As for queries, you might be able to achieve the same thing with usage of command-line tools if it's a `sqlite` database (I am not sure about other SQL DBs). If you want even more control than the settings.json allows, you can use the claude code SDK.
I don't know; I've never done something like that. If no one else answers, you can always ask Claude itself (or another chatbot). This kind of thing seems tricky to get right, so be careful!
If you control the ssh server it can be configured to only allow what you want. Certainly tedious but I would consider it worth while as it stands with agents being well, agentic.
Your story sounds similar to mine. There are some parts of programming at which I know I will never excel. I also don't have time in my life to spends lots of hours outside of work developing my skills. I think it's important to realize that the median software engineer is probably not doing these things either. Maybe the top 10% are? Something like that would be my guess. It's okay to not be in the top 10%!
This is refreshing to read. Sometimes when I come here and look at the posts and comments it seems like lots of people are doing lots of things that are confusing to me. I'm recently coming to terms with being okay about not being able to learn everything that I don't understand and outside of work I've started pursuing non-programming related hobbies which led me to make the comment I did.
Beauty is in the eye of the beholder. My personal taste for the presentation of a piece of writing is that less is more. I usually find artwork that accompanies a text to be distracting. I love reading work that can stand on its own, invoking images in the mind. I also dislike animations that seem to be made for a certain scroll speed.
Having said all of that, I certainly don't think it's bad, nor is it a commentary on the arguments being made. It's just not my cup of tea.
> I usually find artwork that accompanies a text to be distracting. I love reading work that can stand on its own, invoking images in the mind.
But the images are a part of the work, not separate from it, no?[0]
You might have a preference against that, which is absolutely fine, but I think you're making an artificial distinction.
[0] There's obviously a separate conversation to be had about how much that part contributes or detracts with any such work, but the point stands that I tend to view such works as all of a piece including all constituent parts.
> My personal taste for the presentation of a piece of writing is that less is more.
TFA works with iOS reader mode, which is all that matters to me. I use it instinctively as it makes style more or less uniform and lets me focus on the content of the article.
I think when you make such strongly opinionated design decisions on your website, you're deliberately inviting strong criticism. They could have used a readable vanilla bootstrap theme and HN would be actually discussing the actual text content instead of the design, but they didn't, and here we are.
The idea that opinionated design is intended to court controversy or criticism is, itself, very cynical. The corollary to that is that all design should be vanilla to make it as unobjectionable to the widest audience possible.
Design and content are inseparable. When design reinforces the point of the content, that is good design, even if it's ugly, even if it's not aesthetically pleasing to you, even if it's not how you'd do it.
But I'd argue that questing for neutrality is worse than taking a stance, even the wrong stance. Besides which, what one now considers "neutral" is also a giant set of design decisions - just ones made by committees and large corporations, so the blame for its drawbacks can be passed off, and there's plausible deniability for the designer.
Someone takes risks and makes something creative they consider artistic. You're reducing their choices to a question of whether they intended to be popular or to court criticism, flattening the conversation into one about social media credit, and completely discrediting the idea that they had true intent beyond likes and points. That response itself betrays something slightly cowardly about the ethos of neutrality you're proposing.
Actually, HN wouldn’t be discussing it at all, most likely. At least not this much. The design is not only good, it has also successfully incited a passionate response from a bunch of people who don’t appreciate it. Win-win!
That is correct for the collective as a whole, but in his instance, if this wasn't connect to a public github, it would have been substanially more difficult to prove he used a LLM.
> - Is the work faster? It sounds like it’s not faster.
The author didn't discuss the speed of the work very much. It is certainly true that LLMs can write code faster than humans, and sometimes that works well. What would be nice is an analysis of the productivity gains from LLM-assisted coding in terms of how long it took to do an entire project, start to finish.
This is total speculation, but my guess is that human reviewers of AI-written text (whether code or natural language) are more likely to think that the text with emoji check marks, or dart-targets, or whatever, are correct. (My understanding is that many of these models are fine-tuned using humans who manually review their outputs.) In other words, LLMs were inadvertently trained to seem correct, and a little message that says "Boom! Task complete! How else may I help?" subconsciously leads you to think it's correct.
My guess is they were trained on other text from other contexts (e.g. ones where people actually use emojis naturally) and it transferred into the PR context, somehow.
Or someone made a call that emoji-infested text is "friendlier" and tuned the model to be "friendlier."
I suspect that this happens to be desired by the segment most enamored with LLMs today, and the two are co-evolving. I’ve seen discussions about how LM arena benchmarks might be nudging models in this direction.
I think it's not only the potential for self-improvement of AGI that is revolutionary. Even having an AGI that one could clone for a reasonable cost and have it work nonstop with its clones on any number of economically-valuable problems would be very revolutionary.
This is a key question in my opinion. It's one of the things that make benchmarking the SWE capabilities of LLMs difficult. It's usually impossible to know whether the LLM has seen a problem before, and coming up with new, representative problem sets is time-consuming.
