yeah, we called that data mining, decision systems, and whatnot... mapreduce was as fresh and hot as the Paul Graham's essays book... folks were using Java over python, due to some open source library from around the globe...
essentially, provided you were at a right place in a right time, you could get a BSc in it
Yeah, no - quite a chunk of IMO problems are planar and 3d geometry, and you don't really do that at university level (exception: specializing in high school maths didactics)
The cheat code is to substitute it with something like rollerblading. But you'll need to practice it ~3x longer each time, and aint' nobody on HN got time for that.
It is 'incredibly efficient' because it is incredibly good at predicting clicks, conversions, or even conversion values. Which in turn makes it efficient. Sure, there is something called "auction" there, but Sothesby's or Tattersalls generally don't have buyers bidding based on what some machine-learning prediction AI computed in a jiffy (or maybe they do these days, who knows).
Yeah, but this convenience goes well beyond the "one payment button".
If you order food directly, you won't have the delivery tracking on the map. Even within the app, if the restaurant provides their own couriers, you lose the visibility and arrival ETA info.
And 15% might look impressive, but if you are getting your food from a delivery app, you probably don't care that much about food price in the first place.
There is a section 2.1.3 "Online platform studies versus lift tests" in the article. For the marketing tools purpose, you can use either (or some mixture of both). There are pros and cons to the choice.
In my humble experience some time between 2006-8 and 2013. Unless they hid it that well from us interns.
It's not the only thing that changed. Good thing, my manager joined Google back in these older years, so, for instance, he could say to me that I was "expected to rise to L5" in such a way that I knew it wasn't enforced in our org.
The values of G1 and G2 are computed by a complex algorithm, however, that algorithm is agnostic of the position of the ad in the auction. Unlike the constant factors (1+a) and (1+b) applied on top of that.
Other companies in that auction could apply this kind of optimization, too. Perhaps the improvement is not as large for smaller participants, and so, not worth looking into.
At that time, the exchange was a second-price auction, and all parties could submit up to two bids (presumably, the top two bids from their own collection of advertisers). Let's call the Google bids G1 > G2.
Since Google already implemented automated bidding strategies, they would submit to this auction (1+a)G1 and (1+b)G2 for certain fixed small value parameters a,b. Project Bernanke computed on historical data the optimal values of these a,b parameters.
Cue government discovery misunderstanding documentation
An oddly dramatic response to a blog post of someone working on a, what is this flutter, some frontend framework library?
Counterpoint - since I left Google, little birds told me good things happened, for example, cranking down on the travel expenses (that higher-ups used to spent with little to none oversight)
essentially, provided you were at a right place in a right time, you could get a BSc in it