nifty

Yeah, it’s amazing how good SQLite is nowadays. I find I use it for a lot of local apps I make for myself cause it’s just good enough.

I’ve come to the same realization, the less moving pieces you have in your stack the better. Postgres can handle vast majority of persistence needs, and can even double up as a queue. So, there’s really little justification adding another moving piece which you’ll have to synchronize data with.

You’d need a pretty high resolution source image plotted for that to work well, modern digital cameras have 20+ megapixels, and if you take a picture of a low resolution image it’ll be very obvious. I’m sure it’s possible, but definitely not that easy to do.

This concept describes a hardware anchored proof system for proving the originality of and the precise sensor origin of photographic media. The camera hardware would use a combination of secure enclaves and asymmetric cryptography like RSA to digitally sign images at the moment of capture making it possible to verify the physical origin of the file. The camera manufacturer could operate a Certificate Authority where users can validate the embedded signature against the registered public key of the specific device. This architecture would allow verifying that an image is an actual sensor derived photograph.

The neat part here is that the algorithm survives retouching of the photo, so you can still do typical things like cleaning up contrast, color balance, etc., without destroying the authenticity of the image. Given how good generated images are getting, it seems to me that deepfake detection approach is dead in the water, and something like this would establish a proactive foundation for cryptographic provenance of real photographs.

a bit different, it just explains how Prolog works and where it’s handy

Alternatively, it’s basically an advertisement for any company that wants to spin up this sort of infrastructure fast.

I was curious about the NDA bit as well, but I guess it’s the sort of stuff you’d talk about in a tech interview as well. He’s not revealing any deep secrets there. And while none of this stuff is necessarily rocket science, it is always interesting to see how big companies structure stuff that needs to run at scale. A lot of this stuff gets figured out through trial and error, and you can save some time by learning what worked for others.

the video https://www.youtube.com/watch?v=55pTFVoclvE

yup

they don’t have https, there are some sites that still run on http believe it or not

yeah lots of nostalgia there

neat

Yup, that basically sums it up. The key part though was that doing the inefficient thing first allowed them to make progress and to understand the problem better by working through it, which later led to a more clever solution. The moral of the story here is that it’s always better to solve the problem even if your solution isn’t optimal, and then reflect on it.

And I find this extends to a more general principle of just pushing through solving a problem in order to understand it. Then once you’ve done that work, you can find a much better solution. It’s easy to get stuck in an analysis paralysis loop where you try to come up with the perfect solution up front, but it’s rarely possible to do for something you’re doing for the first time. Doing a ‘dumb’ thing like making a giant mysql db that gets you to a working result is the right way to go about it.

here’s a decent music player https://github.com/tramhao/termusic and for file management, midnight commander is an incredible tool https://midnight-commander.org/

The key point being made there is that we can separate the intention from the implementation details. The key aspect of programming isn’t in banging out lines of code. It’s understanding the behavior of the system being implemented, the states that it flows through, and how the user interacts with it. These are the core skills, and they lie at the level of abstraction where there is a significant overlap between coding and design.

I don’t think it’s so much that we don’t need skills to work at a detailed level of code anymore, but that we shouldn’t see them as an essential part. Like you said, there are still people who know how to write assembly by hand, but they’re few and far between, working in specific niches where extreme optimization is required. A general coder doesn’t really think about what’s happening at hardware level at all.

My read of the article was more that we should expand the conception of what we mean by coding to include people who work at a higher level of abstraction as well.

I’m guessing the browser isn’t rendering it correctly?

Yeah, they’ve done a really good job with the tool keeping the model on track. Hopefully, now these tricks will start getting integrated in tools like opencode.

Almost certain that they’re going to study this and take what’s useful. Worth noting it’s not the model itself, rather it’s a harness for agentic coding. But it turns out making a good harness makes a huge difference for the models https://blog.can.ac/2026/02/12/the-harness-problem