futhey

Code quality is just going to get worse. I think that ship has sailed. Its no longer economical for humans to write good code. We're all going to be building better scanners, automated review mechanisms, quality checkers, feedback routers, etc. for the next 5 years.

I could be wrong but my read from following this on Twitter was that it started on Deepseek 4 (flash). It probably started out focused solely on that model with no guarantee the techniques would transfer to anything else.

Because every bit of this webpage was written by Claude.

When I was younger I realized I was falling into specific intellectual traps, and I've noticed LLMs adopt every one I attempted to thoughtfully eradicate in my own reasoning. You can often sound smart by adopting a contrarian position without actually putting a lot of thinking effort into what you're considering. It's a quick escape hatch to sound intelligent; quickly identify a contradiction or a counterargument and state it confidently. Adopt a contrarian viewpoint. You don't have to be right but often you will sound intelligent.

I assume this is just the natural result of asking LLMs to produce text and paying someone three cents to evaluate if it's a good response or not.

Not sure how this is even possible but the trending box for Taiwan just shows the Thailand trending searches in a different order... And worldwide trending searches are predominantly... in Greek?

So the buyout price is never going to match depreciation, and in almost all cases keeping the device doesn't make sense.

The flip side of this, Apple gets more control of the resale market, pushing more buyers into the new device market.

In theory it lets them solve any market problems that emerge where independent resale markets let high quality devices cannibalize sales of new devices.

This is the AI we were promised

Same! I distinctly remember seeing portions of this wallpaper that led me to believe it's procedurally generated and not simply a collection of video files, and had to rewatch it to confirm. There are several points where the date and time match the system date and time down to the second.

Linked spec sheet is labelled as 2.5 years old, chip is 6 years old

I'm being pedantic but I liked your comment. Most TVs today are giant ARM computers, ~95% of TVs ship with ARM Cortex but only about 35% have some variant of Android.

Most LED backlights are wired in such a way that when one LED fails it bricks a significant portion of the panel backlight. You'll knock out entire rows or huge portions of neighbor backlight LEDs when one fails. Basically it's a cheap way to ensure a whole row of LEDs are the same brightness but the tradeoff is one LED fails and it looks like 5% of your screen went dark.

It seems like a good beginner-intermediate thing that'd be approachable to learn with a basic multimeter and beginner level soldering skills.

Small UX thing: I would have loved to not had to inspect element to grab the placeholder text to test it out :)

I bought the color package, enjoying it. It's a personal preference. Hopefully the founders get a variety of feedback and can make a judgement based on multiple datapoints.

That worked. Really well!

But, white on black is really ugly. Even black on white or a simple inversion would be an improvement.

I think it could benefit from the ability to pause and see the transcript, and make edits before the video is generated.

X is so large that it can simultaneously be a giant dumpster fire, the most toxic social network ever to exist, and still have room for honest discussion and good communities.

Gives me some hope for Bluesky, etc. I don't think you need to be Twitter scale or have global network effects to work. Your community just has to choose a particular platform and show a preference for it. You get miniature network effects once your community adopts it.

So if your favorite community doesn't like a particular platform, I don't think they're stuck there, just because it's the one with global scale. They just have to organize an exit.

I thought it was an April fools joke, glad I'm not the only one.

There's a lot of correct feedback about why this is not an amazing start for something starting as a marketplace for talent.

This could be a beautiful start for a Linkedin alternative focused on freelancers, entrepreneurs, etc. Maybe even people in specific industries whose Linkedin currently does more than get them their next job (sales comes to mind, but also folks who work at design agencies and need to get clients, and have a profile they attach somewhere in the about us section).

BunnyCDN has a great product offering, particularly if you've used Backblaze B2 as "ultra-cheap" object storage, the BunnyCDN product is very competitive pricing-wise, and the CDN configures seamlessly with it. And you can set up a cheap image transform proxy on any of your CDNs.

R2 is cheaper though if you storage cost is less than your bandwidth cost, and B2 has a feature to automatically expire items which depending on your design might make it more efficient.

This isn't new or unique. The model seems to be too focused on the task at hand, "cleaning up" unrelated code.

I basically have to branch, commit every time it makes any progress at all, and squash later. There are built-in checkpoints that basically do this.

I actually run this side-by-side with my preferred IDE, and GitHub Desktop (to visualize the diffs). So, prompt -> Claude makes a change -> I view the diff -> I make some edits -> Commit -> back to Cursor.

They're ad-free for Premium subscribers, but some small amount of the premium subscription fee is supposed to be divvied up between everything you watch, so it is technically being monetized.

Well where exactly are my tax dollars going then? /s

I saw someone kill a chicken and get an egg. I thought I'd do the same but almost died.

FYI You can set your default Chrome search engine to be a site search, including one that searches https://www.google.com/search?udm=14&q=%s

Every single other major jet manufacturer is already making air travel cheaper and more efficient every year. It's all airlines who buy from Boeing or Airbus care about. The underserved niche is premium and small, not scale. Commercial aircraft will get 1% more efficient every year with or without new companies (https://en.wikipedia.org/wiki/Fuel_economy_in_aircraft#/medi...).

If we start in supersonic today, and innovate at the same pace for 20 years, you might match the efficiency boom from the "jet age" and make supersonic twice as efficient as today.

Paying someone to break the law makes you culpable. Paying someone to violate an NDA or other agreement with a private company you have no direct association with doesn't ordinarily make you culpable.

The cool thing is that there are solutions to all of these problems, if the more basic problems can be solved more reliably to prove the underlying technology works.

When people realized anyone with a sophisticated police scanner could listen in on cordless (and then early cellular) phone calls, it forced manufacturers to actually implement a bare-minimum level of security on those devices.

I hope this pushes more manufacturers to switch to rolling-code algorithms (like the key fob your car uses), in place of simpler, less secure codes that can be captured and replayed.

He profited from the panic, not the eventual failure of Evergrande. So the court ruled in accordance with the intent of the law.

If they were legally insolvent back then and his information forced them to immediately declare bankruptcy, this short would have been above board.

In Taiwan it's one of the controversial things that comes up every year when there are electricity shortages or water shortages; whether TSMC should get special treatment, or not.

ASML has 5,000 direct and indirect suppliers, some of these have no alternatives or competitors, and ASML is just one of TSMC's core suppliers. Advancing the cutting edge in processor design is a byproduct of global peace and stability at this point.

I've done a few of these. Basically, you need to quote them significantly more than you're currently comfortable asking, because it's basically going to suck up months of your time and you're going to have to re-tool your business to meet their requirements. Let's say $10k-$20k, assuming from your numbers they're 100x the size of a typical customer.

Quote them such a large amount that you expect them to say no, and then make your peace with them saying no. Don't accept a significant discount on what you quote them. Instead, let them go, and wait for another enterprise customer to come along.

Do a call to understand why they want to buy your product ("I wanted to know what you were having trouble with and how you think our product could help") and come up with requirements up front. Charge extra for things like enterprise sign-on if they haven't been built yet.

This customer could make or break your business. If you quote is too low you risk regretting the deal not only on your end, but the potential that your enterprise customer backs out as well when you inevitably hit some stumbling blocks.

Your first enterprise customer is like winning the lottery – it's okay to play, but don't bet everything on winning. Expect to lose, and set yourself up for a big payoff if you end up making it work. Plan to learn from your mistakes either way.