Writing with Low Confidence

[epistemic status: 60%]

It’s easy to frame low-confidence writing (i.e. when the author claims they don’t fully believe what they’re saying, common in the rationalist blogscape) as needless pontificating, but here’s my formal claim otherwise:

  1. Low-confidence ideas generally have higher impacts when they turn out to be true, at least when considering the low-confidence ideas that make it out of people’s mouths.
  2. Entire companies have formed around pursuing low-confidence, high-reward ideas. (Alphabet’s Moonshot Factory)
  3. Low confidence is equivalent to high-risk where the risk is being wrong instead of anything substantial or material, and by qualifying it with a low-confidence tag, you’re nullifying any of the ramifications of being wrong.
  4. If you don’t accept premises 1-3, consider that there’s nothing inherently wrong with needless pontificating. It keeps you sharp, it keeps you entertained, and it’s satisfying. Thinking makes people feel good, especially thinking around others.

Writing Another Road to Rome

First of all, a blog-wide disclaimer: I don’t mean to imply that anything I write on this blog hasn’t been written about before. In fact, most posts will be yet another post about a well-covered topic. I’m doing this for two main reasons:

  • I want to get better at verbalizing my thoughts.
  • I want to improve my writing skills.
  • I want to provide another perspective, among the existing pile of perspectives on a particular topic, in hopes that mine clicks with someone where the other perspectives did not.

The name of this blog was inspired by Screwtape’s post on LessWrong, which stuck with me through the years after I read it.

…if I want to learn multivariate calculus or basic game theory there will be at least a dozen different ways of learning it, from a video with pictures to a terse few pages of a textbook to some goofy edutainment videogame. This is a good thing, and I wish rationality was more like this. Think of your favourite technique: how many different formats can it be found in? Alternately, think of your favourite format for learning something: how many techniques and concepts can be found in it?

For every concept we want more people to understand, we should want it explained in more ways. If we want everyone to be a little more rational, then we need to put concepts into mediums that everyone can understand.

While Screwtape specifically name-drops rationality as the subject, I see no reason why this approach isn’t helpful and good across all fields.

Why You Should Work at a Private Prison

[Epistemic status: low confidence]

I honestly doubt that morally good or bad people exist, but I’m going to use these terms as shorthand for “people who care about reducing human suffering” and “people who don’t care about reducing human suffering”.

Imagine a world where all of a sudden, most doctors purposefully induce complications during surgery, or give you prognoses that hurt you instead of help you. Doctoring would become a stigmatized profession that decent people would no longer do. 

“What? You want to become a doctor? Why? All they do is hurt and scam people!”

This, of course, clearly isn’t the case. The profession innately has the capacity to immensely improve the world. It’s just the case that, in our hypothetical, most doctors are harmful. In our hypothetical, a good doctor would have an outsized positive effective. Not only would they not be purposefully and actively harming people, but they would be genuinely helpful and pleasant, fixing maladies and making the world a better place. You know, like doctors do in real life.

Lately, this is how I’ve been thinking about other stigmatized jobs. An extremely brief list of examples might include: prison guard, debt collector, , cop.

If good people decide not to take morally bad jobs, that means bad people will take the morally bad jobs, and arguably produce more morally bad results.

There are three ways to prevent this: One, somehow convince all morally bad people not to take the morally bad jobs that morally good people already won’t take, so the jobs disappear. Two, reshape reality so the morally bad jobs don’t have a chance to exist. This is the project of many radicals, such as prison abolitionists. The third way is to have good people take the morally bad jobs so, instead of leaning into the capacity for harm that the job allows, they lean into the capacity for good, which largely occurs on the margin.

I think this is equivalent to the prisoner’s dilemma in some ways: defecting is taking the objectionable job, and cooperating is agreeing to not take it. In this comparison, you also have advance knowledge that there’s a sub-group of the population will always defect and take the job.

The important takeaway here is that there’s a cultural understanding that morally good jobs are things like being a doctor. But the reality is that it’s extremely rare to be a doctor on the margin — by becoming a doctor, all you’re likely doing is taking that spot from someone else.

On the other hand, being anything but a bad person at a job that traditionally attracts bad people is extremely on the margin. You’re drastically improving the state of the world by acting neutrally, and even more drastically improving it by being a positive moral force.


