The answer to your question is, Microsoft structures their offering strategically as part of the sales pitch.

There are lots of different plays, but the most obvious version is, Microsoft salesdroid will ask if you're using Slack and come back with, "tell you what, Teams will cost you nothing, we'll zero that." Now $manager can either pay for Slack too, or replace it for "zero cost".

That discount probably goes away next sales cycle, but in a 20k person company, saving that $5/head/month (or whatever it is) means someone hits their bonus target this year.

Tada!

Snack size?

This is Apple's implementation, not Passkeys per se.

I was confused by this when they announced it, because I couldn't think of any reason why you'd want cloud sync for device keys. And it turns out, you don't.

Apple uses "sync passkeys", not device-bound passkeys, and uses iCloud to distribute them to your other iThingies (but of course not other platforms).

I won't use them either, for the same reason. I refuse to sync authentication creds - I will not store them on something I don't control. And even if I were to do so, Linux would need to be included for it to be useful to me.

Anyway, yeah, it Apple didn't need to require cloud nonsense, but they chose to. I understand where they're coming from, I just don't like it and won't do it.

This "feature" actually saved me quite a bit of work at a job I had a few years back. The documentation people were so afraid that anyone who was not a full-time Word expert would irrecoverably screw up the corporate branding (IOW, formatting) of their docs, they didn't want developers to directly edit them. So I was often able to get away with emailing a quick text summary to them, and they had to do all the fidgety proof reading, formatting, etc.

I don't know how they managed to get their jobs done, given that they had no real source control and mainly juggled each update amongst themselves over email and random impoossible-to-find folders on Sharepoint.

Since all the docs had the same basic layout and they were mainly trying to make them look consistent with whatever corporate branding was being promulgated that week, it could easily have been done by writing them in "markdown" and having a script that converted them directly to PDF. Or maybe even learn LaTeX. Then the docs could all be maintained and diffed in github like all the other project artifacts. I didn't even bring that up because I knew that their heads would explode.

They are taking a complete 180 on a core aspect of their interview.

I predict is will have mo measurable impact on their engineering capability.

I also predict their HR will hail this as a great success.

McMegan is a clown. A wannabe Maureen Dowd without the turn of phrase or hint of self-awareness.

I'm totally fine with people like that trashing Bluesky. By all means, enjoy Musk's blog.

And Xitter is exactly that. Open forums don't look like this.

People who are enjoying a party don't sit around trying to convince themselves that some other party sucks more.

"Anyone who believes that exponential growth can go on forever in a finite world is either a madman or an economist." - Kenneth E. Boulding

I have this mental picture a retirement home full of demented nerds flashing printed copies of Goatse at each other.

The thing that gets me is the consistency. It feels as if there's a union foreman somewhere who has been ensuring at least one is assigned to every story since the late 90s.

There is a school of thought which holds that past a certain scale, effective moderation becomes impossible. I've done moderation work and it took tons of judgement.

But it doesn't. For example, the lawyers that keep submitting court filings with fake references evidently have wrong assumptions about what AI is and is not (for now)

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.washingtonpost.com...

250 is less than 1/3 of 500?

Dude, you've got to update your talking points. You sound like you've been watching Fox reruns on VHS in your trailer for the past 20 years.

A more typical use case for now would be using AI to generate some code, and then testing/fixing the code. Not running the AI every time to solve an instance of the problem.

Side note, I wonder if this paper compared AI performance to human performance. You think people can do towers of hanoi consistently?

Everyone should block any traffic to AS32934 at their border. I do, and the internet is so much better without it.