HN user

andrecarini

772 karma

email: andre.carini at acx.net.br

Posts16
Comments137
View on HN
ec.europa.eu 1mo ago

Safety Gate: the EU rapid alert system for dangerous non-food products

andrecarini
2pts0
news.ycombinator.com 7mo ago

Ask HN: Are AI coding agents becoming more stupid?

andrecarini
1pts2
www.youtube.com 7mo ago

IiSU – Front end for game emulation [video]

andrecarini
3pts0
github.com 8mo ago

Show HN: Hacker News Dark Mode

andrecarini
2pts2
www.tomshardware.com 1y ago

Activision takes Call of Duty: WWII offline after reports of RCE exploits

andrecarini
5pts0
news.ycombinator.com 1y ago

Ask HN: Niche technical knowledge not found on the internet?

andrecarini
10pts3
github.com 1y ago

Tab Session Manager: browser extension for persisting tab state

andrecarini
2pts0
email.faircode.eu 2y ago

FairEmail: Open-source, privacy friendly email app for Android

andrecarini
3pts1
waiting-for-blue-robot.gitlab.io 2y ago

Blue Robot Cult: "Godot is a toxic cult"

andrecarini
5pts3
www.nsf.gov 3y ago

Report on U.S. Antarctic Program’s Sexual Assault and Harassment (2022) [pdf]

andrecarini
2pts0
news.ycombinator.com 3y ago

Ask HN: What's on your private torrent tracker that can't be found elsewhere?

andrecarini
3pts4
news.ycombinator.com 3y ago

Ask HN: Why is Meta's Business Suite so broken?

andrecarini
6pts9
news.ycombinator.com 3y ago

Ask HN: State of malware in 2022? Share your stories

andrecarini
1pts2
news.ycombinator.com 4y ago

Tell HN: I used the same computer since 2007 (with minor upgrades)

andrecarini
407pts291
news.ycombinator.com 4y ago

Ask HN: Software you hate but can't replace?

andrecarini
92pts324
datatracker.ietf.org 5y ago

RFC8995: Bootstrapping Remote Secure Key Infrastructure

andrecarini
2pts0

Failing 1/3 of a class if that cohort is genuinely deemed not qualified enough to pass shouldn't be a problem by itself.

But then it raises questions like "are they really unqualified or is the testing methodology inadequate?" and "why was the system unable to provide the necessary growth to such a high slice of the class?". And then the easy way out is to just cherry-pick which students enter the system at all.

My understanding is you're equating `failing a test` to `lacking the relevant skills and knowledge to do a certain task competently`.

The reality is sometimes tests in academia are just not very well made and don't really test what they are supposed to be testing, and that's usually due to multiple reasons like misaligned incentives, staffing shortages and maybe lack of resources / funding.

I don't think the comparison to flight school is relevant enough in this context because it's a too different of a world to traditional academia.

My impression as a driver-in-training is that people are too complacent and forget they are handling a machine that could at any moment kill and maim you, your loved ones and random innocent bystanders. I wish we all were more responsible about it, and I hate the Tesla philosophy of going the opposite way (the touchscreen, the sorta there but not really autopilot, etc)

You have a Minecraft server. You generate money from it (selling VIP packages, et cetera). You could generate more money if you had more players. You can have more players if you consistently DDoS other more popular servers; the experience for these players will be horrible and they might give your server a chance.

Bézier curves are great, but there are certain things they just can’t do. For example:

It would be nice if we had a better explanation of what is wrong with the Bézier curves in the example. I've put the spring Bézier example side by side with the javascript simulated one, had them both trigger at the same time on a keyboard press and I can barely notice a difference; one doesn't look better than the other to my eyes.

As an unpaid volunteer to a multimillion dollar corporation that has just erased a huge collective volunteer effort, listing in writing the reasons I'm unhappy is already way too much effort.

Asking that same volunteer to hop on a video call is just insensitive. They're the one providing free work; if you care about solving the problem and not losing the volunteer force, you should go where they are (the forums) instead of asking them to come to you (video call). They probably don't want to take time out of their schedule to waste their time talking with a community rep. And they probably don't even want to do a voice/video call.

I've been playing with Suno lately (I'm an amateur/hobby musician, for context) and I've been making some tracks for my own enjoyment. I've shown it to friends and got mixed reactions.

