Fusion Energy Will Power the AI Boom [video] (youtube.com)

- I read tons of technical books and manuals. I don't remember all the details of course, but if at work some problem seems related to a given algorithm/situation that I have read in a book, then I know where to find more answers. I have read pretty much all the "books every software engineer should read" and more esoteric ones

- very good Google search skills (now in the AI era, I use LLM as tools just the same way. For me LLMs are just a simpler way to Google things). So, if I need to solve a given problem, I know how to search for previous solutions. This in combination with all the tech books I have read, allows me pretty much to come up with a world-class solution to most of the problems.

- I'm not good at coming up with a solution fast and just by myself (no books, no Google search). I just struggle. I know how to think algorithmically and architecturally, but for implementing a solution I need to rely on external sources. Also, I think better by not being in front of the computer; I either write things down on a piece of paper or I go for a walk or a run to think through the problem

- On the other hand, I can debug code very fast with no external sources needed, just the debugger. Of course, if the compiler/interpreter is throwing at me a very specific error message, I will Google it

- I don't consider myself to have a lot of "raw" intelligence. But in most software engineer jobs one needs to come up with robust solutions, not new solutions. I know how to build robust solutions (because I have seen them before, or I can find them)

I imagine companies are most interested in hiring intelligent people rather than people who are good "searchers" (for a lack of a better word).

And because I'm good at searching, I know how to pass tech interviews (most of the time I know what they will ask, so I can find the solution and improve it). I don't pass interviews because of some clever insight that came to me right away. One problem I see is that more companies are moving towards a "live" coding interview, and of course they don't let you Google anything. I struggle with such interviews.

Comments (7)

uberman · 1d ago

For what it is worth, I would rather hire for good problem solving skills than "smarts".

dakiol · 1d ago

I understand that. But one can have good problem solving skills by using tools (like books, search engines, etc.) right? I find it unrealistic to test for problem solving skills using "just" one's brain.

One's brain (my brain actually) is quite slow, doesn't have memorized tons of things (it has only pointers to actual sources of knowledge, like books/manuals/open source projects/etc.), so yeah, I'm a bit like a tortoise if you leave me alone with my brain only. But I can be Achilles if you let me use books and a search engine, if you let me read manuals, etc. I understand as well that all of this is not feasiable in just 1h of interview, so yeah, that's life.

uberman · 2h ago

I don't really care too much what tools one integrates into their repertoire and I personally think being "good at good" is a valid problem solving skill. By extension, being "good at llm) seems like a valid skill as well.

What I would say is the velocity matters though, so not being confident unless reading a book sounds like an impediment but that could be my own bias.

For what it is worth, I could care less if someone can write code to balance a tree in their head. That extends to a large number of competitive coding questions. I would rather someone use a reference than confidently introduce an off by one bug.

taylodl · 1d ago

You sound like a typical, modern developer to me - and that's not a slam, it's a compliment. I've been developing software for 40 years now and have always worked the same way you do. The only thing that's different is back in the day we didn't have Google, so we relied on books and magazines. We also weren't working with 25 different technologies over several different layers comprising a solution. The world was a lot simpler back then.

Today's LLMs could completely create everything we were creating back then, but guess what? The goalposts shifted. Like they always do. What I've noticed is the more power we developers gain then the more we're expected to deliver. What we now deliver in a sprint would have taken months or years to have delivered back in the day. That's why I'm not worried about AI. It's just another tool to replace developers that'll end up moving the goalposts for what we're developing. Within 10 years a single person will be able to deliver in a day what now takes a team several sprints to deliver. And we'll still be drowning in work.

sfmz · 1d ago

You can train on leetcode sites if that's your weakness.

dakiol · 1d ago

I do, sometimes. But that wasn't the intention of my post (to criticize the current state of tech interviews). My goal was perhaps to see if more engineers think like me (and we all are just pretending by memorizing leetcode exercises because we need to pay the bills) or perhaps among developers, intelligence is rather higher than in the normal population? So, in my case I would be behind the bell curve peak I suppose, but maybe I have been working with highly intelligent people without noticing.

AnimalMuppet · 1d ago

I'm not pretending. I just don't do leetcode. If that's what the company is looking for, hire someone else.

I'm not sure this is the term you're looking for, but I might think of myself as a "deep" developer rather than a "speed" developer. (Of course, for the kind of code that needs a developer like me, I'm faster than the speed people...)