aab0

24 ms·

181.
by aab0 10y ago
> The length of time a copyright can be held on any IP has been continuously lengthened over the past 20 years, making sequels a more tenable venture for movie studios. Strictly speaking, it was lengthened once, 18 years ago, with the h
182.
by aab0 10y ago
That is terrifying. Propaganda for any cause should not be that easy. This makes me even more worried about Trump.
183.
by aab0 10y ago
The ending is a little eyebrow-raising. Both founders are ousted, refuse to talk about it, and we get some pablum about how they are now a 'team' (rather implying that they epically weren't before). What was going on there in
184.
by aab0 10y ago
Normally when you think of an ASIC, you think of a single chip custom-designed for a single customer/buyer. You don't think of a single chip with multiple totally different functions and designs consolidated onto it, which get ena
185.
by aab0 10y ago
Some of the points are bogus, like the rare earths thing; they used to be mined in the USA just fine. But China was cheaper, especially since they didn't price in the negative externalities.
186.
by aab0 10y ago
This reminds me of the similar thing in Microsoft Office recently: it automatically shortened all links in documents, even private ones, with a URL shortener service. URL shortener services are, by design, easy to brute force enumerate. Cue
187.
by aab0 10y ago
Thiel's plan was exactly to wear down Gawker by frivolous lawsuits until either they collapsed or he lucked out; that's why his legal team was searching through thousands of articles looking for people they could contact to drum
188.
by aab0 10y ago
Makes sense from a bounded rationality point of view. We want to be as selfish as possible while escaping punishment and social ostracism; this requires modeling whether others will learn of selfishness, what the punishment will be, what th
189.
by aab0 10y ago
Given their age, it's unlikely they'd both die at the same time and be unable to open source it. Unless they dig in the wrong place in their backyard, I suppose.
190.
by aab0 10y ago
> Wages rose 2.5 percent over the past year, which isn't rapid but does indicate that companies are starting to compete for hard-to-get workers. So, not really.
191.
by aab0 10y ago
I think if you watch a child play anything, even a video game, you'll see that when it's not immediately obvious what to do, a human will just 'mess around' and 'try stuff out' to 'just to see what happens
192.
by aab0 10y ago
Sounds like it. Take into account the impending fine, all the lawsuits in many countries, the family control of the firm... Takata at this point could well be worth nothing at all.
193.
by aab0 10y ago
A highly one-sided presentation. Let me point out some issues: it makes major hay of Moving to Opportunity, but MTO is a notorious failure - it increased crime rates in targeted areas, and showed no improvements on anything until the cited
194.
by aab0 10y ago
Google certainly benefited a ton from all the 'dark fiber' and the glut of bandwidth after the crash. Probably didn't make much of a long-term difference because it didn't solve the last-mile problem to consumers. (&#x27
195.
by aab0 10y ago
"39. Re graphics: A picture is worth 10K words - but only those to describe the picture. Hardly any sets of 10K words can be adequately described with pictures." It's the pigeonhole principle; there are only a few long videos
196.
by aab0 10y ago
> I think you've grossly underestimated the use cases and budgets of the users. I think the point of OP, and of the other magazine article linked in the comments, is that he hasn't 'grossly underestimated' it in the l
197.
by aab0 10y ago
Again: convolutional filters existed long before CNNs did. Yet no one solved ImageNet with some convolutions and hand-engineered features. You can't get that performance even if you shift a set of hand-engineered feature across an imag
198.
by aab0 10y ago
"Here is my personal answer to the second question: deep neural networks are more useful than traditional neural networks for two reasons: The automatic encoding of features which previously had to be hand engineered. The exploitation
199.
by aab0 10y ago
Deep learning can do question-answering.
200.
by aab0 10y ago
This has reminded me to sign up all my emails for alerts, and makes me glad I started using Lastpass back in 2012 and almost all my logins these days are separate and very long passwords.
201.
by aab0 10y ago
Is adding 'SEC. 222' the only thing the law accomplished, and it did nothing else and set no precedents and had no consequences? Does objection to a bad overall law mean you must then refrain forever from using any aspect of it? D
202.
by aab0 10y ago
Computer software and hardware are an interesting case where often, the more expensive something is, the worse it is; the best stuff is often quite cheap. This comes from the high upfront costs meeting very low marginal costs. A cutting-edg
203.
by aab0 10y ago
Amara's law: we overestimate the impact in the short run, and underestimate it in the long run.
204.
by aab0 10y ago
The point is that a whole fleet can rack up daunting numbers in a very short time.
205.
by aab0 10y ago
> “Since introducing this hardware 18 months ago we’ve accrued 780 million miles,” said Anderson. “We can use all of that data on our servers to look for how people are using our cars and how we can improve things.” Every 10 hours Tesla
206.
by aab0 10y ago
> You're comparing a single system developed years ago to everything ever produced with neural networks I am comparing a single system and all its variants and followups to another family. Oh wait, there aren't any variants a
207.
by aab0 10y ago
Absolutely. The point of an autoencoder is dimensionality reduction: boil a big set of data down to a few hundred or thousand numbers in a vector which summarizes it. You could treat it either as lossy compression and store just the encodin
208.
by aab0 10y ago
I would be more optimistic here. AM/Eurisko never showed any replicated results outside of one or two toy domains of simple math & the Traveler's game, and there's always been a lot of questions about how much of that muc
209.
by aab0 10y ago
He only trained it on Blade Runner, I think. (This is doable because a single movie has a lot of frames.) So all of the other movies should be out of sample.
210.
by aab0 10y ago
But they do change the function. You use RELU now, instead of pretraining.

More ›