frigg
frigg on Hacker News
You don't always design bigger systems from scratch as greenfield applications, most of the time you are delivering smaller features in incremental ways or fixing bugs. Even for greenfield projects, you can plan with the LLM and come up with a design you like at the high level but that still does not spare you from knowing (or not) how lower levels work, which is OP's original complaint.
In essence you basically say that you don't read or review the code produced in detail, you just know the high level design of how the system works.
So how did you make it stop?
>but the shift has never been as fundamental as it is today with AI.
How are we so sure? I'm not old enough to have lived through every/many big shifts in our industry but how do we know this is not what people who lived through the compiler-shift era felt? Or people who saw higher level languages like C or C++ be created, or the punch cards getting replaced with a digital display?
This AI stuff feels magical and so much bigger than other shifts, but maybe only in retrospect.
>I remember reading an essay by Richard Stallman calling all this kind of stuff out in the late 1980's and I kind of hate that he got it so right.
What essay is that?
I was also disappointed and stopped reading mid-way. There is, unfortunately, no original idea or thought in this post.
PayPal is the only real company to offer any meaningful support to customers targeted by fraud or in a real dispute with a company. Banks screw customers and for cards you are at the mercy of mastercard/visa and a long wait period for the dispute.
>Wouldn’t you kind of expect to be fired if you release a project under your employers name that’s not even associated with them and hasn’t been cleared?
It seems to be something other people do as well and not out of the ordinary, so no.
You can't undertand because the argument is nonsensical.
So they basically fired all the interns? Can anyone who works or knows someone who works for Cloudflare can confirm?
The models have already plateaued, you don't need latest and greatest.
Yes, even better sometimes.
>they're just as capable as using AI as anyone.
This is a very naive take. AI output still needs to be guided and iterated over, architecture decisions (even at the code level) still need experience to judge correctly.
>battery electric vehicles
>if it is the wrong thing you produce
Says you and not the guys with the big pockets that actually do market research? Until you solve the charging times problem, the mileage problem and the amount of available charging stations problem you will not see wide adoption. Most people don't want them and petrol is still king for many years to come.
I have a hard time understanding what "increased productivity by 4%" actually means and how this metric is measured. One low-digit does not seem high when put into the context and promises, is it?
They are doing this so they can eventually remove the feature entirely in the future.
Most serious banking apps and financial services are still written in Java, it hasn't displaced much of anything. Big data is a relatively 'new' fad that is already becoming less and less relevant.
>Following this hypothesis, what C did to assembler, what Java did to C, what Javascript/Python/Perl did to Java, now LLM agents are doing to all programming languages.
What did Javascript/Python did to Java? They are not interchangeable nor comparable. I don't think Federico's opinion is worth reading further.
Someone might join if they are desperate enough and hop at the next better opportunity, that's what I would do. Aren't you worried this will be a waste of time?
59K for a senior fullstack engineer in Hamburg is crazy low.
>Probably checking in about 10 to 15k lines of code per month on average.
God, these claims are so ridiculous. You guys are probably doing it wrong if you write that much code.
Or "what will surely not happen in 2026". HN crowd is atrocious at predicting anything.
What I've learned from all these disaster stories: have backups for everythig. I have an iCloud+ subscription but also a OneDrive subscription, photos are sync'ed to both storages. On gmail, I set up fwd for all emails to another email address (non-Google related) just in case. Of course you can't do this for every service but do it for the ones you can.
On a meta note, Fuck Apple, I'm so glad I didn't pursue an iOS developer career 10 years ago.
Why is Anthropic not legally responsible for damages here?
These types of problems you usually bring forth in retros. Have you done that, what was the outcome?
Would you learn how to fly a plane to better understand how a car works? The answer is no of course. You learn more of Python and its internal workings to understand Python better.
This was in the news just 2 days ago: "U.S. reports 1.35 million COVID-19 cases in a day, shattering global record". I'm not sure why you're focusing so much on Europe, it's happening in other places as well.
>I imagine they will be the first to be automated.
This has been said over again since the 90's when all sort of tools and "codeless" solutions started popping up. While you can automate some stuff for smaller shops, I suspect web development will trump all the rest simply because of how ubiquitous it is.
I agree with deep technical knowledge but at the same time make sure you don't dig a hole too deep where jobs are few.
In my opinion the best way to future-proof your skills is to stay up to date with market trends and notice when your tools of choice (even in your current job) start to become obsolete or used less and less. I would also pay close attention to the number of jobs for a particular skill.
Just an FYI, HN is staunchly against mongodb. If this is good or bad you decide but you won't hear good things about mongodb from HN.
Did I say I don't know the space? Does it sound like I never did front-end development? You should learn how to parse properly. I tried it, it sucks, and that's why.