Personally, I think it's still uncanny similar to these AI image gens with multiple hands and nonsense details in the background. You can one-shot a lot of passable stuff but the moment you want to put more effort into it (e.g. correct a word or two, slightly change the style of a section) the track gets really messy very fast.

Even if you could generate real-time 4K 120hz gameplay that reacts to a player's input and the hardware doesn't cost a fortune, you would still need to deal with all the shortcomings of LLMs: hallucinations, limited context/history, prompt injection, no real grasp of logic / space / whatever the game is about.

Maybe if there's a fundamental leap in AI. It's still undecided if larger datasets and larger models will make these problems go away.

As another reader who has no idea what any of this is about, I've coerced my favorite LLM to digest it into ooga-booga format in the style of this essay[1]:

# grug see big sky boom

- sky make ooga FLASH but not light, just invisible whoosh (radio).

- whoosh so strong, like sun work many day, but all squish into blink of eye.

- smart sky-people have big ear rock (CHIME). ear rock say: "boom come from there, galaxy far, but not too far (only 130 million fire-circles (light years) away)."

- ear rock also have many little ear-brother rock across land, help point finger very good.

- finger point so good, sky-people know spot of boom smaller than tree forest (13 parsec).

- then, magic glass eye (James Webb) look at spot. see old fat star (red giant) glowing soft.

- but fat star not make ooga boom. hmm. maybe fat star have sneaky tiny angry friend (neutron star).

- tiny angry friend go "KRAK!" → make fast radio boom.

# lesson for tree-brain

- boom in sky still big mystery.

- now smart sky-people can say where boom come from.

- if know where, can watch with other eyes, maybe find secret of why.

- grug think: many sky boom = maybe angry tiny stars yelling far away.

# Ooga booga translation:

"Tree no know why sky yell. But now tree know where sky yell. Soon, tree maybe know why sky yell."

[1] https://grugbrain.dev/

There's also lots more if you're coming from a print design world

Ah, there's another one you just reminded me:

- Poor paragraph justification algorithm; there's no hyphenation system.

This one is table stakes for professional print design.

Well, this is about software you bought in advance. The Facebook comparison doesn't really work because you're not paying in advance to use it.

This software is not guaranteed beyond 0 Unix time.

Grandma goes to the store to buy a game for their grandkid. She sees two game boxes. One says in the front "This game will stop working tomorrow and you will not get a refund" and the other says "This game will stop working in 2030". Which one does she pick?

Lots of bad takes in this thread. The whole idea behind this is just to stop defrauding customers that buy your software and then are left holding the bag. Nobody is asking for developers to keep running server infra for eternity.

Any of the following options are enough to satisfy this proposal:

- Put an expiration date on the storefront and make it clear that your software is not guaranteed to continue working after date X.

- Have your server source code (stripped down of proprietary stuff) ready for public release at EoL.

- Allow customers to reverse engineer the binaries and communication protocol after EoL.

- Package dedicated server binaries with the game and allow customers to connect to it via a LAN or direct IP option.

- Clunky component and variable system; inadequate for more complex stuff with lots of parameters.

- Can't set connectors on Design files (used for documenting the navigation flow between different pages of an app).

- You can set connectors on FigJam files, but if you want to bring your components from Design files then you can't keep the instances synchronised to the component definition. And you can't attach the connector endpoint to some element inside the Design component. It's essentially just an image export of the Design component.

- Prototyping is very clunky and trying to build a flow that has elements reacting to interactions on other distinct elements is either variable hell or downright impossible.

Those are just off the top of my mind. I'm always finding threads from 5 years ago on their community forums with loads of people on the same boat and no activity from Figma side.

I'm really curious about this but searching for "Google Omaha" doesn't return anything related and the linked FOIA request is... just a request, which I assume anyone can write whatever they want in the request.

Notion has been great for me since I started using it back in 2021. Hope they get all the success they deserve.

That being said, I'm honestly disappointed with their overall direction. The main product has not improved much since 2021, except for features I assume are catering to their big enterprise customers.

It's slow, the mobile app has bad UX, the Windows app is a memory hog and I run into issues whenever I build something more complex with it. It's still a great product, but as time passes I look more into alternatives...

Watt The Fox? 1 year ago

I could even hear an actual radio broadcast coming off my speakers whenever they were powered. Maddening!