wrs

A lot of Chesterton’s Fence going on here… I remember thinking this way when I still had a pretty low-level understanding of music (like, “I’m reading arbitrary pitches and pushing the matching keys on the piano”), but once I started understanding bigger musical concepts/chunks, the reason for the notation became clearer, and now I see why it’s a pretty good local maximum for readability.

The majority of music uses a system of intervals called the major scale, which is seven notes with a particular spacing that is determined largely by psychoacoustics (that is, we didn’t just make it up). That spacing can be shifted around to any point on the underlying 12-note system, which gives different keys/modes. On any single piece of music you’re typically using only one of those, so those seven notes are the “skeleton” underlying it all. This is what it means to say “this piece is in D major”. (BTW, a fifth is the fifth note in the major scale, so yes, there is a 5!)

Standard notation privileges these modes of the major scale, which means (assuming you know how to play all the major scales on your instrument) there’s a 1:1 mapping from places on the staff to scale degrees. That means most of the time you’re reading straight off the score into the scale your fingers have memorized.

The key signature is what makes this work. You can even instantly transpose by ignoring the written key signature and reading in another key. Chords are also made readable because they’re usually based on thirds, which happens to be the interval that makes the notes stack up exactly next to each other, so a triad, seventh chord, etc. is instantly recognizable.

So basically the notation is optimized for the common case (the music is using notes from a mode of the major scale) and makes exceptions (accidentals) stand out.

And yes, pianos are a pain because all the (musical) keys are different on the (physical) keys. However, pianos are great because all the octaves look exactly alike (so finding all the Ds on the keyboard is trivial). Stringed instruments like guitars or violins are the opposite: all the (musical) key fingerings are the same, but you have to memorize the positions of all the notes because they all look the same. Pick your poison.

A quick way to lose your engineers’ trust is to claim you “trust your engineers” and then unilaterally deprioritize an engineering task.

Not saying there aren’t clueless or myopic engineers, just that you should be honest about the relationship. (And find better engineers or better product managers, whichever is needed.)

I was quite happy when I could type search keywords in, Google searched for relevant pages with those keywords, and I went and read those pages. Haven’t really been happy since they broke it quite a while ago.

Same is true of Google search “summaries” where quite often I click on the link provided and it doesn’t support the statement Google made.

BTW, this post would be more convincing if it wasn’t written in Claude voice itself!

This president’s view appears to be that if it happens in the executive branch he has 100% arbitrary personal control over it. He’s the CEO of the government and he makes the calls. If someone claims there’s a legal reason he can’t do it, he has mediocre but prolific lawyers to deal with that.

He’s worked this way for decades; why stop now just because he’s got a new job?

Any examples of the previous time a major feature was renamed purely by executive order and justified only by a trade dispute?

I have a similar problem, except I needed to ring the actual doorbell. Currently I'm using a HomeKit-compatible video doorbell that communicates over proprietary radio to its wifi gateway that triggers a HomeKit automation (which presumably runs on my Apple TV as the hub) to turn a Shelly relay on for one second to do the same thing as pressing the doorbell button. I think this all ends up just using my internal network, but who knows. Thank goodness the Shelly relay can get a wifi signal while stuffed into that electrical box in the basement ceiling.

(The actual doorbell button is still there too, so it's a crapshoot which one somebody decides to use in the first place. And yes, many video doorbells have a "chime adapter" to take care of this, but our house's doorbell wiring made them run in fear.)

My current theory is that this reflects a weakness in Claude’s ability to see the “big picture”.

When writing code, I have to explicitly tell it how to structure things at a high level, or the result is sort of a flattened spaghetti. Similarly, when it’s explaining things, it’s not good at pulling out unifying concepts and explaining top-down as a smart human would do. It groups little things together but often doesn’t generalize or synthesize explanatory connections from them.

I’ve been experimenting with explicitly working through a sequence of outputs at different levels of detail, but I haven’t found a consistently successful method.

That’s fair, there are cases where the transaction itself has some worth to the buyer for reputational or other reasons, independent of the item being purchased. Those are rather unusual cases, but they happen.

There are contracts that involve exchanging $1 to satisfy some legal requirement of consideration, but the relationship created is worth much more than $1. And of course there are barter transactions where items are exchanged directly without any specific valuation in money involved at all.

APIs, sure, but do they provide the semi-magical consistency, atomicity, durability, and latency guarantees that this depends on?

