There are two things I try to keep in mind when preparing or delivering teaching. Helpfully, they are both described in xkcd comics. Ten Thousand Average Familiarity The first comic (Ten Thousand) is a reminder that, as teachers, we are privileged to teach a topic we know well to people who are new to it. To me, this suggests two things for us to do: First, learn to enjoy that feeling of taking someone through a process of discovery.

I am currently preparing for the courses I teach in term 2 and one of my objectives this year is to create course websites ¹ with course material that the students can access but that can also be made available publicly in the hope that they might be useful to someone ² . It is also an opportunity to move away from a workflow that discourages improvement and makes editing material cumbersome.

A lot of things are less than ideal in the world at the moment, but I can’t help but be amazed at what one can do with a static website. WebR through quarto-live allows you to run R (you can also run Python with pyodide) in the browser. No server required. Like below: eyJhdHRyIjp7ImVkaXQiOnRydWUsImV2YWwiOnRydWV9LCJjb2RlIjoiMisyIn0= I can’t help but feel this is a little like magic.

I was reading this article in the Guardian. The study it refers too looks like it was written by Confounding et al. but there is no link to the study for the reader to check this intuition. It feels like a basic requirement for any journalist writing about science to provide a link to the study they are writing about. Scientists do not give you anonymous tips!

Like many others, I have been very skeptical about AI detectors. And my default position, up to now is that you should not use them. Ever. Last week, there was a news bit on the Nature website suggesting that there had been huge progress in AI detection.

Last week, I was talking about how I used gpt4 to get myself out of a colouring book pickle, but my (simplistic) solution, which kept my kids occupied, is already outdated with the release of GPTs: OpenAI has released a GPT that is pre-prompted to produce colouring book images. I played around with it and it is pretty fun.

Welcome to ten minute to better writing. If, like me, you find yourself writing a lot in English as a non-native speaker, I hope that you will find some of the tips I share in the videos below useful. I have broken down the material into 10 videos each about 1 minute long, this should help you find what you need quickly.

Is data literacy more important than data analytics? I think it is true for most people that learning data analytics is less important than learning the fundamentals of data literacy. This is why after teaching data analytics for about 6 years, I am now creating a data literacy class. My aim is to equip learners with the tools to evaluate claims that are made with data.

Adopting a historical perspective on things we take for granted is often fascinating. For example, measurement is not something one usually spends a lot of time questioning. But did you know that the meter was last redefined in 1983, only 39 years ago?

Yesterday’s headlines have brought a wave of optimism about the UK’s economy. We’re seeing reports suggesting that the ‘UK inflation set to fall ‘dramatically’ to two-year low’ and about how little house prices have fallen. These stories suggest a reassuring picture: inflation is easing, and the housing market is showing signs of resilience. It’s tempting to feel a sense of relief from these updates.

In a recent column, Tim Harford said: While this claim might seem simplistic, I believe that there is something fundamental in it. Sure, debunking certain claims is time consuming and require skills that we might not have, especially when the claims are made with data. Often, however, even limited amount of poking can help us assess the robustness of claims we are exposed to on social media or offline.