Blog | Hendrik Erz


Centering a Distribution: A Visual Guide

I recently needed to center a few distributions. While a very simple operation, I tend to double-check my intuitions without blindly implementing something. However, a Google search did not turn up anything useful, so I fiddled around myself, confirmed my intuition, and tell you all about how to center a distribution in this short article.

Why your Spellchecker is Probably Smarter than GPT-3

With more and more advances in the development of large language models (LLMs), more and more people feel confident stating that those models have an actual understanding of language and know what you meant. While this is obviously not true, it begs the question: Do language tools in general never know what we mean, or are there cases where they do? In this article, I think about what "meaning" means and whether or not your spellchecker knows what you meant.

The horrifying beauty of JavaScript

When I started programming, I hated JavaScript. It was inconsistent and had weird quirks. Now I can look back on five years of heavy JavaScript coding and have learned to live with the weirdness. However, the language never ceases to amaze, and so today I share an innocent looking code snippet and dive into the horrifying beauty of it.

Should You Use SQLite?

About a year ago I fervently advocated against using SQLite for storing research data, but was since proven absolutely wrong. In this article I explain where and why I was wrong, and share the real reasons why I think we shouldn't use SQLite for research: A lack of skills and time.

We Need to Stop Talking About "AI"

The category of AI is very broad. When someone talks about AI, they could be talking about anything from a very simple computer program to a large neural network. In this article I argue that talking about AI implies talking about expectations, not about actual, purposeful implementations. In order to be capable of talking about and regulating the dangers of AI we have to limit our use of the term "AI" to precisely what it denotes: expectations and possibilities that may or may not be realized some time in the future.

Heat Exhaustion

For two weeks I've had a serious problem of motivating myself. I had several ideas about what might've been the culprit, and I think I have finally pinpointed it: A feeling of existential dread caused by the climate catastrophe that people have already reported.

Crowd Control

Besides actually going on vacation, I usually reserve one week of my vacation for work each year. In summer, I work for a music festival to get a break from my office and do physical work instead. This year, I had an experience that showed me the value of working a completely unrelated job for my actual work as a social scientist.

Is AI finally becoming sentient?

No, it's not. But yesterday, news broke that allegedly some Google chat bot system may have gained sentience. How credible are these claims? While certainly interesting, they do not prove anything – as always.

Thinking in Tables

If you work with data, you can't escape them: tables. Everything from small-scale surveys to huge register datasets are represented using the mental model of a table. While this representation of our data has worked for the better part of recorded sociological history, we are increasingly experiencing roadblocks: our analyses run slow, sometimes our computers crash, and many exciting research opportunities are left unasked, because we fear our computers will be incapable of helping us answer them. In this article, I provide – again – a somewhat technical explanation for what makes data tables so ill-suited for so many tasks, and what we can do instead.

Deep Learning is Witchcraft

Deep learning is a fascinating piece of technology. It basically consists of chaining and stacking together millions of very small functions that, in effect, can predict incredibly complex things. This also means that deep learning can sometimes feel like witchcraft. For example, why do two classifiers almost always perform better than a single one? In this article I'll dig deeper into this curious finding.

You are viewing page 2 of 8