A key element in recommendation systems is the concept of similarity. This is usually the first thing anybody who's building a recommendation system for the first time will try to do is identify which entities or objects in their dataset are similar to others, and by how much (e.g. finding items on a store that are similar to the item that a user is currently looking at or has just purchased).

But the concept of similarity measurement is itself non-trivial, and the systems required to produce high quality search results based on a similarity measurement each bring their own challenges and limitations.

This post explores these concepts, challenges, and limitations.

We often treat data as a purely digital or virtual thing; an ineffable signal that lives somewhere within (and perhaps between) our computers, our devices, and the cloud (AKA other people's computers). However, despite the seemingly clean abstraction, data is fundamentally physical; it is rooted in real-world material components, whose physical characteristics and constraints shape and control everything that we can and cannot do with our data.

We've seen a lot of money, time, and effort spent on the development of LLMs, and the consequent development of tools using those LLMs. And while there is a lot of potential, and some industries are better equipped to leverage it, the actual business value realization has been underwhelming. Some folks are calling the AI hype a bubble, while others are saying it's the next industrial revolution. Time will tell, but I think the truth is somewhere in the middle: underneath the hype, there’s a real and very physical revolution powering this explosive innovation.

Data is the lifeblood of modern business. From daily operations to big-picture strategies, everything revolves around it. But there's a catch: the more data you have, the harder—and more expensive—it gets to manage. You might not notice it right away, but as your business grows, so do the inefficiencies lurking in your data systems.