My impression is that most nutritional research is just p-hacking on extremely noisy data. The only consistent outcome is that eating too much for an extended period of time is extremely unhealthy, regardless of what you are eating. Unfortunately, this runs contrary to the "more is better" mentality of US consumers who throw a hissy fit if food portions are not gigantic.
iRobot refused to incorporate LIDAR scanners, instead relying on some janky computer vision approach. Robots from Roborock and friends know their position with cm-accuracy at all times, whereas Roombas only have some vibe about which part of the building they are in. This makes them much less efficient at route planning and less reliable at adhering to exclusion zones.
The best silicon single crystals still contain impurities at the 1E-11 level. This project is about doing crystal growing in low gravity (the ultra-high purity is only achieved due to the growth process). The vacuum of space is a lot worse than what can be achieved in the lab, especially in low orbits.
Tesla is probably the only EV maker with declining sales for the last two years. Quite a feat in a booming market, and remarkable considering that the stock already has a few orders of magnitude of growth priced in.
Fiber splicers are marvels of technology. They align the fiber cores with sub-micrometer accuracy and produce just the right amount of heat and pressure to melt the ends together. They are also usually very rugged, fully automated, and surprisingly cheap (a few thousand euros). It is remarkable what is possible when the entire internet relies on a technology.
This doesn't match my experience. My previous three laptops (two AMD Lenovo Thinkpads, one Intel Sony VAIO) had essentially the same battery life running Linux as running Windows.
Yes, this is called Bluetooth multipoint and has been common on non-Apple devices (for example Bose) for a few years now. Requires no logins and is vendor-agnostic.
But Vivado doesn't get the job done. The intended workflow is to click around in the GUI until it (hopefully) synthesizes something. The state is then recorded in some proprietary project file that cannot be version controlled or shared with other developers. The workaround is to generate some unholy mess of tcl scripts that automate the clicking, such that one can start from scratch for each synthesis. The scripting mess breaks with each minor release of Vivado, so you need to either never update, or have a separate (~100 GB) Vivado installation for every single project. And if your chip is more than a home-gamer ZYNQ, you hopefully like paying subscription fees for the experience.
That's not true. Transistors were commercialized a few years after their invention, and already the first generation vastly outperformed vacuum tubes in size, weight, and power. Optical computing has been done for a few decades now with very little progress.
Indeed, most societies ended up inventing a mandatory trusted third party escrow called a "legal system" as part of a "state". They usually issue hard-to-copy tokens, solving the double spending problem.
There was no such thing in the UK. ChatGPT is trained to produce text that fulfills the user's expectations. If you put a prejudiced prompt in, expect a corresponding result.
It's an ingenious solution to achieve a "trustless" currency that prevents double-spending without a central authority. Unfortunately, this solves the wrong problem. Spending money usually involves getting a good or service in return, which inherently requires "trust" (as does any human interaction). Your fancy blockchain is not going to help you if you order something with Bitcoin and no package arrives.
GNU Taler is a working implementation of "digital cash" in the spirit of Ecash. Since it doesn't come with its own currency, it cannot be used for gambling. It is quite telling that it has seen essentially zero adoption in the "crypto" scene.
No, it wasn't. Scientists at CERN used DVI and later PDF like everyone else. HTML has no provisions for typesetting equations and is therefore not suitable for physics papers (without much newer hacks such as MathML).
I quite like Sage. Python is a much better language than Wolfram (yes, he named it after himself...). In Wolfram, there is no real scoping (even different notebooks share all variables, Module[] is incredibly clumsy), no real control flow (If[] is just a function), and no real error handling. When Wolfram encounters an exception, it just prints a red message and keeps chugging along with the output of the error'd function being replaced by a symbolic expression. This usually leads to pages and pages of gibberish and/or crashes the kernel (which for some reason is quite difficult to interrupt or restart). Together with the notebook format and the laughable debugger, this makes finding errors extremely frustrating.
The notebooks are also difficult to version control (unreadable diffs for minor changes), and unit testing is clearly just an afterthought. Also the GUI performance is bad. Put more than a hand full of plots on a page, and everything slows to a crawl. What keeps me coming back is the comprehensive function library, and the formula inputs. I find it quite difficult to spot mistakes in mathematical expressions written in Python syntax.
Maybe with the power of a supercomputer, Mathematica can finally launch in less than 30s. I have no idea how a software that still does essentially the same thing as it did in 1988 can be that sluggish.
My biggest worry with Immich is how to future-proof the albums. With photos sorted into folders, it should be no problem to access them in a couple of decades. With Immich, I have to rely on the software still working or finding some kind of tool to dump the database.
This is like comparing a Casio to a Rolex. Both do roughly the same thing, but the markets are completely different. Nobody buys a high-end luxury car like a Taycan because it makes financial sense. The manufacturers know this and price everything accordingly.
US car regulations are weirdly inconsistent. Sometimes they are incredibly strict. You can't have a convex left side mirror and the right one has to carry a stupid warning label. Importing non-antique foreign cars is practically impossible. But then, some obviously unsafe features, such as indicators in the same color as the rear lights, are perfectly legal.
I didn't even feel particularly safe as a pedestrian in Tokyo or Osaka. Despite the good public transport, Japanese cities have cars absolutely everywhere, even in tiny streets that should really be pedestrian zones. Paris is much better in my opinion.
Except these seamless modes have been broken for many years. Also using a containerized Windows means one doesn't have to fiddle with the insane Windows 11 setup process and TPM issues.
It's pretty good. They use XfreeRDP to remote into the container and display individual windows. This somehow performs a lot better than the GPU emulations of VirtualBox or VMware. I guess Microsoft put some effort into optimizing RDP for Terminal Server applications.