vishalblog jr.

Gatekeep, Gaslight, Boyboss

As part of my duties as Head Gatekeeper for Inkhaven 3, I had the privilege of reading over 200 applications. I agonized over each one, having read just about every submitted piece (yes, even the godforsaken 100+ page dissertations on gender dysphoria, climate policy, and John Rawls). Though this was not always a pleasure, it was consistently engrossing, and the last seven weeks have taught me much about the inner life of the upright ape.

It hurts to be rejected over your writing. Such judgments are (by necessity) unusually invasive. In trying to judge the writing, I had to judge the writers too (for, in the end, we always write about ourselves...). So I put my BPD(male) brain into overdrive trying to model the people behind the prose, and sometimes I did not like what I saw. There are so many ways to fail to write well—many directly proceeding from the depths of the human soul. Your writing can fail because you are:

Due to the flood of inbound interest in the residency, I am unable to give personalized feedback to rejected candidates (despite having robust models of each of them still sitting in my head). It'd just be too many emails to write. Moreover, it's not easy to transubstantiate aesthetic judgments into actionable feedback (somehow, "you simply write worse sentences than eighty other applicants" is not what anyone wants to hear). So instead, I'll offer some broad notes about how I'm thinking about rejection, and discuss in detail the most common failure modes.

The Artificial Difficulty of Selection Processes

Inkhaven 3 received more applications than Inkhaven 2, and the median applicant quality went way up. I don't know why this is. Maybe it's because the labor market is pushing more people into trying out funemployment. Maybe it's because Inkhaven's reputation has waxed thanks to word-of-mouth. The most likely explanation is the so-called "Vishal Factor" (a.k.a. ME! God's favorite). In any case, there were many candidates (perhaps a full cohort's worth) rejected from Inkhaven 3 who would have been happily admitted to Inkhaven 2. Them's the breaks. This really drives home for me that the difficulty of a selection process is really sensitive to initial conditions: being better than 140 other applicants is just a lot harder than being better than 80.

At the same time, I feel acutely that the recruitment pipeline for Inkhaven is currently too narrow. I want to create Great Writers, which means we'll have to sample the logosphere widely. I believe the desired number of applications per Inkhaven sits at around a thousand (which would result in a brutal 4% admission rate). If that were to happen, it'd be unlikely that Inkhaven itself would improve in a proportional manner. I think many Effective Altruist talent pipelines have this problem (e.g., MATS underwent a prestige flywheel, and now it's very hard to get admitted despite the program itself not increasing much in quality).

Gatekeepers Gatekeep

I led a five-person gaggle of gatekeepers; four of us read every application. My aesthetic priorities are quite different from the rest of the team. Despite having a polemical brain, I am a very patient reader. I'm okay with writing being tedious, as long as there's an occasional gemerald buried somewhere in there... As a matter of principle, I tried to make sure that every submitted word activated at least one of my neurons. My colleagues did not share this approach, and would (with inappropriate frequency, I must uncharitably say...) skip pieces from unpromising applicants.

I would sometimes be baffled (vexed! furious!) by dissenting opinions across the review panel. Sometimes you will get told things like, "I was totally disgusted and appalled by X's writing, but that's a good thing actually" and all you can do in reply is sob and move on.

Some of my gatekeepers have written about their gaslighting and boybossing, so you can read them explaining themselves too:

Despite all this, we agreed unanimously with surprising frequency. Yes, indeed, art is "subjective," but even highly baroque and alien quality gradients can become legible given enough samples.

Typso

It is really hard to get into Inkhaven if your pieces contain many typos, grammatical errors, and solecisms.

AI Slop

Sometimes people get AI psychosis, and their writing starts Pangramming 40-100%. Sometimes you tell people up front that you will reject them immediately for this, but they don't listen because Claude just feels so fucking good. Unfortunate.

Bureaucratic Ideologues

I was very much prepared to admit mediocre writers who are working in important cause areas, such as alignment or animal welfare. Sadly, these topics are attractors for abominable writers.

I was particularly disappointed by the animal welfare applicants, who were almost uniformly terrible. Frequently their writing was soulless, bureaucratic, and focused on respectability politics. Some of it was actively manipulative. For instance, one piece from an EA argued that banning foie gras is a good cause area because:

  1. the American public already finds it repugnant
  2. it's a luxury item (and therefore difficult to defend against populist critique)
  3. winning a battle over it would be a good way to raise activists

This reasoning strikes me as cynical and political in the lowest sense. A welfare-oriented analysis of animal suffering would not prioritize foie gras bans on the merits.

Sometimes people fall into the trap of thinking that just because they're talking about something inherently important, they do not need to say anything original or interesting about it. Many applicants who write about AI make this mistake. Pal, I already read LessWrong. LessWrong does not need another "Introduction to Mechanistic Interpretability" post. If you really feel you need to write that, you should write it for a new audience (e.g. the New York Times readership, or right-wingers), and you have to write it well. Deconfusion is important, but that alone is not interesting for most readers in most cases.

But who asked tho?

Many applicants wrote extensively about their own motivational systems, and dispensed a lot of advice. This was irritating, particularly because many of these applicants have executive function issues, and frankly should not be giving advice to anyone.

Therapooming

Oh boy, the therapy-speak...

I admitted a number of eccentrics. Part of being an eccentric is being a hot, wet mess, and being honest with your readers about how dysfunctional you are. It's hard having a pnuss, and many of us (all of us?) turn into fools trying to get someone to touch our pnussies. It's good, and human, to be honest about this.

Sometimes, though, the meme can go too far.

There were multiple applications (from both men and women) that illustrated that their authors had a weak grasp on the concept of consent.

For example, one male applicant submitted a piece about how his friends have started warning him he is gaining "a bad reputation" for generally radiating too much unwanted horny energy. I thought it was bold of him to submit this piece, but perhaps it indicated an enlightened capacity for self-reflection... However, I shortly thereafter discovered that all five of us gatekeepers had independent personal anecdotes of this applicant being horny in a bad way (e.g. walking into people's rooms uninvited), so I rejected him and banned him from Inkhaven. Sorry, but I don't want to put you next to women for 32 days.

As another example, one talented female applicant submitted multiple pieces about her bad sexual experiences, about how she keeps on not saying "no" to men touching her in ways that make her feel assaulted afterwards. She talks about how disgusting, stupid, and arrogant these men are, how degrading their touch is. It was really excellent writing, but I don't want to be around it. For Inkhaven, I need to rely on residents being able to say the word "no" when they start feeling uncomfortable sexually. Sorry, but I don't want to put you next to men for 32 days.

Philosophy vs "Philosophy"

I'm a pretentious shit. I love my canonical authors, particularly my philosophers, and enjoy quoting them liberally in my pieces. This is a bit dangerous though, as you run the risk of trading in banalities and sophisms. The joy of analytic philosophy is that it consists of trying to argue as precisely as possible about preparadigmatic areas of thought. Sometimes writers (particularly ex-STEMlords, newly-intoxicated by wordceling) start playing with philosophy, but forget about its precision and just wallow in its preparadigmaticity. Conq (one of my gatekeepers) calls this "throwing banalities up at my screen." One applicant in one of his pieces sadly chose to invoke in quick succession "nothing human makes it out of the near future," "we must imagine Sisyphus happy," and "the medium is the message." A touch inauspicious...

Don't Worry...

It's fine though. Just keep writing (and reading too, you narcissistic fucks). The more you read, the more you see that everyone is bad at something. All you can do is try to be, uh, Less Bad.