3
points
spenrose
Joined 2,976 karma
- spenrose parentThank you for the correction. I should have said “trained”.
- I can't stop thinking about this. WRT Perl specifically, it’s fascinating how the two competitors adopted Unix shell patterns. Python is handicapped to this day by not automagically snarfing up environment variables, etc. But Perl leaned hard into TECO-style gibberish and the meta-syntax that is regular expressions, confronting beginners with arbitrary complexity. It feels like Wall embraced the system administrator side of coding — the side that has an enormous capacity for tracking corner cases and managing impedance mismatches. Wall was trained, perhaps not coincidentally, as a linguist, a field where continent facts really matter. Guido, on the other hand, was an accomplished mathematician. (This is the Dwarf / Elf distinction from Cryptonomicon.)
- I forgot two, er, three:
9. Python got lucky that its inevitable screwups (Python3) didn’t quite kill it.
10. Swift and Kotlin both define programming as serving the compiler (specifically LLVM) rather than serving the coder’s problem. (I haven’t discussed Rust so far since it isn’t attempting to compete with 98% of Python use cases, but if you squint you can see it as going one step further than Swift and Kotlin and in effect forcing the coder to be a sort of human compiler who thinks in types and memory management. This is not a criticism of Rust, BTW.)
0. And behind all of this is Moore’s Law and the demographic explosion of programmers. Python was an implicit, perhaps unconscious bet that if you served people thoughtfully, the tradeoffs with serving the needs of contemporary silicon wouldn’t matter as much.
- 1. Python was designed by testing syntax with novice users to see what they could adopt easily.[1] > 90% of current Python users weren’t born when it was created. They all had to learn, and Python is the easiest language to learn because Guido and his teammates, unlike $LANGUAGE_DESIGN_GOD, approach the problem as experimental scientists rather than auteurs.
2. Python is conceptually compact, dominated by hash tables with string keys. The initial leader in the ecosystem, Perl, is conceptually sprawling and difficult to reason about.
3. Python also took lessons from the Unix shell, a mature environment for accommodating beginners and experts.
4. Python had a formal process for integrating C modules from early on.
5. Python’s management has an elegant shearing layer structure, where ideas can diffuse in from anywhere.
6. $NEXT_GENERAL_PURPOSE_LANG (Ruby, Go) weren’t enough better to displace Python. Both were heavily influenced by Python’s syntax, but ignored the community-centric design process that had created that syntax in favor of We Know Best.
7. Speaking of open source entrepreneurialism, JavaScript has become a real rival thanks to the Web (and node), but it is handicapped by the inverse failure mode: where Go is dominated by a handful of Googlers, JavaScript was effectively unmanaged at the STDLIB level for a crucial decade, and now it can’t recover. (I’d also guess that having to write a module system that works well in the chaos that is Web clients and simultaneously the Unix world is a daunting design problem.)
8. Python got lucky that data science took off.
[1] https://ospo.gwu.edu/python-wasnt-built-day-origin-story-wor...
- Claude Sonnet 4.5 summary of the original paper [https://www.science.org/doi/10.1126/sciadv.adw1280] for middle school students:
How Earth Got Its Water: A Cosmic Detective Story
The Big Question: How did Earth become a planet with oceans and life, when it formed so close to the hot Sun?
What Scientists Did:
- They used a "radioactive clock" made from two elements: manganese and chromium - Manganese-53 breaks down into chromium-53 over time (like ice melting at a steady rate) - By measuring these elements in meteorites and Earth rocks, they figured out WHEN Earth's basic chemistry was locked in
Key Finding: Earth's chemical recipe was set within just 3 million years after the Solar System formed (that's super fast in space terms!)
The Problem: At that point, early Earth was missing the ingredients for life—especially water, carbon, and other "volatile elements" (stuff that evaporates easily when hot)
Why Earth Was Dry: Close to the Sun, it was too hot for water and other volatile stuff to stick to the rocks that built Earth—they stayed as gas and floated away
The Solution: About 70 million years later, another planet called Theia (which formed farther from the Sun where it was cooler) crashed into Earth:
This collision created our Moon It also delivered water and other life-essential ingredients to Earth
The Big Takeaway: Earth needed a cosmic accident to become livable. Without that lucky collision bringing water from the outer Solar System, we wouldn't be here!
