Blog | Hendrik Erz

Panic? Thoughts on the U.S. Presidential Debate

Last Thursday, the presidential candidates for the 2024 U.S. election met for their first public debate in Atlanta. The reactions to the debate were clear: It was a disaster for Biden. While I tend to agree, I think that the debate debacle rather points to a deeper crisis of the Democratic Party: A lack of vision, charismatic leaders, and a hostile environment to do politics in.

How hard can line endings be, really? A horror story from integration hell

They are invisible to the eye, yet crucial for writing: Line endings. While most of the time, it doesn't matter whether you use Windows or Mac, it becomes a central piece of writing when it comes to detecting linefeeds. In this article, I share a short story on how you can really mess this – seemingly simple – task up.

Understanding TOTP Two-Factor Authentication: An ELI5

When logging in online, you usually have to provide a second factor before the website lets you in: A six digit code from your smartphone. Have you ever wondered how this works? After a recent incident where I lost all my 2FA keys, I decided to understand the algorithm.

The Transformer Architecture: A Visual Guide

A few days ago, Grant Sanderson a.k.a. 3Blue1Brown started uploading videos offering a visual introduction to the transformer architecture. I think they are great and you should watch them. Also, they reminded me that two years ago I produced something very similar, but in PDF-format because I like to look at large images that both show the entire process in all its gory glory and the detail at the same time.

A Rant

R sucks, and in this – very opiniated – article, I collect a selection of the various issues I have with this language. Do not get me wrong: R is great for statistics, but for everything else it really, really sucks.

Web Design for Web Scrapers

Social scientists sometimes find themselves in the situation of having to scrape some data from websites since a corresponding dataset does not yet exist. In that case, they are often confronted with confusing HTML code and can struggle to extract the data they want. In this article, I want to provide a primer in web design, the profession of creating these websites. Understanding what web designers think will help disentangle the issues many of us face when looking at some HTML and trying to extract precious data from it.

Installing Linux: 18 years later

I installed Linux on my work computer. After having been unable to properly use it for the better part of the past three years, I asked IT whether I could wipe Windows off the machine. It turns out to be one of the best decisions I did regarding that computer, and now I have not one, but two workable computers. In this article I compare this experience to my first one 18 years ago, and conclude with whether I can recommend Linux for productive use.

The Telemetry Fallacy

Should you collect telemetry data from your apps? The short answer is no, but the reason why is more complex. It has to do with user's intentions, data poverty, and the simple fact that one does not collect telemetry in an Open Source app. In this article, I report on my (possibly half-baked) experiences with what a good app really needs.

Dedication, Social Pressure, and Spite

Today something a bit more personal: Recently I asked myself the very existential question of why I even continue developing Zettlr, given that there are so many alternatives out there. The reason turns out to come back to three fundamental driving forces: Dedication, Social Pressure, and Spite. But especially spite.

Join us at NSA 2024: Pretrained Language Models for Sociology

On my own behalf, I want to invite you – the readers of this blog – to submit abstracts to a session that I will host at the NSA 2024. My colleague Sebastian Gießler and I want to discuss in an open session the ethical, methodological, and theoretical pitfalls in utilizing pretrained neural network models in sociological or general social scientific inquiry.

You are viewing page 1 of 11