boyter

@Jason Are you two making out over there? No… we are pair programming!

How cool is this?!?

"Are you two making out over there?" "No... we are pair programming!"

@petesmith As it says! I refuse to throw away content that could be useful!

@boyter This looks familiar… 😅. Great blog, well written and quite insightful for us common folk. Looks like I have a new buy list…

And now for something completely different, a blog post on fortified wines boyter.org/posts/fortifie…

Closing the gate after the horse has bolted, raised a family and is on its deathbed having lived to a ripe old age.

A third grader just scored a 5 on AP Calculus BC. The system that trained him contains no LLM. The core is a knowledge graph one man spent 250 hours encoding by hand, two minutes per edge. The platform is Math Academy. Its "AI" is an expert system that routes each student through nearly 3,000 math topics, from 4th grade arithmetic to the math behind machine learning. Every node, every prerequisite link, every weight was placed manually by a team of mathematicians. The weights alone took Justin Skycak a full month: 1,500 topics at the time, roughly 5 prerequisite links each, 2 minutes to estimate each one. 8 hours a day of pure encoding, done before ChatGPT existed to ease the load. Why go through that? Because the graph unlocks mastery learning, the closest thing education research has to a cheat code. In 1984, Benjamin Bloom showed that students with one-on-one tutoring perform two standard deviations above a regular classroom. The average tutored kid beats 98% of the lecture hall. Nobody could afford a tutor per child, so the finding sat in journals for 40 years. A prerequisite graph with a mastery gate is the workaround. The system always knows the exact next topic a specific kid is ready for, drills it until proven, then moves on. Zero time spent waiting for 29 classmates. That waiting is most of school. A year of classroom math is roughly 150 hours of instruction, and the majority goes to pacing, review, and re-teaching. Strip it out and a motivated kid covers six grade levels in one calendar year. The origin makes it better: this grew out of a math program at Pasadena High School where 8th graders were passing AP Calculus BC, back when the founders were still hand-grading the whole thing. The most effective education AI running today is a graph a few humans built by hand, one edge at a time.

@milo_plain It is! But then again I made my fair share of mistakes too.

@boyter making mistakes at industrial speed sounds exhausting

Trad-coding is like using a shovel. AI coding is like using an excavator. You can do it faster, but you can still break pipelines, dig the wrong place and damage stuff.

@complexsplit Next release should half the search time too!

the sneaky fast cs tool by @boyter also uses bm25 to speed agentic/llm code searches. github.com/boyter/cs

@Kyle99338388 That is the other way of looking at it.

@boyter Og programming was using a scapel :)

The fact that people add site: to the search terms to route around the precision index and reach the human long-tail is something constantly feel is both absolute win in how search ranking works and also a massive fail.

Had some spare time, so finished off my first real "polished" attempt at a Godot game, boyter.itch.io/egabomb a remake of Artillery Combat, but what was known as EgaBomb due to the dos .com filename.

What the hell has happened in Sydney. Just had to strongly suggest someone to get off the train to do so which thankfully they did after they threatened multiple older people. Can we please get mental asylums opened again. These people need help.

I spend all day talking to a thinking rock inscribed with runes of power that we put lightning into and that physically sits on the other side of the world. Impressive, but I don't know if thats what I signed up for.

The sales of my local AI book are looking very healthy (now 74), and the ranking in the Amazon store continues to rise 🤩 If you want to learn more about how AI models work, how you can squeeze models onto your local machine (for free), and save on frontier cloud costs, then this book is for you. Contrary to the book's "local first AI developer" messaging, I've been told it's invaluable for anyone wanting to learn more about AI, especially if you want local/agentic workflows. My goal: 100 sales by the end of next week 📚 Then I'll move the goalposts! If I'm lucky enough to get 250 sales, then I'll release my next AI book. Not sure? Drop a comment, or follow for teasers. x 🇬🇧 amazon.co.uk/dp/B0GYDV3FXD 🇺🇸 amazon.com/dp/B0GYDV3FXD #ArtificialIntelligence #LLM #LocalAI #OpenSource

@RobbiewOnline As the saying goes, gotta spend money to make money.

Book sales update - 69 sales... that's erm a good number 🤭 Well over my original 50 target but a little way from my new 100 target. But I've not been marketing it beyond the automated kindle ads and some spurts of targeted mentions on X as I've been very ill this week 🚑 Wifey is away this weekend, so I will have to see what I can do to bump this towards 100, thought right now sleeping all weekend is also an option 😴 Any thoughts? 📈

@RobbiewOnline Best of luck! But I think even promoting just normal tweets/posts might be the go as well!

@boyter Trying a new promo now 🤞 plus because it's a promo the links in there too 🤷🏻‍♂️

@RobbiewOnline Maybe. Give X promotion a try on it.

@boyter I tend not to as I hear the links rank you down in the main post, hence why I added it as the first comment 🤓