This was beautifully written. I give it +4.
I like that it includes opposite examples. Is nature gentle, friendly, and harmonious? Or is it indifferent, hostile, and murderous? This is the wrong question to ask.
When I first read this post, I was already aware that "I believe in X" could mean either "I believe X exists" or "I support X". I reacted to this by mentally translating "believe in" to "support" to remove the ambiguity.
It is not hard to see where this bucket error comes from. It comes from religious faith. The Apostle's Creed begins "Credo in Deum Patrem omnipotentem" which translates to "I believe in God the Father almighty". Here "belief in existence" and "support" are collapsed into a single concept.
... (read more)I believe in God,
the Father almighty,
Creator of heaven and earth,
and in Jesus Christ, his only Son, our Lord,
...I believe in the Holy Spirit,
the holy catholic
Here is my system prompt. I kept asking Claude "Evaluate and critique the user custom instructions." until Claude ran out of substantive criticisms and requests for clarification. Claude said of the final version:
What works especially well
The sycophancy/bluntness section is excellent.
I don't know whether this is a good sign or a bad sign.
The prompt is long, but tokens are cheap now. I haven't done any systematic testing of different prompts, or even of my prompt vs no prompt. A lot of it evolved over time in response to specific errors. I copied parts of it from Gwern, and the anti-sycophancy section is influenced by Zach Davis.
... (read 1609 more words →)My operating system is [REDACTED].
I usually code in
For the third passage, he was not comparing the USA in general with Germany in general. He was comparing the stagnation of progress in theoretical physics:
Dirac is here, which benefits the physics here greatly, although it does not remedy the depression of theoretical physics. —
...
The general depression that prevails here and is certainly sufficiently well known to you, nonetheless, measured by European standards, it is still "God's own country". Somewhat frightening, though, is that it looks roughly like Germany in 1927, so that one does not quite clearly see whether the amplitude of the depression is different, or only the phase. Perhaps it is the amplitude after all.
In February I will be back in Europe, and at the end of April in Berlin.
The idea of "swapping people" only makes sense if you believe in an immaterial soul. Then you could grab John's soul, and stick it into me, and grab my soul, and stick it into John, and now I am John, and John is me.
But if, like me, you think that you are a mathematical mind that is physically implemented on a human brain, this idea of "swapping people" is incoherent. There is no way for you to turn me into John or vice versa. If you rearranged all of my atoms into a perfect copy of John, there would be two Johns and I would be dead.
Strong upvoted. Is it missing the ending, or
There is no evidence showing that trans women outperform cisgender women by any significant margin [...] trans women get well within the expected ranges for cis women within around 3-4 years.
A meta-analysis of 24 studies found that transwomen were still stronger and more muscular than ordinary women after 3 years of hormone therapy:
In transwomen, hormone therapy rapidly reduces Hgb [haemoglobin] to levels seen in cisgender women. In contrast, hormone therapy decreases strength, LBM [Lean Body Mass] and muscle area, yet values remain above that observed in cisgender women, even after 36 months. These findings suggest that strength may be well preserved in transwomen during the first 3 years of hormone therapy.
Also, it... (read more)
Well, sort of. Your comment provides evidence for habryka's claim compared to a counterfactual of 0 commenters finding their way here from that thread, but it provides evidence against hybryka's claim compared to a counterfactual of 2 or more commenters finding their way here from that thread.
Overall rate of engagement remains low. It was stuck at 11 comments for 21 days before your comment, and has been stuck at 47 karma for at least the last 19 days. Karma and comments are imperfect proxies of readership, but they are what we have. In the counterfactual world where this post wasn't unlisted, I expect engagement would have been far higher.
Seems like we mostly cleared it up
Yes, your reply makes your position clear. I don't feel like taking the time to edit my comment, but thank you for offering to edit in any changes.
Also, you definitely have my sympathy for the amount of time you have burned on this! I would not want your job.
tl;dr: My recent Question post was censored[1] by habryka. The post was critical of habryka's decision to ban Said Achmiz on current evidence, and requested additional evidence.
I recently read through all 10,309 words of habryka's blog post explaining his decision to ban Said Achmiz. I also read through thousands of words of comments. I did not find the examples given of Said's behavior a compelling reason for a ban, but I did identify a key crux which would justify a ban:
... (read 768 more words →)Said has been by far the most complained user on the site, with many top authors citing him as a top reason for why they do not want to post on the site,
To be honest, I can't remember using the technique a single time. The highest-value thing I read recently was object-level: Zvi's post on how to use Claude code, the most useful part of which linked another guide.
I consider the consequences of my actions all the time, both consciously and subconsciously. So perhaps I already have a "good enough" version of this technique running. The likely outcomes of my actions are usually obvious; I rarely have cause to step back and ask myself "what are you about to do and what do you think will happen next?"
When I ask myself, "In what situations would you benefit from being more aware of what you... (read more)