For months, I have built a new home for my online content: www.turntrout.com. I brooked no compromises. Over 2,220 commits later,¹ I’m ready to publicize.

Welcome to The Pond!

I don’t want to be on LessWrong anymore. Briefly, the site—and parts of the rationality community—don’t meet my standards for discourse, truth-seeking, charity, and community health. For the most part, I’ll elaborate my concerns another time. This is a happy post.

Inspiration for the website

I am pleased to present the initial release of www.turntrout.com—I have many, many more planned features.

Many folks see the first dropcap and think of gwern.net. Some of my site’s features were inspired by gwern’s site, but some others were convergent design choices. For example, I forked the Quartz static site generator, which already included hover previews for internal links. However, gwern’s site inspired inline link icons, dropcaps, linkchecker, and cryptographic timestamping.

The serif font is the open-source EB Garamond—that choice inspired by the beautiful Garamond of ReadTheSequences.com. However, most of this website’s design was by my own taste.

The Pond makes me feel graceful and grateful and proud. It’s my home, and I’ve worked hard towards perfection. I have so much hope and so many plans for this website!

I hope this site encourages me to write more. I miss writing and sharing. I miss feeling proud and grateful to be part of a community. This website will probably not turn into a community per se, because I don’t plan to enable comments. But I still hope that when I write, and you read, and you write back with your thoughts—I hope we can bond and exchange ideas all the same.

What you can find in The Pond

I’ve imported and remastered all 120 of my LessWrong posts. Every single post, retouched and detailed. I both pin down my favorite posts and group the posts into sequences. I’ve also launched the site with three extra posts!

1. The design of this website
2. Can transformers act on information beyond an effective layer horizon?
3. Intrinsic power-seeking: AI might seek power for power’s sake

The research page summarizes my past and present research interests, along with short retrospectives on the older areas.

My dating doc

Like any good trout seeking a mate, I’ve prepared my nesting grounds with care. While trout typically build their nests (called “redds”) in gravel stream beds, I’ve taken the initiative to construct mine in digital form. Female trout are known to carefully inspect potential nesting sites before choosing their mate—and I encourage similar scrutiny of my dating doc.²

Embedded preview of my dating doc

Transparency in reasoning and predicting

I’ve criticized the loose analogical reasoning which permeates the rationalist community’s AI risk arguments. But criticism is cheap. I want to hold myself to my own high standards.

Theodore Roosevelt

“Citizenship In A Republic”; delivered at the Sorbonne on 23 April, 1910

The poorest way to face life is to face it with a sneer. There are many men who feel a kind of twister pride in cynicism; there are many who confine themselves to criticism of the way others do what they themselves dare not even attempt. There is no more unhealthy being, no man less worthy of respect, than he who either really holds, or feigns to hold, an attitude of sneering disbelief toward all that is great and lofty, whether in achievement or in that noble effort which, even if it fails, comes to second achievement. A cynical habit of thought and speech, a readiness to criticize work which the critic himself never tries to perform, an intellectual aloofness which will not accept contact with life’s realities—all these are marks, not as the possessor would fain to think, of superiority but of weakness. They mark the men unfit to bear their part painfully in the stern strife of living, who seek, in the affection of contempt for the achievements of others, to hide from others and from themselves in their own weakness. The rôle is easy; there is none easier, save only the rôle of the man who sneers alike at both criticism and performance.

I hope to speak loudly and carry a small ego. I want to enjoy my wins and honorably acknowledge my mispredictions.

Tracking my mistakes

Inspired by Scott Alexander’s Mistakes page, I’ve written my own. The list is short because I’m still filling it in.

Bounty for bad analogies I’ve made since 2022

Analogies can be useful; analogies can be deadly. For an analogy to be useful, it would do well to highlight how two analogous situations share the relevant mechanisms. For example, an analog computer obeys the same differential equations as certain harmonic oscillators. By reasoning using the “analogy” of an electrical circuit with such-and-such voltages and resistances, we can accurately predict physical systems of pendulums and springs:

However, in AI alignment, folks seem to be less careful. Does “evolution” “finding” the human genome tell us anything about the difficulty of “inner alignment” in “selection processes”? What are the proposed mechanisms?

I think that I am more careful—and that I have been for a while.

$50 bounty for analogies without mechanistic support

Criteria:

1. The analogy was made since July 7, 2022. That’s when I posted Human values & biases are inaccessible to the genome.
2. The analogy is in a post on turntrout.com.
3. The analogy is not supported by mechanistic justification because no such justification exists.

If your claim meets the criteria, I will also credit your name on a list—alongside the called-out analogy. I will comment how I changed my mind as a result of realizing I gave a weak argument. Lastly, if the analogy lacks mechanistic justification but such a justification exists, I will pay $10 and edit the article.

To claim your bounty, submit your find.

Post edits are public and cryptographically timestamped

Each post states when it was published and when it was last updated. The updated link points to the file on my GitHub repo where the edit history can be inspected.

I also took a moment to enable:

Cryptographic timestamping

The design of this website

I concatenate the sha-1 commit hashes of all commits being pushed to main and hash their concatenation with sha-256. Using a slight variant of gwern’s timestamping procedure, I use OriginStamp to commit the sha-256 hash to the blockchain by the next day.

By committing the hash to the blockchain, I provide cryptographic assurance that I have in fact published the claimed commits by the claimed date. This reduces (or perhaps eliminates) the possibility of undetectably “hiding my tracks” by silently editing away incorrect or embarrassing claims after the fact, or by editing my commit history.

Fatebook prediction tracking

By embedding Fatebook.io predictions, I get the benefits of PredictionBook without the clunkiness. Fatebook embeds show how my beliefs change over time and overall encourage me to make more falsifiable predictions.

The future

Over the last few years, my life has lost a certain touch of magic—a touch of aspiration to do better and of excitement to not be doing it alone. I will speak more of this de-magicking. For now, I say: May this site bring back a touch of magic.

☙❧

Find out when I post more content: newsletter & RSS

Thoughts? Email me at alex@turntrout.com