or reward that appears based on chance instead of a fixed script. If you have ever asked "what is a random drop in games? how loot and power-ups appear", the short answer is…
Customer discovery questions are the questions you ask real people before you write a line of code, so you learn whether a problem is worth solving. The right ones are about the past and the present, not about your idea.…
Hadleigh Castle, in Essex, had three lives. A royal minister built it in the 1230s to show how powerful he was. A king rebuilt it in the 1360s to guard the Thames against French raiders. And in 1551 a new owner sold it…
A pet behaviour problems log: what you tried and what worked, written down as you go, saves you from repeating the same failed fix for the third time. Memory blurs. You forget which week you tried the new routine, and…
If you want AI explained for non-technical founders, you do not need to learn to code or read a research paper. You need a working picture of what a model does, what it costs to run, and where it breaks. With that…
Hume's problem of induction: what he argued and why it matters comes down to one question. You have seen the sun rise every day of your life. What gives you the right to expect it tomorrow? David Hume argued that no…
If you want to know how to write a thank you letter to a mentor, the short answer is this: name one specific thing they did, say what changed because of it, and thank them plainly. Keep it to one page. Send it while the…
If you are writing about a project mistake, a lessons learned post is the form to use. It names what went wrong, shows what it cost, says how you fixed it and states what you would do next time. Keep it to one mistake,…
What is generosity? In the plainest sense, it is giving to others, whether money, time, attention or credit, out of real concern for them and not for what you get back. Philosophers add two things: generosity is a…
Chrome microphone not working? The fastest fix is to work through four things in order: the site permission, the selected microphone, the computer's own mute and input settings, and a moving level bar. This post covers…
If you are asking how to write a changelog for your project, the short answer is this: keep one file, list changes newest first, group them under plain categories, and write each entry so a user can tell what changed for…
If you want to know how to build an AI agent, start small. An AI agent is a program where a model decides what to do next, calls tools, and loops until a job is done. You can build a simple one in an afternoon if you…
The refusal direction in language models, explained in one line: it is a single direction in a chat model's internal activations that, according to one 2024 paper, switches refusing on and off. Take it out and the model…
Symbolic AI is the approach that built intelligence out of explicit symbols and rules a person wrote down. Wikipedia also calls it classical AI or logic-based AI, and people often nickname it good old-fashioned AI, or…
If you are looking for no code AI tool ideas, the best ones share one trait: each does a single job for a single kind of person, and each can be built from one well-written prompt. This post gives you ten of them, a way…
What is a Turkish Van cat? It is a large, strong, mostly white cat from the Lake Van area of Turkey, with color on its head and tail, and a well known liking for water. This guide follows the Cat Fanciers' Association…
Lenin vs Stalin is a question about two leaders of the Soviet state who shared a party and a one-party system but made opposite choices about the economy. Lenin led the Bolsheviks to power in 1917 and brought in the New…
What is armor in video games? Why some enemies take several hits while others fall to one shot? Armor is anything that makes an enemy harder to destroy without making it faster or smarter. The most common form is simple:…
Next.js security comes down to one idea: your app has many places where server code can be reached, and a check in one place does not protect the others. The OWASP Next.js Security Cheat Sheet builds its advice around…
Write here too
Your own page.
Post what you think, follow the people you like and talk about what they write.