I appreciate how convenient it is to have statistical analysis tools available directly in BQ SQL, but are linear and logistic regression really considered "machine learning"?

Actually most of online ads ctr prediction machine learning systems are still logistic regression based until very recently. It's simple but very effective.

Just because it's not a neural net doesn't make it not ML. Also, can echo the other folks here: linear models and basic logistic regression are still competitive.

I wholeheartedly agree with you. I'm suggesting that it might be time to have the counter conversation that just because it has numerical statistical analysis does not make it ML. I've been around the field for a couple of decades and understand linear regression is covered in the first pages of chapter one. If that is the standard, then Excel can claim to have had machine learning capabilities for years.

Yes, it definitely is.

Honestly, those costs were the last thing anyone worried about. There were so many local donations of fuel, water, food, etc. Some of the folks I helped evacuate offered cash donations to cover those things, which we turned down because we didn't need it and knew they would.

The interesting thing is we self-organized. If you had a flat bottom boat and a truck, you helped your family first. Then there were twice as many people asking for help. It was easy to help because so many people needed help. I think technology would have been in the way, to be honest.

During the 2016 floods in the Baton Rouge area, I was part of this effort. Flooding disasters are fog of war. Victims are psychologically paralyzed with disbelief and loss, and there are others who prey on them. It's part of our Southern culture to help one another and get things done. If the CN helps protect the vulnerable while evacuating and saving lives, so be it. You can't sit around and wait for help that never arrives.

Google is the problem. I thought they didn't want to be evil.

If you have to ask this question, maybe you shouldn't have applied for the position.

Because Rust!

Regarding on-host filtering (edge analytics), my experience has been it's because of performance, and I agree with the security angle, too.

This is seriously cool, in my opinion. In fact, I've been working on something kind of similar in Go with more support for network monitoring and vulnerability management that would feed into GrayLog.

Are you guys hiring?

I had a first interview with a company where an engineer walked into the room with an eight page C++ exam. No interview. No meet-and-greet-tell-me-about-yourself. Just write, using pen and paper, several complete C++ programs for really hard academic-style problems (combinatorial optimization, dynamic programming, machine learning). Oh, and you had 1 hour. It wasn't an interview as much as a graduate level final exam.