I'm no longer considering myself a good developer either, the agent is way better at details. Not so much at following a general pattern that makes sense.
Great advice.
If at all possible, I'd say favor part-time, so it gets you up in the morning, but doesn't completely fry you out by the time you come back home.
Hi,As I am coding day in day out using Opus, I was wondering if people here had a view on the effect of input style on the quality of the agent's production?Like if I treat the agent like shit, with a great deal of pressure and shaming, does it perform differently than if I treat him with the utmost deference and calm?And if so, which should I be using?I am usually talking nice to my agent,...
That's the bottom of the problem. You can think all you want in your spec doc, you won't fully solve all the problems that you will face during implementation.
That's why ports are easy with LLMs, cause the code implicitly carries all the spec.
Sure, there may be a gap between the code and the idea behind the spec, but it can be argued that the gap between the spec as a document and the idea behind the spec is just as wide, and in fact doesn't take into account the constraint of reality.
LLMs generally suck at taking the initiative. Tell it to build a report, it's going to do it, but then you'll have to iterate a lot to make sure it's producing it the exact way you want it.
Like is the motivation to please Trump and ask for stuff/be left alone? Or is Zuckerberg really gone full right wing globally?
This is not a data center, this is a data town.
I’ve been a professional developper for nearly 20 years now. Started on my Amiga in 1992, went on to get a CS degree in 2006. I’ve worked on video codecs, financial systems, math-heavy workloads, frontend, backend. Done the C#, the Python, the React. I made games for my wife’s bday in pico 8. I made some cad stuff for my 3D printer.
Programming has always been a great source of joy.
And yet I have zero regrets using AI to write my code. None.
That thing is so fast. I remember an exceptionnally good programmer who used to spent time honing his typing skills so that he can program faster. he was so good his bottleneck was how fast he could type. AI is like this to me. It’s just helping me write faster.
There’s a reason we call them assistants. They are. Like actual assistants. They just assist me with doing the stuff that is tedious to me.
I am not arguing “love the product, not the tech”. It’s still the tech. I still look at the code that’s produced. And I still want it to look nice. “Rename this, rename that, abstract this away, you should do that loop first, oh no you gotta start over from scratch”. I tell the agents.
I don’t feel like I am not programming. It’s still the same job, just a lot faster and without the sore wrists.
We’re technologist, we write stuff to make people’s life simpler. And this time, the progress is make our lives simpler.
EDIT: Yeah, obviously there are dozens at 3 bucks on temu.
But god, it's so true. According to the latest dentist you'll see, everyone who's seen you before has done a terrible job. They're worse than plumbers...
The worst that could happen is you get fired of course, did you expect anything more serious could happen?
taking the geometric aspect first and defining operations on points in the plane, showing that it works to match multiplication and rotation.
https://www.wsj.com/finance/regulation/flurry-of-suspicious-...
Deciding square root of -1 is fine and use it in calculations as if it existed is confusing. Why does it work with sqrt(-1), and not with dividing by 0 ?
introducing the imaginary plane as: "hey someone found you could just draw them, and it works even better!" is even worse.
To teach complex numbers, you should start by defining them:
1 is here, i is here, multiplication for 1 is like in R, but multiplicating i by itself goes to -1, and then derive the calculation rules then.
And only when you went all the way to exp(i*theta) = cos(theta) + i sin(theta)
Only then you should explain what they are for. That they are a way to treat real numbers as a projection of these things that behave nicer in calculations. Deriving trig functions from that simple formula is trivial.
It's obvious that open weight models can do that. The only steps needed are hacking a server, downloading themselves, run themselves. The only difficult part is finding a server to hack that also has enough computing resource and disk space. With the cost of running LLMS going down and number of powerful connected devices going up. It's inevitable that this is going to happen...
I don't think regulation is the right way to fix this, but I think it's a problem we shouldn't ignore.
I am a parent with two young kids. My older one is difficult. He has had behavioral problems since very early on. Looking at the current set of "disorders", he checks all the boxes for OCD. I am putting "disorder" in quotes, because I don't consider it an illness to be defiant. It's annoying AF, but in the wild he'd be just another ape. Btw, there is no "cure" either, the "treatment" is to help parents dealing with it.
So my kid does talk back to teachers. He does hit his friends. He probably steals. He probably lies. I dont know the full extent.
Day in, day out we hear:
- kids these days are treated like kings
- their parents let them do whatever they want
- kids hit their teachers and nothing happens
- schools level plummet cause kids are making a mess
- it wasn't like this in my days
and so on and so forth
As the parent of my kids, I can assure you that we've never been permissive or lax with them, and neither have been their grand parents, or their teachers. The older one getting under our skins easily, we did use quite a lot of "force" in trying to stop him as the article puts it.
I think there might have been a time when those children were meant to be "broken", as in beaten so much so you suck the life out of them. My father got whipped, not sure he came out much wiser for it. So yeah, maybe there was a time when the "bad seeds were rooted out" and life was better for everyone else.
Luckily, many professionals are aware of this, his teachers understand how it works. The people we've seen were very helpful.
My personal experience is that: it's not permissivity, it's physically wired in the brain. If kids overall have more problems than they used to, look not for parenting problems, look for environmental problems.
Maybe it's the ecology, maybe it's increased overall stress. I don't know. Just don't blame it on parents.
Most of us try our best.
https://news.ycombinator.com/item?id=49698936
I haven't digged myself but many people have argued that Anthropic used accounting tricks to appear profitable to investors, but they removed some costs from their calculation.
I think the message to investors is that operating the model is profitable. Like the income they get from paying customers pays for the infrastructure of running the models.
- it was uncomfortable af
- it was unsafe af
BUT
- it was cheap
- it was robust (so many are still running here, it's insane)
However, the car looks really decent, and they're trying to make it cheap, and somewhat barebones. Tbf, the new Renault 5 didn't come out too bad either.
After the disaster of the e-mehari, this is a good surprise.
I guess what disturbs me is that I’d rather have one screen in my pocket than tens of screens in my house.
But more importantly I ultimately think that one of the unstated goals is to reduce screen time. And setups like these are only one voice command away from beaming youtube videos all around your house.
Sure, it's a little more hands free, and projecting around definitly is nicer than screens. But projecting has a lot of problems too. Shadows, projection surface, ambiant light, heat are difficult to handle, but to me the worst is that it's not persistant. You can't just project 24/7 your family planner on the wall.
I'd really rather have something that prints, or better draws.
You want a recipe => print it, don't beam it.
You want a map of kyoto => tiny robot walks around the page and draws it for you. And you can draw over it if you want.
And it doesn't have to be exclusive, you can do both beam and print. But I think protable printing / drawing has a lot of unexplored potential.