a modern class theory, MDMA and treatment effects & ChatGPT fuzz
CC#51 - Socially Acceptable Anxiety, Demand for Remote Work & Banana Bread
Hey there and welcome to ✨ CuratedCuriosity - a bi-weekly newsletter delivering inspiration from all over the internet to the notoriously curious.
Things I Enjoyed Reading.
👑 How Elites Abandoned the Masses
I would call this article “a thoughtprovoking class theory for the 21st century”. Not sure if I agree with all of this, but some things definitely resonated and got me thinking.
Unfortunately, that’s exactly what new elites are doing today: they’re implicitly encouraging people to take little responsibility for their outcomes while taking maximum responsibility in their own lives. They’re criticizing the same accountability structures that they themselves use to rise up.
The new elites say: You don’t need religion, religion’s dumb. You don't need family, family’s oppressive. You don’t need marriage, marriage’s sexist. You don’t need your relationship even, it’s holding you back. You don’t need kids, you’ll hate being a parent. You don’t need to send kids to gifted programs, it won’t make them better anyway. You don’t need to work harder, everyone else should work less hard.
You don’t need to do any of those things, but which of those things do the elites do? All of them.
Today’s elites are not only denying that they’re elite, they’re denying other people the opportunity to become elite by claiming it’s impossible, minimizing the effects of hard work, or celebrating the opposite of what they practice.
🩺 Perfect Doctors Heal, but Bad Doctors Reveal
I have to say to some extent this article is a bit of a ‘random’ mix of (1) a write up of the history of MDMA and (2) an illustration of why being good in predicting treatment effects makes it hard to measure them in a causal setting. I think both is interesting, so I can recommend reading it. But if you only happen to be curious about one - it has two parts (‘Act 1’ [MDMA] and an ‘Act 2’ [Treatment Effects]) -so you can easily skip one if you want.
If our Perfect Doctor assigns each person to their best treatment, then our Perfect Doctor will have created a biased dataset in that comparing the MDMA group and the mindfulness group will not reveal a causal effect of either treatment on outcomes. In fact, depending on the distribution of the treatment effects themselves, the Perfect Doctor, as I will show you in a simulation, might even create a dataset in which the average treatment effect is negative but the naive comparison is positive and statistically significant.
The point of the following exercise is drive home a very specific point I think I don’t hear enough — the problem in causal inference is not heterogeneity. The problem is the rationality and efficiency of the underlying treatment assignment mechanisms. The better we get at predicting treatment gains, the better we get at assigning people to those treatments with the highest gains, the more it will look like those treatments are hurting people. The problem in causal inference is rationality and efficiency, not heterogeneity, and that is one of the gifts — perhaps even the main gift — that economics brings to the broad field of causal inference. Let’s illustrate this with a simulation.
😳 Socially acceptable anxiety is still anxiety
There are certain anxieties which are very much accepted or even encouraged in our current societal structure - can we reduce them by showing empathy and simply refusing to enable them?
When I take a little too long to respond to an email or a text, I start to worry I’ve offended the other person, so instead of simply replying with an apology, I put off responding even longer. This cycle continues until the only way out is to fake my own death. (..) I think a good word for it is anxiety. The hallmark of anxiety is worry without results:(..) I don’t improve my email etiquette by avoiding my inbox. (.. ) What’s surprising is that nobody finds my anxious tendencies at all concerning. If you tell people that you manage your stress by tossing back a couple bourbons, or crying in bed all day, or pulling the wings off of flies, they get a little worried. But when you tell them that you manage your stress by thinking the same useless thoughts over and over again, hundreds of times a day, feeling very bad the whole time, they go “oh yeah, me too."
Food for Thought.
⚽️ A wake-up call that we should not only be critical about statistics posted by people/news outlets we dislike (e.g. Trump, Fox News in my case) but also question information by seemingly credible sources that affirm our personal worldview.
Note: This should by no means justify what happens/happened in Qatar - despite this specific number being incorrect there is certainly a lot of things that went terribly wrong.
🎓 Does ChatGPT pose a threat to the education system?
💻 In terms of an employer’s perspective, I think it would be interesting to get some measure of application quality to see whether remote job ads also receive higher quality applications (or if changing a job ad from on-site to remote basically just results in a bunch of additional low-quality applications).
Random Stuff.
✍🏻 After all the GPT fuzz: A reminder that the purpose of writing is not solely producing text
GPT agrees as well. Interesting also that it seems to refer to itself as human.
🍅 Scientific Language vs. Used Language can sometimes lead to confusions. Fun Fact: In 1893 the U.S. Supreme Court issued a ruling that classified tomatoes officially as vegetables (although they were aware of the ‘true’ botanical classification).
🪥 In case you haven’t spent enough time yet playing around with GPT-3 - here is an article on how to use it / optimize for prompts to assist in creative work. Can’t wait for that toothbrush that communicates with smart toothpaste (and hopefully orders a new one when I run out).
Personal Update.
After an initial baking fail I have started a series of banana bread experiments - I think its going well (improvement has been slow but steady) but in case you have any great recipes in mind, send them my way! 🍌
Had great fun at the first post-COVID edition of Copenhagen Data Beers - thanks to some seriously cool people for organising. Looking forward to the next edition!