They work fucking fantastic, until they don’t.

If you have more than 3 brain cells to rub together maybe. But that requires recognition of the knowledge chain challenge -> experience -> learning -> knowledge. Hell even the last two (learning -> knowledge) seems beyond a lot of people.

Personally I blame our educational system. It does a shit job of teaching people to learn. It does a great job of teaching people that learning isn’t fun and should be avoided.
Combine that with ‘stranger danger’ and helicopter parenting keeping kids indoors and keeping them from unstructured unsupervised activities that lead to organic learning through experience, and you get 18yo’s graduating high school with virtually all of their learning being in school settings and virtually none of it fun. So they do it when necessary or when they don’t realize they’re doing it, and otherwise avoid it. It’s really sad.

I got lucky or maybe I just had really good parents. But somewhere along the line I figured out that learning could be a ton of fun if it’s on a subject you’re interested in and if the instruction matches your learning style. So I taught myself all kinds of shit, usually by diving face first into projects I had no idea how to finish and by the time I was done I was good at it.
Like in the early days of VoIP- ‘I’m bored so let’s download some sketchy open source bullshit and route all my phone calls through a random free provider in Washington State. Oh it only runs on Linux? Guess I gotta figure out how to make Linux work then.’ Add a bunch of evenings and weekends of fucking around and without realizing it I’ve now got a marketable skill. And that ‘sketchy open source bullshit’ became Asterisk, one of the most common open source PBX systems around today.

Thing is- if Claude had been around, and I used it, I’d just have said ‘Claude, I need to install CentOS linux, download the latest snapshot of Asterisk out of SVN, compile it, and set up an IP PBX. Write me a script’.
It’d think for 30 seconds and then give me the entire answer. And I’d learn fucking nothing other than how to copy and paste from Claude.

It’s like all the companies that are firing their junior engineers in favor of AI… they are shooting themselves in the foot. Where you do think senior engineers come from? Do you think they hatch from eggs, fully trained and experienced and ready for a technical leadership position? No you moron they start as interns or junior engineers and they gain knowledge through experience. Get rid of the junior engineers and wait a generation or two and there will be no senior engineers to direct Claude.

Depends on the situation. Not every place to live is walkable, or should be.

Past a certain level of population density- ABSOLUTELY. EVERYTHING should be designed with walkability in mind.

Below that though, cars can often make more sense. If you live in a suburb, there’s too much distance to walk usually, and too few people on any given route to justify regular mass transit.

Personally I’m a real big fan of ‘third mode’ transportation- bikes, scooters, powered skateboards, hoverboards, segways, etc. Things that go faster than walking, slower than cars, and (crucially) take up almost no pavement space. Get enough people on those and things can get better fast.

I’ve had this exact conversation in real life.

I do IT work. Someone who was a heavy AI user asked me ‘how was I integrating AI into my workflow’ as if it was a foregone conclusion that I AM integrating AI. I told him I wasn’t, that AI may be useful for finding answers to some things or photoshopping funny pictures but most of what I do is intellectual, specific to certain situations and requirements, which AI generally doesn’t understand. That I can’t risk AI making actual changes to my environments, so if I let AI write a plan and then I proof it myself and then do it, it’s easier just to do the work myself. So that leaves the only thing AI could do for me is writing, and I’m good at writing, so I overall have little or no use for AI.

Dude looked at me like I was an alien. Like he legitimately could not understand why an intelligent person would voluntarily reject AI. Like I said I rejected cars and I ride a horse to work. Does not compute. Here’s this supposedly smart person who knows more about tech than the next 2-3 people in the office combined, and this person is completely rejecting the holy grail.

Honestly she should sue for malicious prosecution. She was driving the vehicle legally with insurance, she proved that to the officer by showing her paperwork, and then, knowing that she had insurance and the complaint was baseless, the officer wrote the ticket anyway.

This should be an open and shut case.

This is hilarious.

I bet Muse got paid. If I were in the band I’d say absolutely sell the thing- the Muse AI thing is going to crash and burn anyway so might as well get paid especially if they’re too dumb to realize MuseAI would be a better handle anyway.

