I really don't get having a ton of MD files lying around. And the possibility of having to edit a thousand MD files when the metadata frontmatter changes.
A single SQLite database implements columns/metadata handling, and comes baked-in with FTS and BM25 ranking too.
> LLMs are terrible at accurately summarizing anything. They very randomly latch on to certain keywords and construct a narrative from them, with the result being something that is plausibly correct but in which the details are incorrect, usually subtly so, or important information is omitted because it wasn't part of the random selection of attention.
I don't know what you've been doing, but the summaries I get from my LLMs have been rather accurate.
And in any event, summaries are just that - summaries.
They don't need to be 100% accurate. Demanding that is unreasonable.
The LLM meeting-summary bot in Teams seems accurate… unless you were in the meeting, and also closely read the summary afterward. It misrepresents what people actually said all the time.
Depends on topic, often what they consider important isn't what is important and details that are essential get out of view. I'm having good success with youtube video, not as much with technical docs.
Yes, search and summarization is where LLMs shine. I use them all the time for that, and much less for code generation. I would say search > summarization > debugging > code gen/image gen
The newest Macbooks have insanely powerful hardware (I have an M4 Macbook Max). Yet they do not feel as speedy or instant on my machines with i3. There's always a perceivable milliseconds of latency, with response time from the keyboard to the screen. As someone who has tons of key bindings, I find this tolerable, but it can get a bit grating compared to just how instantaneous everything is on my Linux.
The way sidebars feel is really "sticky". This has got worse with SwiftUI. The List component used for this has notoriously poor performance and a really inflexible API.
The one interesting thing (as a heavy user of both OSes) is that since the past decade there now are plenty of high-quality games (if those count for apps) on the Linux, that still don't work as well or as plentifully on Mac.
Linux is bound to be the number one gaming machine in time; general apps aside.
Yes, that's a game changer. I also wonder why businesses and governments continue to use Windows and macOS. In many aspects, Linux is more intuitive, almost all major software supports it, and the free office software is fully featured, beautiful, and free.
I use CopyQ. Love it because it's so cross-platform, and consistently works across my Mac and Linux machines with minimal fuss; it handles images really well too.
I have some doubts about your statements. If you can't get a job at FAANG or the major startups, sure - that makes absolute sense. Competition is crazy.
If you can't get a job at all, that makes no sense, and probably speaks more to the fact that you might be an academic parrot who can talk code, but not write code, as an actual software developer with practical abilities. And that goes more to your lack of practice, rather than lack of knowledge. The world is not to blame in that regard but you.
Alternatively, try different domains - your skills might work little wonder in the software world, but might move mountains in other (especially labour-intensive) domains.
While I love this insightful analogue, your statement seems exactly like the kind of text you copy-pasted from some LLM, which you then regurgitated to Hacker news with some modifications.
It even ends with that trademark conclusion-style statement... which is a hallmark of ChatGPT output.
FWIW Zig has error handling that is nearly semantically identical to Go (errors as return values, the big semantic difference being tagged unions instead of multiple return values for errors), but wraps the `if err != nil { return err}` pattern in a single `try` keyword. That's the verbosity that I see people usually complaining about in Go, and Zig addresses it.
The way Zig addresses it also discards all of the runtime variability too. In Go, an error can say something like
unmarshaling struct type Foo: in field Bar int: failed to parse value "abc" as integer
Whereas in Zig, an error can only say something that's known at compile time, like IntParse, and you will have to use another mechanism (e.g. logging) to actually trace the error.
I think a better word may be "explicitness". Zig is sometimes verbose because you have to spell things out. Can't say much about Go, but it seems it has more going on under the hood.
I was about to dislike it, then I realized: the author's aggression comes off as... human.
I don't know if the aggression itself is an intentional mockery of the cool, dispassionate, pseudo-poetic tone that AI mostly writes in.
But the passionate visceral disgust is definitely human.