onion2k

15 ms·

301.
by onion2k 5mo ago
You can't have 100% uptime. It's unfeasible, especially for a startup. You should be telling your customers that downtime might happen, sometimes for reasons beyond your control, and that if it does then you'll do your best t
302.
by onion2k 5mo ago
That's one footgun, but then pushing that into production and actually deleting things rather than queuing them to be deleted later after a sanity check until the system is stable, and not informing users that the 1 year policy exist
303.
by onion2k 5mo ago
Some people think it'd be reasonable to have every car reporting it's GPS position and speed to the government to stop people speeding, or to have facial recognition cameras on ever corner to catch criminals, or to make everyone c
304.
by onion2k 5mo ago
Being a polymath was once doable. Any reasonably intelligent child who can write some poetry, understand basic calculus, and has a working knowledge of chemistry would be heralded as a polymath if you dropped them into the Victorian era.
305.
by onion2k 5mo ago
Jensen Huang was saying humanity is going to need 1000x the current energy production in the future. He might not be wrong.
306.
by onion2k 5mo ago
No it isn't. My salary won't increase if the AI budget reduces, or if the AWS bill goes down, or if the board decide not to do a retreat to Davos this year, or anything else. My salary is very strictly limited by the salary budget
307.
by onion2k 5mo ago
It does mean that, but in a situation where people don't know what the productive use is you have no other option. It's like that famous quote about advertising that says "Half my ad spend is wasted, but I don't know whi
308.
by onion2k 5mo ago
I'd bet that the goal is for people to 'game' it though. By pushing people to use AI more they'll try it, experiment with it, 'waste' time on it ... and from that they'll learn about it. That's the en
309.
by onion2k 5mo ago
The problem with this way of thinking is an assumption that problems would be caught "if only the developer thought about it some more." That's just wrong. Being able to foresee problems is great but you'll always have
310.
by onion2k 5mo ago
De-orbitted 1-2 ton steel rods will have about the same effect - ie. you can hit many strategic objects of your attacker. The orbital kinetic strike weapons that have been proposed in the past are usually 2 ton titanium rods that would hi
311.
by onion2k 5mo ago
Something being deorbited will probably break up into relatively harmless pieces that mostly burn up though, and there's no nuclear material involved so even if a massive chunk hits the Earth that's not going to have a huge impact
312.
by onion2k 5mo ago
Anybody else wanting to get there would have to get permission. The USA, China, and Russia have all successfully tested anti-satellite weapons. If anything, any company that operates a constellation of space-based data centres would need
313.
by onion2k 5mo ago
And if it's a website promoting a local tree planting initiative or whatever is it really a good use of budget to get everything hardened so much. Given the fact lots of sites like that have Wordpress 'databases' of form su
314.
by onion2k 5mo ago
A VC knows that out of 20 investments: 10 will go to zero. 5 will return the money. 4 will make decent returns. 1 will be a unicorn. And that is why VC funds never fail! Except ~25% of funds don't return anything to their investors
315.
by onion2k 6mo ago
The prototype is never the whole application. In a world where people (stakeholders, Product, and dev teams alike) want the prototype to be the full set of MVP features, this is not true.
316.
by onion2k 6mo ago
I recall a few times everyone promised, if this gets promoted then we will rewrite it from zero. Never happened. Why would you do that though? If you have a working 'prototype' that's handling the demand, has the required f
317.
by onion2k 6mo ago
You do need to ask whether or not Sonnet or Opus are overkill for a lot of work though. If Gemma4 with some human effort can achieve the same result as Sonnet then it's arguably a lot more cost effective as you're paying for the
318.
by onion2k 6mo ago
If your tool regularly lies, gaslights and produces wrong results, that's a tooling issue. It's a human issue if you don't recognise that the code it's generated is wrong. That will never change no matter how good the
319.
by onion2k 6mo ago
Now of course you could just reprompt your LLM to change the HTML - but when I already have a clear idea of what I want to say in my head, that’s just another roadblock in the way. It's usually faster though, so you get to spend more
320.
by onion2k 6mo ago
Yes, and this is obvious if /users/ exists and returns a 400 if the ID is required. That way you can tell the difference between /users/ being there and expecting and ID, and it not being there.
321.
by onion2k 6mo ago
You can return whatever HTTP response code you want, but if you care about knowing whether your site is working being about to look at the logs and see "That user requested a page that doesn't exist" being different to "
322.
by onion2k 6mo ago
You could get the agents to output something structured and then use a deterministic test if you're worried about that.
323.
by onion2k 6mo ago
Agents are probabilistic systems. A common mechanism to get a reliable answer from systems that can have variable output is to run them several times (ideally in separate, isolated instances) and then have something vote on the best result
324.
by onion2k 6mo ago
I am. It's not very nice. Arguably nicer than suffocating though.
325.
by onion2k 6mo ago
We suffocate them to kill them when we pull them from the sea. That's quite mean. Few people would advocate the humanity of killing a cow in the same way.
326.
by onion2k 6mo ago
Software is going to pile up because developing it is now cheap. It won't, because right now we're busy exhausting the vein of good-ideas-we-wanted-to-build, and that's the source of all the good stuff you listed. When that
327.
by onion2k 6mo ago
I think it's more subtle than that. There are a lot of measures of what a 'good developer' is, and one of them is 'shipping things'. AI is specifically accelerating that part of the industry - it's much easier
328.
by onion2k 6mo ago
Lots of very successful software was built using a waterfall approach. It's a methodology that works well if you know precisely what the end result needs to be. That doesn't make it appropriate for everything - if you don't k
329.
by onion2k 6mo ago
The scientists are being killed by space fireballs!? This is conspiracy bigger than I thought!
330.
by onion2k 6mo ago
given that you can read the code, stars seem to be a completely pointless proxy Imagine you're choosing between 3 different alternatives, and each is 100,000 LOC. Is 'reading the code' really an option? You need a proxy.

More ›