wg0

10 ms·

211.
by wg0 6mo ago
It's an illusion of intelligence. Just like when a non technical person saw the TV for the first time, he thought these people must be living inside that box. He didn't know the 40,000 volt electron gun being bombarded on phosphor
212.
by wg0 6mo ago
Looks like you either have not worked with any human or with an LLM otherwise arriving at such a conclusion is damn impossible. The humans I did work with were very very bright. No software developer in my career ever needed more than a par
213.
by wg0 6mo ago
A wise man from Google said in an internal memo to the tune of: "We do not have any moat neither does anyone else." Deepseek v4 is good enough, really really good given the price it is offered at. PS: Just to be clear - even the m
214.
by wg0 6mo ago
If the workhorse underneath is a Raspberry Pi with ghosty? Or must be dedicated specialized for purpose hardware?
215.
by wg0 6mo ago
Sounds like a security risk to me if the link gets misplaced.
216.
by wg0 6mo ago
Very simple. Name it to pgbackrest-AI and add the line: "AI driven backups with smartest world class models optimizing every byte stored via deep AI analysis." With that added, a million dollars is just chimp change. YC alone woul
217.
by wg0 6mo ago
I have watched it several times. Every time it hits different. It surely is a masterpiece.
218.
by wg0 6mo ago
Anyone remembers the GitOps thingy called flux? Weave was the company name. Git and Kubetnetes configuration cannot go hand in hand. You cannot go back in past indefinitely because cluster state might not be that reversible. If so, git is u
219.
by wg0 6mo ago
Docker swarm is good enough for most cases if the black plague called Microservices isn't in your stack. Basecamp is running successfully with Kamal and surely making more money than all the Kubernetes experts combined on this platform
220.
by wg0 6mo ago
Software is code and code is documentation. Do you know MAME that is documenting the arcade consoles as code? And that they work is just a side effect and is not an intended goal? That doesn't mean that MAME is a trivial project. Far f
221.
by wg0 6mo ago
Interesting take. We are not going to talk about Office, Windows, Adobe or Autodesk products here. Neither Linux kernel. Just classified ads or e-commerce platforms such as gumroad and shopify are complex enough that a single person cannot
222.
by wg0 6mo ago
>The combination of technical skill and the judgment to know when the AI is wrong barely exists in the market anymore. I see a talent pipeline collapse in next 5 years. "Software engineering is over coding is a solved problem"
223.
by wg0 6mo ago
I use claud CLI or OpenCode. The "Regenerate" example is just to illustrate that same prompt would produce different output each time. You're rolling a dice.
224.
by wg0 6mo ago
> Writing detailed specs and then giving them to an AI is not the optimal way to work with AI. It is NOT the way to work with humans basically because most software engineers I worked with in my career were incredibly smart and were damn
225.
by wg0 6mo ago
>Then implementation which I still use Opus for because I get a warning that my current model holds the cache so Sonnet shouldn’t implement. This is based on the premise that given detailed plan, the model will exactly produce the same t
226.
by wg0 6mo ago
The author is not a native English speaker it seems. They might mean "few weeks ago" and the phrase "couple of weeks ago" might not be exactly as "Vor ein paar Wochen" in their mind rather could be as "few
227.
by wg0 6mo ago
I write detailed specs. Multifile with example code. In markdown. Then hand over to Claude Sonnet. With hard requirements listed, I found out that the generated code missed requirements, had duplicate code or even unnecessary code wrangling
228.
by wg0 6mo ago
A heavily vibe coded CLI would have tons of issues, regularly. LLMs over edit and it's a known problem.
229.
by wg0 6mo ago
Impatiently waiting for the AI bubble to pop so that both environment and a common consumer and buyer of computer parts can have some relief.
230.
by wg0 6mo ago
It is happening in EU.
231.
by wg0 6mo ago
Few years ago, recruiter at a very well known retailer was on phone with me for a Python backend role. I asked him how many people were on my team that I'm going to join. He told me some 80 developers. Bit surprised and thinking probab
232.
by wg0 6mo ago
Capitalist tech bros have lost their marbles at this point.
233.
by wg0 6mo ago
Practically speaking, the $20 plan is useless anyway given the limits and the way Claude Code consumes tokens with very little caching.
234.
by wg0 6mo ago
The best thing for humanity, economy, technology, society, progress and environment is that this scam should come down ASAP.
235.
by wg0 6mo ago
How are the usage limits compared to Anthropic?
236.
by wg0 6mo ago
Notice the pattern that Chinese providers are now: 1. Keeping models closed source. 2. Jacking up pricing. A lot. Sometimes up to 100% increase.
237.
by wg0 6mo ago
Someone exactly said it better here[0] already. [0]. https://news.ycombinator.com/item?id=47817982
238.
by wg0 6mo ago
I have said similar things about someone experiencing similar things while writing some OpenGL code (some raytracing etc) that these models have very little understanding and aren't good at anything beyond basic CRUD web apps. In my ow
239.
by wg0 6mo ago
I wouldn't know for sure and I'm clearly guessing. But if you look at the units who's executives are leaving (Sora and such) and then put two and two together - a different plausible picture emerges that they might have been
240.
by wg0 6mo ago
If you have the setup within server fully scripted and automated (bash, pyinfra or ansible etc) and backups are in place then recovery isn't that hard. Downtime for sure maybe couple of hours for which you can point your DNS entries to

More ›