Why This Matters: If Earth needed such specific, lucky events to support life, habitable planets like ours might be much rarer in the universe than we thought.
- Look at your examples. Translation is a closed domain; the LLM is loaded with all the data and can traverse it. Book and music album covers _don't matter_ and have always been arbitrary reworkings of previous ideas. (Not sure what “ebook reading” means in this context.) Math, where LLMs also excel, is a domain full of internal mappings.
I found your post “Coding with LLMs in the summer of 2025 (an update)” very insightful. LLMs are memory extensions and cognitive aides which provide several valuable primitives: finding connections adjacent to your understanding, filling in boilerplate, and offloading your mental mapping needs. But there remains a chasm between those abilities and much work.
- So many articles should prepend “My experience with ...” to their title. Here is OP's first sentence: “I spent the past ~4 weeks trying out all the new and fancy AI tools for software development.” Dude, you have had some experiences and they are worth writing up and sharing. But your experiences are not a stand-in for "the current state." This point applies to a significant fraction of HN articles, to the point that I wish the headlines were flagged “blog”.
- 3 points
- "Since 1995 nearly all of my professional work has been enabled by, and published on, the web." https://jonudell.net
- 9 points
- Some headlines, such as this one, are catnip to up-voters* despite the article's contributing nothing to the established discussion c. 2015, let alone 2025. I don't know how you disrupt this dynamic and redirect to "go read X", where X is _Team Topologies_ or whatever, but it would improve HN.
* (not a criticism; the topic is important to hackers)
- My hypothesis is that about 300 people whose identities were formed by participating in the Slashdot / LWN / etc. communities c. 2000-2005 are active HN participants in 2025. They saw the dream of Linux beating Windows fail—and worse, they saw Macintosh become the high-status alternative to Windows. They saw the Olde Internet of hand-coded web sites be swamped by the arrival of humanity using smart phones, and they hate it. They are like 60 year old sports fans upset about the rise of analytics, or '70s rock fans bemoaning hiphop, or Socrates berating scribblers for displacing orators. Evidence: the 400-point popularity of dozens of recent stories on Firefox minutia—Firefox does not matter, nor does Brave (note: I worked for Mozilla for four years). The many, many stories about reviving the pre-smartphone Web. Probably other topic clusters I am forgetting—Web standards?
- I suspect it's driven by the Olde School Linux / Free Software contingent of HN commenters / voters. Here is an example[1]:
"The #1 story on Hacker News at 2023:08:21T15:41Z is a 2021 discussion of Linux desktop packaging tools. Hypothesis: HN story up-voters are heavily drawn from Free / Open Source Software folks interested in issues that were broadly discussed in "tech" two decades ago (Linux for the desktop!) and are much less broadly discussed today."
That anodyne observation garnered 5 downvotes. I mean, of course it was silly to treat Linux desktop packaging tools as the most important story in tech in 2023! Overall the dynamic feels like Wikipedia: people who participate are atypical, and nothing annoys them more than one's pointing out that they are atypical.
- I argue that we should start calling them "pattern processors": https://x.com/sampenrose/status/1877200883613659360
- "the only partisan issue I've taken a strong stand on through the years is being pro free speech"
Unless I have missed something (possible, LMK), Ben's silence on Musk's partisan use of Twitter is, in fact, a stand. Contrast with John Gruber on Tim Cook and Trump[1]: Gruber's beat is Apple and Cook's choice clearly rises to the level of "Apple agenda item". In fact, you can extend my point to the larger partisan battles among the tech elite. Reid Hoffman is a significant partisan player. Staying silent on that, while wise from a don't-piss-off-important-sources perspective, does a disservice to the truth.
[1] https://www.axios.com/2025/01/03/tim-cook-apple-donate-1-mil...
- Excellent. Wish he had cited Thompson's "Reflections on Trusting Trust" from half a century ago:
* https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_Ref...
- 1 point
- 2 points
- 1 point
- 3 points
- 3 points
- Trade Wars are Class Wars: https://www.amazon.com/Trade-Wars-Are-Class-International/dp...
- What We’ve Learned From A Year of Building with LLMs: https://applied-llms.org