timenotwasted

It's a fun way to push for a lighter web but without a way to distinguish the complexity of the sites on the list it's really not all that useful. It's kind of addressed in the FAQ "The whole point of the 512KB Club is to showcase what can be done with 512KB of space. Anyone can come along and make a <10KB site containing 3 lines of CSS and a handful of links to other pages" but without a way for the user to distinguish the site complexity at a glance I'm not sure I understand the point. Regardless of the FAQ the first few sites I clicked on while yes were quite light in size but also had nothing more than some text and background colors on their sites. Also any search site is going to be at the near top of the list e.g. https://steamosaic.com/

Complexity would be a subjective metric but without it I'm not sure what you take from this other than a fun little experiment, which is maybe all it's meant to be.

There is not enough nuance here to really understand the friction that this would have created but I've found the more conducive environments that engage and value healthy debate are made up of individuals that have the "strong opinions loosely held" mindset. This, on the other hand, sounds like stubbornness and wearing everyone down but again a lot gets lost in the details so I'll opt to not jump to conclusions.

I love this kind of stuff, makes me nostalgic for the early days of the internet. There are still so many fun, interesting things to find out there. However, in a twist of irony they seem so much more difficult to discover now.

I had an opportunity to work with OSM data (nowhere close to something like FlightAware would need) at the beginning of the year and this comment put into words something I felt deeply at that time — "OSM is amazing, but as a global project with thousands of contributors, its data can be... quirky.". I have tons respect for this kind of stuff after working on that project and seeing the massive amount of time that gets invested to make things look correct for the use. It's the quintessential if nobody notices then you've done your job extremely well.

It's really great to see the shift that has been taking place away from unicorns, growth for the sake of growth, and all the chaos that drove throughout. Maybe it's my own personal bias but I feel that these stories of low, slow growth; small teams, small wins but consistency are becoming more norm. While I realize there is still plenty of froth, it's inspiring and makes me hopeful for an industry shift in that direction.

"And when we launched after a year in private beta, almost all of our 100 beta users converted to paid customers." — That's a neat stat and one I'd be extremely proud of.

Aside from the value debate it's pretty crazy how far and how fast things have moved since GPT3.5 circa 2022 which is the early onset of OpenAI becoming more formally known mainstream.

I have an old 2080TI that I use to run Ollama and Qdrant. It has been ok, I haven't found it so good that it has replaced using Claude or Codex but there are times where having RAG available locally is a nice setup for more specific queries. I also just enjoy tinkering with random models which this makes super easy.

My daily drivers though are still either Codex or GPT5, Claude Code used to be but it just doesn't deliver the same results as it has previously.

This is actually a really interesting offering to have available as someone who needs DEV tier PG databases for a better testing pipeline on a shoestring budget.

I was referring to something more akin to a legend like you have in the examples "(examples: hard-coded secret, weird crypto mode, gnarly logic)." where I could click "hard-coded secret" (not the best label but you get the idea) and it would filter on those instead of the slider.

This is very cool and I could see it being really useful especially for those giant PRs. I'd prefer it if instead of the slider I could just click the different heatmap colors and if they indicated what exactly they were for (label not threshold). I get the underlying premise but at a glance it's more to process unless I was to end up using this constantly.

It does feel like AI has really started to level the playing field for some of these industries that are black boxes. Close family members have fed medical data to Claude and ChatGPT and had much more useful interactions with care providers than previously possible. Was it possible to sort this out before? Sure, but not without a lot of research, now it is become much more accessible and that is a great thing.

The absolutist type comments are such a wild take given how often they are so wrong.

And now academia needs to adapt since they can't just assign out all the learning to the students. All this is doing is exposing how broken the entire concept of higher-ED learning as been for quite some time.

The 4x price jump over the old Claude 3 Haiku was surprising, especially without image input support.

It’s cheaper than Opus but still a tough sell if you’re looking at CpT which for me gpt-4o-mini still can't be beat.

As a parent with a couple of kids about to be the age of wanting to watch YT I'm very interested in this!

Playing around with LLMs to generate fictional stories and I'm working on a journal of sorts based on a traveler that goes to various parallel worlds not unlike our own except for one or two minor but key differences. He reports on his findings there and how it affects social, political, and economics for that timeline. A new timeline drops every day and I have a lot more planned to take it beyond the simple blog format it's in now but it has been a fun challenge to learn how to do some deep prompting with AI to generate random entries without veering off into crazyiness or getting locked onto the same thing over and over again.

Here is the site: https://twistedtimeline.com/ - Feedback is always welcome!

"It’s important for my software to get names right because it sends out emails and messages on the users behalf."

This is the critical part your requirement, since it's important to get it right for emails then Full Name would be the correct option. If you needed the name for business or legal reasons then you'd want to go the other way.

As someone who has tried and failed at many points throughout the years at building a game I couldn't agree more with "Build a game, not a game engine" it's hard not to get sucked into the game engine black hole though.