I'm curious to know what are the biggest companies on the GitHub list. It looks like some of these methods may work fine for a company hiring a dozen people a year, but when you're a large tech company (like your list), hiring 10,000 people a year and interviewing probably 50-100x that, you need a method that scales up and is more cheat proof.

Netflix is in the list.

While the term isn’t very precise, in this context “brainteaser” means precisely the kind of questions asked by Google, etc.

Mmmh, no, I don't think so.

Brain teasers are typically understood as questions such as "How many ping pong balls can you put into an airplane?", and other silly, non-coding related questions.

This type of question has been banned in most tech companies for years.

Given that you are the one who used the term, I’m curious why you used it in the way I described then and not the way you just did. The article here clearly is including LeetCode style questions as “whiteboard questions” — this is explicitly stated. Maybe you hadn’t read the article? That is fairly likely, given that your comments about physical whiteboards are addressed in the article as well.

That's a Fermi problem, not a brain teaser. Brain teasers are problem with one solution that you solve by a spark of insight (or memory) with no solving process.

Serious question, because I see this view all the time: how are the algorithms design, architecture, and coding questions asked by Google and friends brain teasers?

They tend to be contrived, artificial, aggressively timeboxed, unrealistically isolated from normal tools one might use, and the desired solution tends to be one that depends on having one or two central creative insights.

Whether they’re a good or bad interviewing technique is a separate topic — I only mean to describe the usage of the term as I’ve seen it in the last 5-10 years.

(Also, to be clear, not every question asked by Google necessarily falls into this category. System design questions in particular tend not to.)