Correct. I’ve found ChatGPT chokes after about 10,000 lines of code. Beyond that, you’d have to start modularizing stuff a lot. I have little interest in doing that, so I’ve mostly restricted what I’ve done to smallish projects. One day I’ll take on a bigger one, but in the main I’m not interested in being a programmer with or without AI.
It is an anti-interest of mine. I only do it because it must be done.
When AI gets bigger context windows so I don’t have to manage so much myself, I am going to start altering Firefox and making it better on my own.
First goals:
I’ll probably think of more. I could do almost all of this now with Claude Code + ChatGPT. It’d just be too much of a time investment at the moment. In a year or so as AI improves, I’ll give it a go.
I’ll probably never release it. It’ll just be for me. But for the first time in history, it’ll be possible for people who don’t want to be full-time programmers to unfuck clown-ass decisions by absolute idiots.
Eventually, I will go with something like this when all-SSD NASes become practical for home use.
25Gbs NICs are fairly affordable now, specially if you buy old but-still-fine off-lease enterprise gear. I technically already have infra endpoints (NVMe SSDs, etc.) that can push more than 25Gbs across the network if I had the switches and NICs to support doing so, but I just wouldn’t use those capabilities that much. But when I’m tossing 8K video and VR stuff around it’ll be a lot more useful.
And then I’ll show that data the meaning of haste.
5 hidden bottlenecks that are quietly killing your home network speed.
Trust me, this is not a problem I have. My network is insanely blazing fast.
It’s wild that I’ve been using computers since 1980. I can’t even really remember a life without them.
Obviously, they’ve changed an enormous amount and are no longer “bicycles for the mind.” But I’ve had some fun along the way, at least. And that matters.
Stonewalled by Citrix’s new AI “Customer Service” model : sysadmin
This is so common in my industry. I’ve had to do that with HP, VMWare, Citrix, and others.
Me: I’m calling about your SDN product. (Note that I called the correct support line.)
Them: We don’t offer SDN at this time.
Me: Yes, you actually do. I am using it and have five licenses I bought six months ago.
It’s amazing how often that sort of thing happens. The idea that corporations are “efficient” is only something a true libertardian could believe.
1328244 – [WebExtensions] Firefox customization API request
This is when Firefox actually died as a real competitor in the browser wars.
I and others told the Firefox devs exactly what would happen if they went down the locked-down path. They still did the stupid anyway. Which just goes to show you cannot make people think. They simply will not do it. Perhaps cannot.
Damn.
You’re talking 7–10 kW of power to run something like that, depending on current disk sizing and exact server hardware used. Depending on where you live just running all that alone would be $1,300 a month in power costs on a residential plan.
That is some serious storage.
Sopro TTS: A lightweight text-to-speech model with zero-shot voice cloning.
This tool only requires 3-12 seconds of audio to do voice cloning. And it’s not even a very large model. The Reddit “wisdom” is that voice cloning takes hours of audio and/or is “impossible.”
And, sure, voice cloning did require that much audio — in 2018. Things are changing faster than people can keep up.
Unfortunately, jobs like technical writers, editors, proofreaders are becoming useless.
The commenter in the image above is correct. MBAs don’t care that it’s good. They care that it’s cheap. No matter what any of you think or your most impassioned pleas, that will win the day. Or lose, depending on your perspective.
Really glad I bought 128GB of RAM when it was at its very lowest and a separate machine with 128GB of memory in it as well.
That was good timing given mem prices now. The 128GB of RAM sticks would cost 5.5x as much now.
I worked for a sub-unit of a German company that made exactly those kinds of products. And I kid you not, their unofficial motto was, “We do not innovate.”
And in this case it was exactly what our customers wanted. These were all extremely large (many Fortune 100) enterprise customers that prioritized the tried and true, the stable, and the predictable. They did not want anything that changed like the wind every month or two, as is the usual tech industry way. If a product had been out and tested for 10+ years, they might consider using that “new” tech. And then adopt it over the course of the next decade.
More of the tech world should be that way.
This is the sort of thing I saw someone just a few weeks ago claiming that AI would “never be able to do.” Well, it’s doing it now. It’s only going to get better, too.
Too many people commenting on AI are only familiar with what it could do three years ago, or even a year ago. And that’s not good enough. These days, that’s like being familiar with a 1914 car or a 1928 car.
The OS is the malware. The ads are just evidence that you’ve installed malware.
Back to this. IPv6 makes sense if you have more than a few hundred thousand endpoints and a large engineering staff to manage it all. It’s tons more complex and fiddly than IPv4, so if you have, say, three full-time network engineers for IPv4, if you put in IPv6 you’ll need around a dozen (or one me, but I am rare).
It then does in fact totally make sense! Since IPv4 will always be around and IPv6 is as mentioned vastly more complex and often poorly implemented in hardware, you’ve just increased the difficulty of managing your network 50-fold at least. But some other problems do in fact go away.
Which is why mobile providers use IPv6 — they have the large full-time networking staff to handle it and millions of endpoints which makes even poorly-designed IPv6 a sensible choice.
And all of the above is exactly why no one else should use it.