Hacker News Reader

Me.

I was the conceptologist. I learned by fitting things into chunks.

Then I got to the world of tables of apis and frameworks and shitty microservices and all of it ad-hoc, and found that my mind, tuned to do what I found valuable and beautiful in the world, was worthless.

That's how I thought until I started going deep into some technologies that have very strong technical foundations:

- Databases: Postgres, ClickHouse - Back-end frameworks: Ruby on Rails - Front-end frameworks: Elm

I lost count of the times I was reading the documentation of these projects and had my mind blow thinking "The person that came up with this concept was a genius! I'm glad I get to learn it."

This is how I feel about FeathersJS. It just feels like a really sane way to design an API

Wow. That’s exactly how I feel. I’ve always focused on concepts over specifics, thinking that I’ll figure out the details as I do a thing. People don’t seem to want that, though. Maybe I’m looking for the wrong kind of work.

I think I've solved this for myself. I'm a concept-first learner too, but don't have that much of a problem learning ad-hoc details of random libraries and tools. I think what's going on is that, at some point early on, my brain figured out that if the thing doesn't seem to fit a coherent, sensible concept, then there's likely a story explaining the delta. I started to think mostly in terms of guessing what kind of problems and tradeoffs the author faced, which would explain the peculiar design. I found myself searching for information about past iterations, previous use cases - anything that could let me correctly guess at the story behind the seemingly random thing.

This works for me pretty well in software. I still have problems with things requiring memorization, though - like some bits of chemistry, or the grammar of German language.

I'm incapable of retaining the ad-hoc specifics. Most of the shit they want me to learn I use once and then it's on to other shit. I'll never improve like that.

Please let me know if you ever figure out what kind of work fits...

I feel like the move would have been to go very deep on something. Too late for that now.

Wow, this sounds like me. I'm good at conceptualizing and rarely needed to study for my comp sci classes. I was actually decently skilled when I first started, working on monoliths. Now I'm struggling because of poorly documented apis, incomplete requirements, and overbearing tech debt or messy technical design. If only I could find a job with an orderly structure.

Right???

I started out in a 300 kloc c++ codebase.

I've never had a better environment since =/

This is exactly how I feel, and why I don't understand why tech interviews are the way they are right now

> Wait a second, who here honestly did not practice on-demand learning at university? You did, didn't you?

I have no real ability to do that, and even when I try to cram it feels like cheating (take that up with my therapist.)

What I did in college is figure out the classes I was likely to take next semester, then furiously study before the semester started until I had a learning buffer of a few weeks. Then during the class I tried to study deeply, always ahead of the class and exams, and ideally the buffer would be zero by the final.

I did the same thing. I'd always try and read ahead so I had read the material BEFORE the teacher covered it. This was an immense help in understanding. Seriously, it helped so much.

I was always surprised that my fellow students didn't do any reading or note taking before the teacher covered something in class.

The textbook was usually better than the teacher anyway.

> You had to because you were studying for a specific exam, which you would fail if you only understood the deeper concepts but not specifics of the questions on that exam.

I mostly operated that way on a class-by-class basis in grad school, but at some point it became a conceptual understanding. The "aha" moment came in grad school when I was a TA for a class I didn't do particularly well in as an undergrad. During a help session, I couldn't recall a couple specific formulae, but I was able to infer and derive them based on my conceptual understanding of the topic.

I think the real difference was that I was now teaching the material, and in doing so, had to approach things from a totally different angle than studying for the test.

Neither of those modes alone suffice. At an undergrad course you will have the concepts all forced at you in a class, so that later when you are learning on demand they can click.

Love the analogy. For years after school I found myself going into "test mode", a timeboxed intellectual fight to the death. Works great when the work gets distilled to a number and thrown away, not so well when you have to keep working with your results...

When I studied for my Applied Physics degree there were exams twice a year to weed out those who couldn't hack it but these exams could contain questions on absolutely anything that was part of the whole course.

However the class of one's degree (first, second, etc.) depended only on the finals and the oral examination of the final year experimental work. The intermediate examinations counted for nothing except to justify your continued study.

On demand learning doesn't work in such an environment you have to have simultaneously a broad and a deep understanding of the system of physics not just isolated bits of it.

I thought the OP was silly because surely everyone knows that you have to do both.

School grades is a case of Goodhart's Law, as soon as you try to get good grades the grades stop being a good measure of your learning.

Interesting to see a lot of deeper concepts-focused people in here. I'm the opposite. I have to start from on-demand or nothing will stick. I work best when I can tinker first, then dig deep and get explanations for things I actually observed, roadblocks I encountered, etc.