Personally, I think social media and academic publishing timescales, rewards and social conventions don't match very well. Social networking feels transient and impersonal. I would like to take some time to form my opinion about a paper, not jump in with a post. Maybe some comment box would be ok to write a couple of nice things about a paper, but it doesn't feel the place to write harsh criticism or have complex discussion where things could be misunderstood. Rather than write in the public record, if I think the paper has a deep flaw I would prefer to contact the authors first. This can be followed up by discussions in your own papers. Others may have different opinions, of course.
A surprising number of scientists seem to use it, likely because of the now terrible atmosphere for scientists on Twitter/X and the emptiness of bluesky.
What's the good of a perfect railway line if it never gets built? What happened to the capacity argument? There is likely a good optimum between the cheapest and most expensive possible for capacity and speed. We could all fly around in supersonic aircraft, but there's a reason we don't.
Isn't the problem that the requirements for line were "gold plated"? If they'd put in another standard rail line instead, it would have increased capacity, taken up much less space, would have been much cheaper, would have caused less disruption and would have had a clearer business case.
Pedantically, it's Einstein Probe (EP), not Einstein. Confusingly, in Chinese it has another name, Tianguan, commemorating the observation of the supernova which created the Crab Nebula.
Stunt Island was pretty good. However, there are more unusual games out there with reasonable budgets, like Death Stranding, The Talos Principle, The Outer Wilds, Portal and X4:Foundations. Even games involving shooting like Control or Alan Wake 2 are driven by unusual story telling.
With google I can't find things I know exist. For example, it doesn't find several of my github repositories with unique names that have been there for years. With old google you could drill down a couple of pages of results, but they're not there now. It also prefers worse sources for materials, for example some blog which poorly explains some API, rather than the original documentation.
Although Intel processors are efficient, modern AMD processors have much higher idle power usage, due to their chiplet design. They typically use at least 20W more power.
Unfortunately the LibreOffice Impress is pretty bad. PowerPoint is what keeps me using Office. LibreOffice formatting looks bad on the screen (it resizes bitmaps to make the fuzzy) and the user interface for constructing slides is both clunky, slow and buggy. I don't understand how the UI can be this slow in 2025 on a fast PC. Basic functionality, like resizing an equation, just doesn't work.
I am the upstream author of the Veusz scientific plotting package: https://veusz.github.io/
Github: https://github.com/jeremysanders and https://github.com/veusz/veusz