imh

46 ms·

421.
by imh 10y ago
When you have free parameters, you are actually opening the door to a whole family of predictions. You're saying that of this entire family of models, you'll choose the one that best fits the data. Without free parameters, you hav
422.
by imh 10y ago
When your train is carrying hundreds of passengers, the cost of somebody driving it is pretty small per passenger. In a car alone, that cost isn't split, so is more expensive. That makes it more worth pursuing.
423.
by imh 10y ago
As rtx said, the difference is choice. We could save a ton of lives where people choose to put themselves in danger, but we'd save them by killing a smaller number of people who choose not to put themselves in danger. I don't th
424.
by imh 10y ago
I don't think there's a clear answer to that question, but my gut says that would be immoral. What we'd be doing is saving a ton of lives where people would have otherwise gotten themselves killed through their own choices, b
425.
by imh 10y ago
On their webpage, waymo breaks down that 30k into smaller pieces. 94% of those deaths were due to human choice or human error. At least 71% were due to speeding, alcohol, distraction, and drowsiness. It's probably safe to say that driv
426.
by imh 10y ago
Maybe you can think of score voting as expressing differences between candidates. I think X is this many better than Y, but only this many better than Z. But yeah I like approval for the reason that I think score is complicated to think t
427.
by imh 10y ago
That's such an inethical way to do it. You're suggesting that we just let it learn from wrecks in the real world, instead of trying to prevent those wrecks by other means? This, the Tesla strategy, feels despicable.
428.
by imh 10y ago
You know, I have no idea how to make stuff more accessible. If you don't mind, could you tell us about how you browse the web, so others like me can make our content more universally accessible? Any recommended resources for that kind
429.
by imh 10y ago
Having just watched Black Mirror and Westworld, I was totally bummed about how much near-future sci-fi is so goddamned depressing. Yesterday I decided to reread a story I love, "The Lifecycle of Software Objects," coincidentally a
430.
by imh 10y ago
I meant normalize in the social sense that people become more comfortable with google tracking them everywhere instead of feeling creeped out by it. It doesn't have to be new. Most, or at least a large fraction of, google maps and andr
431.
by imh 10y ago
>Because Trusted Contacts works even if a phone is offline, Thelma requests Elliot’s location and in five minutes can see that his last known location was in the middle of the canyon. The cynic in me thinks this whole use case is intende
432.
by imh 10y ago
>The states, if challenged, could assert that a federal ban on marijuana is unconstitutional on its face as it flies in the face of both the 9th and 10th amendments (and possibly the 5th, with maybe a sprinkling of the 14th, depending).
433.
by imh 10y ago
>Federal statues also supersede state law, under the supremacy clause of the constitution. How does that mesh with the states legalizing marijuana?
434.
by imh 10y ago
That arms race is like some bizarre analogue to bitcoin mining.
435.
by imh 10y ago
I think that's the joke.
436.
by imh 10y ago
I think that many of the corner cases (background downloads) are cornery enough to justify giving users the options. The UX is hard, totally, but it's worth it.
437.
by imh 10y ago
>As a result of their fight, the FBI folded, rescinding the NSL and unsealing associated court records rather than risk a ruling that their surveillance orders were illegal. This scares me. I know why we have a legal system that allows t
438.
by imh 10y ago
Is it really? At $50k/year, planning to save for things most people have to deal with is hard. You have to actively plan for it, and still might not be able to. At $500k/year, it's easy. At a million or a billion, it's s
439.
by imh 10y ago
I'm pretty sure open floor plans are so prevalent because they are so much cheaper. I've been places where they actually switched out all our desks for smaller ones so we could fit more employees. These proposals are at least stab
440.
by imh 10y ago
Doesn't matter. All that matters is that the errors are correlated. If voter turnout is biased in one state, then it likely is in another. The reluctance was just an example.
441.
by imh 10y ago
538 made an interesting point about these kinds of biases. Essentially, polling errors are correlated. You can't unbias your estimate, but you can realize that if voters are reluctant to state their Trump preference in Wisconsin, then
442.
by imh 10y ago
Let's say I'm reporting on a study that sets out to see if there is wage discrimination against earthling immigrants in favor of native martians. The study shows that earthlings earn more than martians on average, but once you con
443.
by imh 10y ago
There are two different kinds of fake news. The first type is just shitty journalism. Stuff that isn't fact checked, doesn't tell the whole story, actively misleads, or what not. Thats the kind that has a spectrum. The second type
444.
by imh 10y ago
The general motivation for a p value is that you can model what your data should look like under the assumption that your model is correct, but you can't really say what your data should look like under the assumption that your model i
445.
by imh 10y ago
I think the part I missed was why the solution to a crashing currency is to print more money. Let's say Venezuela cannot produce widgets, but they are necessary to modern living. Venezuela's currency drops in half, so now widgets
446.
by imh 10y ago
You seem knowledgeable. Why can external currency changes (USD to bolivar) cause such strong inflation internally? I would think that at some point no one in the country can buy an iPhone, but how does that turn into no one being able to bu
447.
by imh 10y ago
You can do quite a bit without making strong distributional assumptions. There are nonparametric tests for all sorts of things (and the basic ones are covered in good first intro to stats books too).
448.
by imh 10y ago
I think any intro stats book should do the trick. As far as I know, the material in a first stats course is pretty homogeneous. I'm not a biostatistician, but I happen to like this book [0] for introductory stuff. Amazon says you can g
449.
by imh 10y ago
Nope, we're switching to "AI" these days.
450.
by imh 10y ago
This reads weirdly. He talks about how MXNet is the best choice without comparing it to other frameworks. That's the whole point of choosing between things. I'm sure they did the legwork to make this decision, and some insight int

More ›