2^1 is one followed by how many zeros?

He is most likely talking about a keyboard with this layout:

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

The improvement works recursively. For example, you can multiply a 25x25 matrix by treating it as a 5x5 matrix of 5x5 matrices. So, instead of 96*96 (9216) integer multiplications, you have 95*95 (9025).

The savings compound, so what was a ~1% improvement for 5x5 becomes ~5% for 3125x3125, and so on.

The graph is fooling your eyes; the growth is still exponential. If you hover your mouse over the last few data points in the "Total Coronavirus Cases in the United States" graph, the numbers are (most recent first): 68211, 54856, 43781, 33592, 24192. Subtracting the previous day from each number, the deltas are: 13355, 11075, 10189, 9400.

As you can see from the numbers, new cases are still accelerating. The percent growth over each previous day for the above samples (again, most recent first): 24.3%, 25.3%, 30.3%, 38.9%. Still nowhere near linear.

Scroll to the bottom of the article and click "Read the full report here". It's unfortunate that the link to the report is so hard to find - every single one of your questions is answered in the PDF.

This includes their criteria (with examples) for what constitutes misinformation. It also includes their full methodology, rationale and limitations (Annex 1 on page 50), which should address your concerns about which 100 videos they included (and why) out of the 5000, and where those 5000 came from.

Here's a link to the report: https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fdrive.google.com%2Ffile%2F...

Except that it is precisely how cryptographic hashes work. If you don't believe it, I invite you to play a game. Choose a random phone number in the 213 area code, give me the first 6 hex digits of the sha256 hash, and I will reply with a list of around 1-3 possible phone numbers (with a very rare worst case of 8 possibilities.)

Here is an example Linux command line that you can use to get the hash:

$ echo -n +12138675309 | sha256sum | cut -c 1-6
4b7da2

This is an example of a known plaintext attack. Knowing or guessing the area code of a phone number narrows it down to 10,000,000 plaintexts. Compare this to the number of different values for the first three bytes of a hash (16,777,216), and it should become apparent why the attack is possible.

I know you were modded Funny, but the Japanese actually use the English loanwords "plus driver" and "minus driver". And in many cases, "plus" may even be more accurate, considering there are several other plus-shaped designs (JIS and PoziDriv for example) that are incompatible with Phillips drivers.

Something didn't seem quite right, but I just couldn't put my finger on it...

You can have any 4-digit ID you like, as long as it's "8374".

Not to mention the hassle of claiming those bitcoin tips as income, when it's not exactly clear yet how to do that properly.

You might have used the regular dictionary instead, and learned that the word "photog" has been in use for over a century. Congrats, you're one of today's Lucky 10,000!

Allow me to explain how things work here at Slashdot.

First, you read the headline. Advanced users might also make mental note of the Slashdot editor who posted the story (this helps to frame your reactions to the story.)

Now, there are two differing schools of thought as to what to do from here. One camp jumps straight from here into commenting on the story, having already taken in sufficient information at this point to begin forming and expressing opinions. The other camp will read some or all of the summary before commenting. They claim the latter method helps them in identifying and avoiding commenting on duplicate stories.

However, at no point should you ever actually read the articles (this was where you made your critical mistake.)

It's just not done.

I've been playing for about a month under Wine without my account being banned. That said, one has to wonder just how "extensively" their tests were done on Linux. Try running it on any 64-bit kernel, and you can't even get past authenticating unless you're using either a patched version of Wine, or the "setarch i386 -3" workaround. The Warden routine gets stuck in an infinite loop without the patch. More details on this Wine ticket: http://bugs.winehq.org/show_bug.cgi?id=30849

Despite their community manager's assertion, there most certainly are gaps in their testing.

If you examine their demo file, you'll find the 5 static JPEGs inside.

http://cdn01.lytro.com/media/lytro/lyt-37/lyt-37.lfp

What exactly was the point of that "demo"?

So, in other words, this is the facial recognition equivalent of a CAPTCHA?