Rants and Opinions

This is just whatever happens to be on my mind at a given moment.

Amazon S3 Shows How LLMs Get Things Wrong
Ask an LLM for any information on how to do security in Amazon S3, and you'll get a bunch of crap. There just happens to be this one change to how S3 works that shows clearly the effect of frequency on LLM training data. I show you how cutoff dates and LLM's reliance on frequency lead to bad results in LLM output.
My First AI Coding Impressions
First Ai Coding Impression
Causes and Creators I Support
2024 was the year that I shifted my support toward individual creators and causes. So I've been trying to spend my money on causes, systems, services, and people who support the future that I want. I do not support AI and I don't support anyone whose primary contribution is using or promoting AI. I'm trying to make the world a better place by supporting people who help people.
Migrating iRedMail from MySQL to LDAP
These are my notes from migrating my [iRedMail](https://www.iredmail.org) server from a MySQL-based installation to an LDAP-based installation. This blog has a bunch of scripts that turn SQL queries into LDIF files that will create the LDAP structures you need.
Bias and Limitations in AIs and LLMs
Artificial Intelligence (AI) and Large Language Models (LLMs) have really obvious limitations. I don’t see how we build a brighter, better future using models that are unable to escape the frequency distribution of our past.