I asked ChatGPT to generate a Seinfeld episode for me. It generated a story where Kramer starts to collect mustard sauce, and becomes obsessed to find the perfect one. I thought the idea was hilarious. But then I asked it to generate the dialogs, and they were crap. (But, then again, it was a Seinfeld episode, maybe the dialogs were on point).

That's how Joshua Wars II started.

When I taught programming at college, I actually had to teach students to read error messages. Not once or twice, but multiple times every semester.

I tried it to see what kind of results I could get with it. It happens that I have some different implementations of the Fast Fourier Transform that I use to benchmark these kind of things. What I found out is:

It doesn't implement every Python module. I couldn't get the array module to work. But this might be in their future plans.

It can get a little picky with variable types. For example, multiplying an integer with a complex won't work, or trying to print an integer using a floating point format. Maybe they're working on it too.

It can take some time to compile and run.

While the scripts indeed run faster, it was nothing close to 10x the speed, much less 100x. In general, I got a 2.5x speed up. It was outperformed by Pypy in every test I made.

Just for the record, I have the same algorithms implemented in C, and Pypy performs comparably to C. Disclaimer: they are not optimized, instead, I made an effort to make the same operations as much as possible, with the intent of comparing speeds. Also, not a scientific assessment, so take it with a grain of salt.

There isn't really any proof that Waymo vehicles are driverless, only that the driver isn't in the vehicle. If you compare a dozen dubiously driverless vehicles to a million collecting data every moment and already doing a pretty good job at autonomy, which do you think will succeed?

The fact that the driver is not in the vehicle does not mean that there is no driver.

I believe that this was figured out in the amateur observer community before NASA confirmed it. I saw it go by in my twitter feed. Not sure I can find it now...

Linear Energy Transfer

It's not radioactive. When high-energy particles strike it, they break apart into lower-energy particles that are more harmful. So, you need enough shielding to catch the particles that break apart in the top layer of your shielding.

While radiation mostly results in transient errors, it can also cause latch-ups that blow a gate off of the die. For rad-hard equipment in space this is mitigated by silicon-on-insulator fabrication, so that you can't have a latch-up between the substrate and a silicon feature.

We could have REAL ZOMBIES! YEAH!

:-)

I hear you. Going to be an interesting future, if we live.

Someone who broke their neck and was suffering from paralysis. You can control a chair or exoskeleton. People who can not hear today have cochlear implants, this is not all that different and might (eventually) work better. Or speak, or see. Other people who are disabled in various ways.

Oops. To continue: I'm just not seeing that it is so hard to access. But then, I use a Linux desktop, which has always come with a proper PDF renderer, there has never been a need to download "Acrobat".

Web pages use SVG to render vector graphics. It uses the exact same imaging model as PDF and is implemented in all modern browsers. The web in general has taken a lot of lessons from Adobe because Warnock and Geshke, in the PostScript Red Book, got so much right about how to build an image model that many GUI developers are still learning today. If you start with a PDF, it should be possible to machine-translate it to SVG and present it as a web page.

PDF exists because it is trivial to generate it from the document renderer meant for printing. Although I have once in a while run into an improperly scaled PDF meant to be printed 8-up, I'm just not