Apple's mistake was building *privacy* into their AI model. Nobody else did that, and it crippled Apple's solution. Apple pushed 5GB of AI model data to everyone's phone, and everyone complained about the space usage. Next up their AI is slow because it is using resources on the phone instead of big data centers. Apple did what everyone asked for, but users were ultimately unwilling to accept the compromise.

Personally, I liked the Apple solution better. On Android, if I lose internet for 2 seconds, and say to my phone "Call Bob Smith" it sits there for several minutes then times out with "try again later". BUT IT GETS WORSE: The local hardware transcribed the text perfectly. So there was no need for a server to be involved at all!

Apple's old approach was the right one, so it is really sad that they botched it.

EFF resources on age verification
I thought I understood the issues and options and trade-offs, but after reading this I realize there is much more to it than I considered. This is a really good resource, and some of the articles under it are bite-sized so you can send to non-techies and make them go "oh yeaah.... I hadn't thought about it that way.... maybe this isn't such a good idea....?!"

I am not the OP, but perhaps each vendor should not have to architect this function, since it is such a high security risk. Instead, use infrastructure like apt, yum, chocolatey, Windows Update, Steam, etc.

You are right. And also, the advertiser/data broker won't want the data for "only people who don't pay." They are going to be very interested in those people.

Many people avoid Meta products because of the predatory pricing model. In that model, "payment" means giving up our privacy and control of our data to advertisers and data brokers. Instead, offer a model where we pay directly, and give us control of our own data again.

Any other deal is pointless. Any "premium" service is for their benefit, not ours.

Without copyright law, the GPL would still be necessary. The GPL prevents someone from modifying software that someone else created and keeping those changes hidden.

Powershell *could have been* the best shell scripting language ever. But Microsoft destroyed it.

The core problem with scripting languages is that they rely on lexical parsing of the text output from a command. You run a command to get a list of the employees, and you know that the 3rd column of the output contains a string with the date they were hired, so you used sed and awk to skip 2 commas, then any spaces, then grab the first four digits, and now you have the year of their hire. Great! But this is difficult and likely to break if the output format changes.

Powershell commands returned objects, so you run the cmdlet and grab the "Hiredate.year" property. That won't break, and if it is removed the code *should* throw an exception. And the brilliant part is that it still worked like a textual command-prompt when used interactively, because Powershell displays the objects as text when you typed a command! Amazing! It is the best of both worlds.

And it can call .NET so it has all the power of a real language like C#!

Yet Powershell is the least reliable, most likely to break in unexpected ways, scripting language that I have ever used. My teams have forbidden it from anything more than a few lines because we have been so burned. So where did Powershell go wrong?

First problem: Powershell's CD command doesn't work. The current directory is always c:\windows\system32 unless you do some wacky stuff. There is a reason for this, but it is dumb and not worth it. Just never use Set-Location in Powershell.

Next, even though Powershell has exception handling, it ignores errors and just plows on. So that example I gave about Hiredate.year doesn't actually work. If they change the output, the code silently breaks, completely defeating the entire purpose of the language. Plus you have the old problem of: 1) CD/Set-Location $myDirectory, 2) delete *.*, and if the CD command failed you just deleted c:\windows\system32. Genius!

Next, Powershell has lots of DWIM (Do what I mean) commands. So suppose you call .Count on something, Powershell could decide to return 0 of that thing has no count property. Or it could decide to return the count of properties on it. Or it could decide to return the Count property of the first object in the collection, if the collection has only one element. It depends.

Powershell makes extensive use of "implicit" variables like $_ and global settings. So you can call a command that changes those and your script behavior changes mid-run.

Return values are broken - there is a return command, but it doesn't do what you think it does.

Overall, Microsoft didn't learn any of the lessons from PERL, PHP, or Python about what makes a language unpredictable or inconsistent. They took the best idea to come into scripting in 20 years, then combined it with the worst mistakes of those languages.

The current US average tariff rate is 17%. But go back to 2024, countries were charging small single-digit percentages.

