You're missing the point where: in complex systems, sometimes optimizing code is both a high effort undertaking, and can totally not pay off. Having done hundreds of such exercises on our software over the years, it's liberating to have an idea of how to make something faster, being able to validate it without the fear of having to throw it all in the trash if it fails after days of work. What is still important is being able to provide proper guidance - we even built new tools to allow an AI agent to analyze memory usage in more depth, and instructions on how to benchmark in cloud environments where shared CPU usage and VM reallocation happen all the time and confuses the AI all the time with measurements

Also, not every task is worthy of applying (human) galaxy brain consideration. Sometimes a task is satisfied by cheap + dumb work.

They added a whole set of skills and python scripts for interacting with office files using libreoffice headless, it's not writing the XML files directly

Right it's mostly python scripts instead of generating the entire XML, but they do make changes to the XML directly to get around python-pptx limitations: https://github.com/anthropics/skills/blob/main/skills/pptx/S...

Ironically, the Claude PowerPoint add-in actually does a lot of direct XML because the Office.js library can't even do charts

For me it was even worse than that. One laptop has copilot blocked by policy, and it shows a blocking modal every time you press the key by mistake

If I remember correctly the status page was not precise before the acquisition - so take with a big grain of salt the 100% pre-acquisition values

I remember the status page being quite accurate before the acquisition.

I don’t like this whole casting of doubt upon sources without providing superior or even alternate sources.

It makes it hard to discuss when one person presents a source and another says “I’m not sure that is accurate.” In a vague way.

What am I supposed to do with that? Research more sources that may or may not align with how you feel?

It's worse than that - the feature exists in the paid OneDrive app, it lets you scan, edit, add pages etc just like the office lens. It just doesn't save the output - scanned files just disappear in the void.

Seems like this is going to be the year of AI slop being released everywhere by Microsoft. Just wish they'd put as much effort into a post morten for this one as they're doing for a diagram on a blog post https://github.com/microsoft/onnxruntime/issues/27263#issuec...

Maybe I'm not seeing it but wouldn't it be just easier to pump the river water back up instead of letting it flow on the sea?

Except that the amount of water needs to be augmented as it disappears through usage and evaporation. So you’d run out pretty quickly.

Large scale desalination might be necessary but I think the article is downplaying the brine issue. I don’t think we’re running out of table salt so trying to use it for harvesting metals will be an environmental challenge at the scale being proposed here.

I feel like the brine issue is a red herring. Coming from first principles the ocean is unbelievably big such that no practical amount of desalination will change the oceans salinity. (Plus most desalinated water will make it back to the ocean eventually anyway)

It really is just a matter of distributing the brine water over a large area.

There are some minor breaking changes like the order of iteration is not always the same as the official Linq implementation, or Sum might give different values due to checked vs unchecked summing. Probably not an issue for most people, but a subtle breaking change nevertheless.

See here for more info: https://github.com/Cysharp/ZLinq?tab=readme-ov-file#differen...

To add on that, you can define your lambdas as static to make sure you're not capturing anything by mistake.

Something like dates.Where(static x => x.Date > DateTime.Now)