There’s a trend of doing impressive things by pushing many of the hard problems into S3 and assuming S3 “just works”, then not bothering to explain how S3 works. I guess we do the same thing all the time with other miraculous solutions to hard problems, like assuming that superscalar out-of-order processors “just work”, but in cases like this it makes for an unsatisfying explanation. Especially because S3 is a proprietary product, not an algorithm.

On the contrary, you would be demonstrating that, to at least one buyer, it was worth that. By definition.

I think the argument you’re really looking to contradict is this: if 1% of a company is worth $1B to somebody, then somebody would pay $100B for the whole company. That does not follow (and indeed the acquisition price of a company is almost never exactly its “market valuation”).

“Behind” = less capable, less used, less resourced. Safety and alignment don’t necessarily correlate with market dominance. If they did, no safety regulations would be needed.

A lot of people wanted to talk about this at the time, and the election result was after the training cutoff, so I’d guess they threw this in there to reduce waste due to hallucinations and/or web searches.

There is no valid public interest in having secret police.

Good news, we now know up front what an independent port would cost, and it’s not much. So no reason AWS, Google, and friends couldn’t bang out their own port if they want, binary-compatible with this one.

What we don’t know for sure is whether there is any copyright for LLM-generated code. The license might be irrelevant!

That question makes me think about Boeing. Or NASA. Or Enron. Reality always wins, no matter what management and Investor Relations says.

I have seen elevators, usually in a big building with a lot of elevators, that do act that way (presumably using weight sensors?). I’ve read that some elevators also have a “kid mode” that does the opposite: if some kid pushes all the buttons as they leave (just for the lolz), the elevator cancels the calls.

That is just RDMA itself and seems to be aimed directly at distributed inference (no surprise there!). Now they need to update the SMB driver to make use of it.

I don’t think “you” in these titles ever really refers to an “average reader”. Some familiarity with the field is required. Imagine how non-programmers (and many programmers) feel about some examples I just Googled:

“You Could Have Invented Parser Combinators”

“You Could Have Invented Container Runtimes”

“You Could Have Invented Git”

Given the references to “mathematicians”, I think this reaction is more about an unfamiliarity with the concept of applied mathematics, which is ironic for practitioners in a field containing so much that is (or should be) regarded that way.

Software used to be all about “discrete math”, logic, maybe a bit of queuing theory…and now suddenly linear algebra and statistics became important. Don’t panic, it’s just another textbook on the shelf.

A few models ago it was not unusual to find that Claude had written tests whose assertions looked correct, but were so thoroughly mocked out that they ran no real code at all — a Potemkin test.

I haven’t seen it do that in quite a while, but it was an interesting failure mode!

If you look at the reasons for rewriting Bun [0], they weren’t about efficiency and they don’t apply to Claude Code. It’s already written in a memory-safe language, and doesn’t have to interface to a library with a tricky set of invariants.

[0] https://bun.com/blog/bun-in-rust

In the specific case of ship retrospectives, I prefer to call them “postpartums”.

But why isn’t it called WARPA now?

I think you misread the article. The skepticism isn’t about whether it happened. It’s about whether it’s a reason for models to be locked behind “trusted partner” firewalls. Offensive hacking capabilities are the same as defensive. Hugging Face had to use a Chinese model to defend against this because they weren’t allowed to use OpenAI models to do it.

Yes indeed. Sorry if I was unclear in my wording...I meant:

* Fretted cellos would have the same fingering positions as unfretted cellos (frets are placed where the notes are, they don't set where the notes are).

* All parallel-stringed instruments have modular fingering patterns like this (though not exactly the same fingering patterns, if their string interval differs) so this is a good principle to try no matter your instrument.

Learning both guitar and cello is kind of like learning both QWERTY and Dvorak. Same intellectual principle, whole different muscle memory.

They say they started in 2002, but same here, I've had hundreds of domains on Dynadot since the early 2000s with no problems.

Why would you think Analog Science Fiction isn’t on the internet? It’s right here in the obvious place. [0]

[0] https://archive.org/search?query=analog%20science%20fiction%...

While cellos, violins, etc. don't technically have frets, the notes are of course in the same layout as instruments that do have frets, so this same method works.

(It's funny how different instruments live in their own worlds. Once I showed a movable scale fingering diagram for guitar to a cello player and they had no idea what I was talking about, even though the cello has the same kind of patterns, as would any parallel-stringed instrument. Another cello player did know about scale patterns but called them "universal fingerings".)

No, the diagram looks like a binary literal but it's actually backwards from that. Lane 0 becomes the LSB, but that's on the left of the diagram.