If you fuzz the names they won’t mean the same thing anymore, and then it’s no longer the same test. If you remove the whitespace the LLM will just run a formatter on the code. It’s not like the LLM just loads in all the code and then starts appending its changes.
I've never had a LLM try to run a formatter on my code with probably a few thousand hours logged driving agents (driving 4+ agents at once in most of those). Fuzzing makes the semantics slightly less immediately obvious, but LLMs are more robust to this than you or I, the biggest difference is the reduction in memorization carryover. If it feels like too different of a test for you, not sure what to tell you, but I know the world would appreciate a better way to test for training set contamination if you can figure one out.
I've done it? I have a benchmark called scramblebench that will do rewriting to evaluate model performance degradation with symbol replacement and layers of indirection.
Jonathan Haidt has a lot of good material on this. He is leading the charge in encouraging parents to delay giving their child a phone until high school and not allowing them to have social media accounts until age 16.
How do Asian countries and top-performing countries deal with this?
We should do whatever they do.
On that note, we should also segregate kids by academic desire and achievement like Japan and China. The bullies and underachievers hold back those who are academically excellent. We do this in limited instances, but not enough to really count.
In Japan and China, high-stakes entrance exams come earlier and play a stronger role than in the U.S. In China, the zhongkao (high school entrance exam, around age 15) and gaokao (college entrance exam, age 18) largely determine access to selective schools and universities. In Japan, competitive entrance exams for high schools (age 15) and universities (age 18).
That's really underselling it. Gaokao determines where you can live, where you can work, who your friends are, occasionally how much your family values you. They shut down airspace and conduct military/police patrols during examinations to sniff out cheaters. It's only the very wealthy who can just uproot their lives and send their kids to an Ivy/Stanford/Oxbridge/MIT and just skip the whole thing.
Responding to the OC, this is a downright awful solution to the current education problem in the U.S.
I grew up a white kid in a very (90+%) Asian community. IMO, the biggest difference I observed comparing my white friends from other communities to my Asian friends in my community was the expectation of excellence. For the Asian kids, either they were succeeding, above and beyond, or they were a failure. "B is for 'Better not come home tonight', A is for 'Adequate'", as the jokes went.
And some of those kids still struggled. But the response was to push harder. Didn't get adequate grades that school year? You're not doing anything fun this summer, you're studying. Needless to say it was a culture shock going to college and meeting people who were shockingly cavalier about potentially failing classes.
There were some responses about educational expectations, but I would love to hear how folks in these Asian countries specifically deal with cell phones, social media, and these general media/online distractions.
Cultural pressure towards education, and phone bans left and right. Also, people are still addicted to their phones, including kids. But more controlled, I guess.
I only know through cultural osmosis and not real data but it sure seems like the expectation is for the kids to be up till midnight grinding away on homework.
As someone with difficulties early on in life and thus showing behavioral issues (what you describe as bullies and underachievers), I went through a system like this and I despised it. N=1 but segregating children at early age based on the behavior they're showing, i.e. the difficulties they're having, felt kinda cruel. It worked academically I guess, I ended up ok, but for many it just meant they just simmered in an environment of mediocrity and rarely made it out.
I get it especially with younger ages, but on the other hand if the student is persistently disruptive they should be removed for the sake of the other students. It's also unfair that 1 student hinders the education of 20+ others.
I do agree there's a disparity between educational outcomes in men and women - but I don't think you can immediately draw your conclusion:
Baked into it is the assumption that current education models fit both genders equally. Boys respond better to active learning and competitive techniques than the more passive techniques used currently. (Could we just as easily draw the opposite conclusion if our current educational culture was geared towards boys?)
Another thing to consider is the various programs that incentivise/enable girls to get into various subjects (in my n=1 experience I had much fewer programs (programming, robotics, maths, etc.) to join despite being already very interested and strong in those subjects).
By comparing age groups directly we are also not controlling for the fact girls mature faster making them better students earlier in life. We are also not considering tail effects of a normal distribution: e.g. top 5% of all students are male, but majority of students in the top 50% are female.
Maybe the solution is to segregate schools on gender, but that doesn't immediately equate to boys crashing and girls excelling.
I agree and I don't think gender-segregated schooling is a wise idea. But the argument is Kryptonite to those who favor school segregation because they realize that they more likely than not would end up in the loser group. Works wonders on race baiters too, who has to come up with "reasons" why girls beating boys is the result of "unfairness" while whites beating blacks is "natural".
As I said, there is a vast overlap between boys and girls. Boys even do better in some subjects, notably mathematics and (some) sciences[1].
In the same way that if we streamed per-subject, there'd still be a significant number of girls in the top set for maths, if we streamed by performance overall, there'd be a lot of boys in the top schools.
Nothing about streaming implies gender segregation, so I'll ask again: do you have another reason for being against it?
That's why I wrote "generally". There are many countries, subjects, and years of education to compare so you can always find some statistic that bucks the trend.
As for queries, you might be able to achieve the same thing with usage of command-line tools if it's a `sqlite` database (I am not sure about other SQL DBs). If you want even more control than the settings.json allows, you can use the claude code SDK.