This exactly.

Manifest V2 is not just a plugin, it touches a LOT of the code. Keeping it in the codebase will become increasingly difficult for any 3rd party.
But this isn’t just about Manifest V2. It’s about the very concept of choice- if Google decides you don’t get something, then nobody gets it. Call it open source if you want, if all rely on Chromium then edge, vivaldi, opera, brave, etc are just reskinned Chrome with extra shit.

It’s sad that Firefox and Webkit are really the only other choices. I wish there was more. Sadly these days it’s all about fast javascript and that’s not a super easy thing to build.

For lack of better choices, I agree that Firefox and its variants are what we should be pushing these days.

Shocking, but not surprising.

The more I think about this, the more I think it’s a really big problem how much Web surfing runs on Chromium. While standards are far more prevalent now than in the days of Internet Explorer’s de facto monopoly, the end result is similar- just as MS had control of the Internet through IE, Google gets a lot of control through Chromium.

The recent removal of Manifest V2 is a perfect example. A change literally nobody wanted forced upon the whole world. And while others like MS may be keeping the Mv2 code around, that won’t be tenable forever.

The problem isn’t Mv2 or uBlock or anything else. The problem is control. That one company has the ability to disable something millions of users want, across multiple browsers, because they say so.

And thus the more I think about it, the more I conclude it’s important to get as much browsing OFF Chromium as possible.

Disney didn’t use Steamboat Willie for a while, they didn’t lose it…

I’m not claiming Twitter is great.

But this is obviously going to fail, always was obviously going to fail. It’s a PR stunt, nothing more.

Twitter will sue them, they’ll settle for giving up the name and some token fine, and hopefully they’re on the map. Or, their hail mary long shot will pay off and they’ll get to call themselves twitter in which case holy shit they’re golden. That’s the strategy.

It’s stupid- they’ll lose in court and I doubt whatever they offer is more compelling than what’s already out there.

In general I agree- the concept of red tape, environmental evaluations, etc is a good thing.

But we all know what the road to Hell is paved with.

Key is to find a middle ground- that lets a fab be built in an environmentally friendly way but also rapidly.

Is this a surprise?

Semiconductor production is a national priority for China.

This 100%.

HBM has massive profit margins. It’s also hard as hell to make, as it involves multiple stacked wafers. This vid is a good catchup on that and the rest of this situation.

China will HAPPILY flood the non-HBM DRAM market with memory ‘only’ 2x as expensive as last year. And every PC maker, automaker, gadget maker, etc will happily buy it because as Valve famously put it, “They give us a price every month or something, and they say ‘you can buy that many’ and it’s yes or no, and if we say no then they never talk to us again.” Source interview. If China says ‘we have consistent price, we have consistent supply, we can meet your needs’ and the quality is decent then it doesn’t matter if it’s slower or DDR4 or whatever people will love them for it.

And play that ahead.
The big 3 are focused on HBM and most of their capacity is already sold through 2027
If China can scale up production of standard DDR4/5 and sell it reliably at 200% of normal price (aka, below current market rates) then there’s less incentive to build fabs. So they will let China have the consumer market with their blessings. And thus China will have made themselves essential in yet another product supply chain.

This was easy to predict.

Chinese government is not stupid. They have a strategy (more so than USA does). Years ago that was to dominate global manufacturing and supply chains. They were successful. Now the goal is to reach parity with Western chipmakers and/or get ahead. It’s happening.

The AI boom was like a golden gift to them. Semiconductor prices are through the roof. So if they can produce almost anything at all in quantity it will sell for inflated prices.

While the tech companies of the world jerk each other off with AI, China is building DRAM capacity. CXMT is the first and will not be the last. They will scale quickly. If they have the process perfected they will build a fab in the time it takes USA to conduct the environmental review for the consideration of allowing a permit application to build a fab. They will not build one fab they will build 5 or 10. I’m proud to be American and I love my country but I compare USA to China on the ability to execute large projects like this and China will win every time. We have too much red tape, too many lawsuits and insurance and moving parts. China executes faster and more decisively.