US 2.2%, Japan is 2.8%, EU 2.9%, UK 3.5%, Canada 3.8%

He is referring to the feature where Amazon broke-out the price and the tariff separately, like how it shows the price, sales tax, and VAT as separate charges. Amazon backed-off from the plan after Trump threatened called it a "hostile and political act."

A coworker who is a mechanical engineer developed a web-based tool 100% LLM coding. He had no prior coding experience. It wrote HTML for him, and took his JSON data and just threw it into the JavaScript files directly. It won't scale because the whole thing is client-side, and it added a bunch of junk files that do nothing. But he produced a 100% working tool that does exactly what he wants. He reached out to me and asked how to actually deploy it somewhere for a demo. I made him a web server, and he used the copilot in VS Code to deploy it. I keep asking him how things work, and he shrugs.

This was originally a prototype to show management that this tool is worth developing for real. But it works well enough that the team just uses it internally. He showed me that he added logging, and I asked how it works, and again he shrugged. The logs go into an azure storage blob that he can view. I have no idea how that works since it is entirely client-side. I bet the AI hard-coded some API key or credentials in there - but it's an internal tool so it kinda doesn't matter. At this point he is using features of Azure that I don't even know, so in a weird way I am jealous even though I'm the professional software engineer with 25 years of experience.

Come to the utopian city of Rapture! Where the artist does not fear the censor, where the scientist is not be bound by petty morality, where the great are not be constrained by the small! Want to be stronger, faster, smarter? Plasmids are the key! Book your flight now, use the promotion code GATTACA for a 10% discount!

Even if you embrace the online storage, the even obfuscate where that storage is. So I am reading "Widget Maker v1 Mechanical Design.docx" and I think "Oh, what other documents are there on the Widget Maker? So I want to access the FOLDER that it is in, and it is nigh impossible to find it. It's just in the ether somewhere.

OneDrive destroys the power of the desktop metaphor that Microsoft helped to make so popular. What was great about good ol' share drives is that they looked and operated exactly like a file on disk. But today, just renaming a file is hard because they do everything possible to not show you a useful Explorer/Finder view of things.

I do this exact same thing and they can pry POP3 from my cold dead hands.

IMAP is for people who want to leave all their email on someone else's mail server and be subject to their tools for searching/archiving/storage.
POP3 is for people who want to have full control of their email. I have more than 25 years of emails stored locally - every email I have ever sent or received. I have no interest in syncing that to some remote server via IMAP. Nor do I want my local storage to be "secondary" where I have to go through extra steps to download them and archive them. POP3 is perfect - download everything, leave it on the server for 30 days just in case.

Don't forget: Microsoft said that Windows 10 is the last version of Windows.

The article is correct though: The real reason is that the horrible parts of Windows 10 were optional before, and they are now required in Windows 11.

Also note that the web in general does this same stuff by default, and nobody cares. People log-in to Chrome using their gmail account, then happily browse an advertising-laden web while Google tracks and sells their every move. They log-in to Pinterest and Facebook and Tiktok and whatever, happily sharing their data. So the market has spoken: Nobody cares enough about the surveillance to actually change their habits.

How long before we login to Slashdot with Gmail credentials? This is like the last site left on the "old web." All 100 of us left. *shakes fist at cloud*

I have always dreamed of being cyber-augmented. A camera and a brain implant, so that I could look at a sign and mentally think "zoom in" and read it from a distance. Or look at some acquaintance and have it display: "Joe Smith, at 37. 2 kids Joey and Kaley. Joey just joined Cub Scouts. Kaley auditioned for the part of Dorothy." That would frieking rooock! Or to listen to a speech and have it pop-up with fact checks.

We have the tech!!!!

But the data stream is owned by assholes!!!

When we imagine this, we don't imagine that the video + your location + the last time you made a bowel movement is streamed to an advertising company who offers this as a free service so that they can beam targeted ads into your brain. That's now how this is supposed to work. Do I like the idea that a computer can warn the school principal that a student brought a gun into school? Yes! But not if that means the camera also tracks the student's every movement and knows who is on their period. That's not cool, and it's not worth it.

Technology futurism isn't evil. But the population at large ceded control of the internet to the least worthy of humanity. We can get to a really cool future only if people flock to systems that are designed around people controlling their own data.