It’s  pretty intuitive that there already are good people who enter these fields. I imagine some of them are capable of doing good work (for example, there are a lot of judges who strictly prescribe rehabilitative sentences). But I also imagine that bad or corrupt systems have an immune system built into them that prevents them from being reformed, and which prevents individual actors from acting in ways that are against the norm. A prison guard who is really nice to the inmates will probably be bullied by his peers. In fact, I imagine bullying and informal threats from managers are the primary ways that this kind of reform is prevented.

Consumer Anxiety

Since college ended, I’ve felt this anxiety that I’m not watching enough TV shows, or playing enough games, or reading enough books, and whenever I start one, I feel obligated to finish it, even if I start to lose interest. This is especially true for classic anime for some reason: Cowboy Bebop and Neon Genesis Evangelion hang eerily over my head like clouds, both of which I’ve seen the first few episodes of, and I think if I didn’t treat them like canon that I was required to watch, as items begrudgingly on my todo list, I’d probably enjoy them much more.

The same thing has been happening to me with my pocket queue, in which I have approximately 4,000 articles at the moment.

I’m still working on how to free myself from this weird narrative I’ve generated where I feel beholden to large amounts of content like this. I’ll update here when I have more information or a better idea of what’s driving this behavior or how to remedy it.

Blog Design Opinions


There’s a difference between “readable” and “interesting”. I think too often, there’s the idea that “readable” means “simple, boring design and layout”, whereas in reality, the designs that tout readability (I’m looking at you, Medium) are so monotonous that they feel like a carbon-copy of everything else. Readability doesn’t just mean that the text is easy on the eyes, it means that the text grabs and holds your attention.


Don’t use stock images. Not ever. Make sure the image is actually a picture of the thing you’re talking about, not just a reaction image or something in the same general milieu. Make the images personal. Alice Maz uses images wonderfully in her essay on Minecraft economies – it helps that she took all the pictures herself, and that the chosen images augment the text and increase understanding. Just be careful when you add an image — make sure that it’s relevant and facilitates communication. A bad image is worse than neutral: it cheapens your words.

When to Write

Write when you want to record your thoughts and refine them, get feedback, or share them with the world because they seem like they could be useful. Apart from that, I don’t think there needs to be any guidelines around the content of a blog, especially a personal one. Just follow your heart. Your blog will develop concentrations as you publish more.

On Leaderboards

The Problem

Global leaderboards are populated by people you’ll never measure up to unless you devote a large portion of your time to a single game. Very rarely will you ever be in the top 1k, or even 10k. local/friend-only leaderboards are generally too sparse

Leaderboards should promote friendly competition, feelings of pride, and feelings of wanting to work harder (within reason). They should inspire people to play your game and talk about it. Global leaderboards with the most insane, far-fetched scores being most visible do the opposite.

Possible Solutions

Tiered Leaderboards

e.g. “I’m ranked #17 in my global bracket/league of 100”

Possible problem with tiered leaderboards: You’ll probably jump between different brackets way too much for it to be usable, given a game with more than a few thousand active players.

Possible fixes: a “most improved recently” leaderboard; you’re assigned a new cohort/bracket every month or so based on prev performance, & you stay in that bracket for the rest of the month.

Geographically Scoped Leaderboards

e.g. “I’m ranked #27 in Charlottesville, Virginia, and I’m ranked #516 state-wide, and #21,592 nationally”

Write While It’s Still Fresh

There’s a tendency to only start writing about a topic once you feel a level of mastery over it, and I think this hurts the process of learning. The “Toolsday” podcast mentions this, how it’s easy to forget what the actual hurdles of starting up a new project were and instead focus on, as someone more distant from the learning process now, what you imagine the hurdles would be for new people.

Similarly, in his book Pragmatic Thinking and Learning, Andy Hunt talks about how Experts in a topic are often much worse at teaching than people who are experienced beginners or newly proficient in that topic because they rely much more heavily on intuition (intuition which relies on having context) than standardized rulesets that can stand on their own. Being able to write from a context-less perspective, something that’s much easier for newbies (as they are relatively context-less).

So write about an idea while it’s still fresh. You can always edit for clarity and simplicity, but it’s much harder to capture the sharp, poky bits of a topic when you’ve learned to intuitively account for them, and your audience, which is presumably using your post or article to learn something they’re still novices at, will need to know about those sharp, poky bits to be able to stay enthusiastic and feel like they’re not insane for encountering them and having trouble with them.