So give it a year or two and China will have a fleet of pretty good fabs. They will sell memory for way less than what the others do. And then just as China made themselves indispensable for manufacturing, they will make themselves indispensable for DRAM.

Yeah it’s awful.

I actually found the best peanut butter at a local store. They have a machine that looks like a drink dispenser but it’s full of different kinds of peanuts- you grab a jar and push the button and it grinds the peanuts into peanut butter right there in front of you. Resulting machine looks something like this.
There’s a few different options like plain, honey, spicy, etc. I just get the plain one and stir in some salt when I get home.

If you let it sit for a while it separates (the oil comes to the top and solids sink) but you just stir it up and it’s fine.

Funny thing is- flavor is actually a good bit better than most of what you get in the jar. And there’s zero bullshit, just peanuts and salt to taste which are all you need for great peanut butter.

I’ve not been able to find any solid intel on what actually happened to the wafers.

Thing is though- cutting wafers and packaging can be done at any fab so in theory they could find an older fab with excess capacity and pay them to do some of it.

Binning however requires some more specialized equipment, that from what I understand is more specific to the type of chips you’re making. This is where you test each individual chip- out of a wafer you’ll get some great ones, some good ones, some bad ones, and some that don’t work at all. Thus ‘binning’ is taking a stream of chips and sorting them into bins by their maximum speed, stability, etc.
You might find a fab that has the equipment to bin DRAM chips but not the means to manufacture those chips. Still it seems an odd use of resources to get into the chip production business when your core business has nothing to do with chips.

I’d like to see a lawsuit like this against OpenAI.

They bought a combined total of 40% of the world’s 2026 memory production- not as usable modules, but as finished wafers. Those wafers have to be sliced into individual chips, which are then tested (binned) and packaged into modules, which themselves get mounted on a PCB with some support circuity to make a DIMM.
OpenAI (as far as I know) doesn’t have the equipment to do that.
So chances are those finished wafers are either in the trash or sitting in a warehouse somewhere.

Question becomes- to what end? To drive up prices for other AI companies? To manipulate the market and then sell those wafers in small batches at great premium?

IMHO- unless they can show those wafers have been processed packaged and put in servers, they should also be sued.

I wanna know what’s happening with all those finished wafers they bought.

If they aren’t being made into memory modules I think someone should go to jail.

Understand that people are speaking from frustration. Users are tired. We are sick of having things we don’t want shoved down our throats and told that we want them. This is not just Firefox, this is most software.

If I was Mozilla CEO I know exactly what I would do. I would double down on the users.

Immediately put out a press release that Mozilla will not for as long as I’m in charge make one single dime selling user data. Put in our very corporate charter that we are required to collect as little data as possible to make our products work. Also make a public promise that any AI features which aren’t 100% local will require a very big opt-in and we will try to avoid shipping any such things at all.

Focus on speed. Chrome started getting market share in the first place because they advertised it could render a web page in under 100ms. So that’s what I would shoot for. Screw everything else, the main rendering parts of the browser should be fast, threaded, and stable.

Part of that would be to include some script selection processes in the browser itself. This would partially be like an ad block but more like a priority system. Right now you go to a news website and there’s a good chance you’re pulling tens of megabytes of JavaScript that tracks everything and actually runs a fucking auction in your browser where advertisers are bidding on the right to show you an ad. This does not help the user. So I would focus on developing a system that identifies what JavaScript code renders the bulk of the web page and what is for things like ads, the add code goes dead last. That way the content of the page loads very quickly.

Then I would basically license ublock origin and include that functionality in the browser itself. I would throw Dev time at optimizing the hell out of that. And that would be one of the questions asked at first run, do you want to block advertisements? If user says yes then ublock is enabled. That alone will probably get a shit ton of users, because it will do the same thing as Chrome did years ago, just make the experience of web surfing better.

I would stop reinventing the UI every two years.