A very old NTSC TV will be 525 scan lines @ 30 frames / second => 15,750 hz.

The source of the sound is the transformer which resets the scan line back to the original side of monitor.

Why is it more noticeable when the computer isn't sending a signal, but the monitor is on?

This is very cool. You can start with an image, generate a mesh for it, render it, and then compare the render to the image. Fully automated training.

continous training

Interesting to hear the author complain about Android, which today is held up as the one part of Google which knows how to ship product.

This has been the best app for years.

What I find both amusing and frustrating is that the linked article uses the word "accuracy" 110 times but never explains what is meant by it. For example in the headline "Over 500 ‘Migrant Caravaners’ Arrested With Suicide Vests", what is meant by rating for accuracy? Is it the number 500 and whether it is low, exact, or high? Is it whether the people were migrants? Is it whether the vests were explosive? Is it whether the story is in any way truthful? I really have no idea. Is "accuracy" an alias for "truthful" or something else? The article doesn't explain.

From the article:

Participants were randomly assigned to then either judge the veracity of each headline (accuracy condition) or indicate whether they would consider sharing each headline online (sharing condition)

So they use "accuracy" as an alias for "veracity". But they don't really need to define it, because the study is about what the participants think is accurate, not the researchers.

EDIT: Now that I've made it down to the methods section, I see that the wording they actually used was "We are interested in whether you think these headlines describe an event that actually happened in an accurate and unbiased way." So, their measurements refer to whatever the participants interpreted that question to mean.