alkonaut

7 ms·

61.
by alkonaut 4mo ago
An aerial bomb is equally indiscriminate. You make a judgement about the target value and the estimate collateral damage/civilian casualties and you fire if the equation is reasonable. Same here. If it's a muddy square kilometer
62.
by alkonaut 4mo ago
I find deepseek-v4-pro to be every bit as good as sonnet tbh
63.
by alkonaut 4mo ago
Yeah that's why it probably needs to be more than 1 charge in the battery. Unless you do N back-to-back charges during peak time, the charger isn't utilized enough. And to do N back to back charges you need about N car batteries a
64.
by alkonaut 4mo ago
> if in average you have 2 cars charging for 5 minutes every hour, you can draw 166kW continuously instead of having bursts of 1MW consumption) You definitely need to have that to not load the grid with 1MW, but the question still remai
65.
by alkonaut 4mo ago
Validation to avoid mistakes is, as they point out, good. I'd even go so far as to extend it so that I reject those without any tld (without any dots) just because it's 99.999% a mistake and I don't care about the person who
66.
by alkonaut 4mo ago
Does the location help though, if the company isn't trusted? I can't even visit the webpages of these companies from my enterprise network
67.
by alkonaut 4mo ago
Copilot CLI with DeepSeek, writing code in mostly C#.
68.
by alkonaut 4mo ago
Copilot CLI with DeepSeek, writing code in mostly C#.
69.
by alkonaut 4mo ago
I don't think painting and photography (Which was when painters perhaps feared for their jobs at first) are equivalent. If I paint, it's not because I want an exact "photo" of a person or scene. The fact that its painted
70.
by alkonaut 4mo ago
Some, but actually surprisingly little. The thing is that my hobby projects are usually extremely "prior art". It's like "make a renderer (just like one that 1000 people have made before)". It's not novel. I ju
71.
by alkonaut 4mo ago
Yes, and that's also why I'm really effective at using AI when the context is one I'm comfortable with. Still - it doesn't feel like I'm actually doing anything. The product was never the goal for me. The craft was.
72.
by alkonaut 4mo ago
But "Is the user a kid" is already a switch that I (a parent) switch on in the device and that the kid in question can't switch off. That bit seems like a solved problem? Why would anything else even be needed in that space?
73.
by alkonaut 4mo ago
Does the law really require third party? Because having on-device functions configurable by parents doesn’t seem terrible at all.
74.
by alkonaut 4mo ago
I spent most of the last two years making a rather large hobby project. A 3D renderer for cad/visualization. It's pretty hard and slow work because getting anything wrong is usually hard to debug. Get a sign wrong and you have a b
75.
by alkonaut 4mo ago
Absolutely would. But if we can categorize 1/3 of defects as ”bugs” 1/3 as ”missing or incomplete functionality” then there is 1/3 in the middle which we can’t decide. So the dichotomy is kind of weird, and making ”rules” and
76.
by alkonaut 4mo ago
Not even a development team can (or should!) be able to agree on whether something is a bug. Just because there is never a complete spec. It’s a mental state of the team. Or expectations among stakeholders.
77.
by alkonaut 4mo ago
There are things that are clearly bugs and clearly features. And a lot of things that are somewhere in between. It’s not black or white and as such perhaps not worth categorizing? Especially with dramatic processes like ”always fix all bugs
78.
by alkonaut 4mo ago
> do you mean in the short or long term? The answer to that is sadly "yes". > prioritizing so-called "quick wins" only quickly wins the codebase more tech debt, that puts the project on a sure path to development
79.
by alkonaut 4mo ago
That's the simple version of it yes. I outlined a more complex version of it in a parallel answer. In short: lacking a complete specification of what to do, it's often impossible even within software teams to tell whether somethin
80.
by alkonaut 4mo ago
Some times there is, and in many cases there is not. Distinguishing a feature from a bug requires having some kind of spec. In some cases you can have obvious sign the behaviour is unintended/not desired, but in other cases it's n
81.
by alkonaut 4mo ago
> It doesn’t matter that the language you use is memory-safe, if you didn’t design for correctness or have no process that will eventually lead you to fixing all bugs. After many years in the business I have come to a more pragmatic view
82.
by alkonaut 4mo ago
And this isn't a "Windows PC" in the traditional sense. The reason people run Windows in the enterprise (and for some desktop home uses like games) is still hardware and software compat. I run it for work because we make wind
83.
by alkonaut 4mo ago
The trouble this time is if this goes bust I see a huge crash. And if it's wildly successful it seems worse.
84.
by alkonaut 4mo ago
Won’t (presumably) all the market actors converge on similar pricing? If OpenAI stopped operating on subsidies and charge the true costs and their most token hungry customers are the ones that switch to Anthropic and others, then their pric
85.
by alkonaut 4mo ago
I mean all else being equal, I'd just take the ergonomics (dependency management, build-system/multi-targeting, modern language features, no h-files, ...) and run. I can always just write at the level I want (safe rust, unsafe rus
86.
by alkonaut 4mo ago
Either your model needs 20+GB or it doesn't. This whole hardware race is about local AI, little else.
87.
by alkonaut 4mo ago
People buy Windows Laptops because they need to run windows. But everyone saw the M1 and realized that local AI basically made integrated memory the future. People can't choose between "A computer that is good at local AI" an
88.
by alkonaut 4mo ago
I do windows (Desktop) dev for work, and am interested in running local AI. I'd be really happy to have a machine with 64GB shared memory and a decent x64 emulation.
89.
by alkonaut 4mo ago
It would be a steal at $5k tbh, with todays RAM prices.
90.
by alkonaut 4mo ago
That works both ways: The macs run MacOS which for my work at least is a non-starter. I write win-only apps and sell to win-only customers who run only windows-machines. I do want to run local AI models on my machine though so I really do s

More ›