25% of American adults use "Do not track", but hey, non-geek Gizmodo visitors obviously have more trouble finding the "Do not track" option in their browser settings than the average American. So much for the credibility of the Forrester Research study. And for Gizmodo, wittingly distributing bogus research results.
Agreed. The main risk of a self-driving car is probably not in reacting too slowly, but in scene detection breaking down. For that you need better architectures, not faster architectures.
"We had the benefit [...] of knowing what our neural networks look like, and what they'll look like in the future,"
Really? If they take their neural network development seriously I don't think they know what their networks will look like in ten years. It's a research area in the middle of a transformation. Using architectures molded into hardware is probably just costly and will act as an antagonist to innovation. I don't think having 2000 vs 200 frames per second right now outweighs that downside.
If you would be sure about the 90-95% efficiently then there would be no need to be "leaning towards" the infeasibility of doubling that efficiency, isn't it?
I hope 1M is enough to fix that mess. Successfully switching between user sessions is a matter of luck; Useful panel functionality is only provided through sometimes abandoned third party extensions; For anything other than the five-or-so options in the system settings you need to resort to something like a "tweak-tool"; Then there is the wildly unnecessary "swipe up screen" that get's in your way. All in all a very bad experience. I'm switching back to 16.04 for now.
I think what they actually mean is that the costs of hosting a website and a couple of pdfs should be provided for free either by public or private funds, just like providing, editing and reviewing the scientific content is done for free by the people who signed the open letter.
I can't believe a fee of $30 per online article copy is in any sensible way proportional to the actual costs the publisher has to provide that article.
That's a lame argument. It implies a very narrow reading of "keeping people safe", a term that is wide open to interpretation. An opposing argument may equally cite keeping people safe as a reason for strong encryption.
The problem with these kinds of standards is that they usually settle on the cheapest, crappiest thing. MicroUSB is the perfect example, exploding/burning/shorting chargers are another one.
I fully disagree! My chargers have never exploded. I thank the EC on my knees everyday for imposing standardization of power supplies for phones.
Whether crypto-mining or not, some pages seem to use a disproportionate share of cpu time for the content they're delivering. Some form of cpu usage indicator per tab would be helpful, similar in vein to the speaker icon on tabs that produce sound.
Although to me this is clearly a case of A.I., parent has a point that not all types of problems lend themselves to be treated as reinforcement learning tasks, which seems to have been the key to success in this case.
Yeah, but I think it's obvious that popularity of information in Google's ranking algorithm is used as a proxy for truthfulness/correctness (or in a wider sense: utility), and by and large that approach seems to work well. However if it turns out that in certain (predictable) circumstances popularity is not a good proxy, fixing that is an improvement.
If I query Google for information, I am not interested in falsehoods just because they are popular.
If you are afraid that Google won't serve sites telling you that earth is flat, Obama is a muslim, or the Las Vegas shooter was a f*ng n*r, you can add an extra term "fake" to your query, just like you have to add "xxx" if you want to see pussies.
The frequent stops of buses at predetermined locations may be an opportunity to recharge the battery at these locations (half a minute of charging every five minutes amounts to 6 minutes of charging per hour), perhaps eliminating the need to make a long stop for a full charge.
Slashdot Mirror
25% of American adults use "Do not track", but hey, non-geek Gizmodo visitors obviously have more trouble finding the "Do not track" option in their browser settings than the average American. So much for the credibility of the Forrester Research study. And for Gizmodo, wittingly distributing bogus research results.
In the sense that people in China using Google search won't see what their government does not want them to see? Congratulations Google!
Agreed. The main risk of a self-driving car is probably not in reacting too slowly, but in scene detection breaking down. For that you need better architectures, not faster architectures.
To speak with Knuth, premature optimization is the root of all evil.
"We had the benefit [...] of knowing what our neural networks look like, and what they'll look like in the future,"
Really? If they take their neural network development seriously I don't think they know what their networks will look like in ten years. It's a research area in the middle of a transformation. Using architectures molded into hardware is probably just costly and will act as an antagonist to innovation. I don't think having 2000 vs 200 frames per second right now outweighs that downside.
allowing for 16 total antennas and ensuring that no matter how you hold your phone, the signal won't get blocked.
"Four antennas ought to be enough for anybody." -- Steve Jobs
Well, stop, right fucking now.
Yeah, can't stand these fucking exclamation marks.
If you would be sure about the 90-95% efficiently then there would be no need to be "leaning towards" the infeasibility of doubling that efficiency, isn't it?
I hope 1M is enough to fix that mess. Successfully switching between user sessions is a matter of luck; Useful panel functionality is only provided through sometimes abandoned third party extensions; For anything other than the five-or-so options in the system settings you need to resort to something like a "tweak-tool"; Then there is the wildly unnecessary "swipe up screen" that get's in your way. All in all a very bad experience. I'm switching back to 16.04 for now.
I think what they actually mean is that the costs of hosting a website and a couple of pdfs should be provided for free either by public or private funds, just like providing, editing and reviewing the scientific content is done for free by the people who signed the open letter. I can't believe a fee of $30 per online article copy is in any sensible way proportional to the actual costs the publisher has to provide that article.
That's a lame argument. It implies a very narrow reading of "keeping people safe", a term that is wide open to interpretation. An opposing argument may equally cite keeping people safe as a reason for strong encryption.
Yeah, they wanted to call it iFixit but somehow that name was taken already.
The problem with these kinds of standards is that they usually settle on the cheapest, crappiest thing. MicroUSB is the perfect example, exploding/burning/shorting chargers are another one.
I fully disagree! My chargers have never exploded. I thank the EC on my knees everyday for imposing standardization of power supplies for phones.
It's not clear to me what problem he is trying to solve.
Dump truck bombs make you look inept to the other world powers.
In a way, you are saying that using Dump trucks makes you look like Trump. <ducks/>
if you have access to a system you can do whatever the fuck you want with it.
At least, that's what Trump himself seems to be doing...
Whether crypto-mining or not, some pages seem to use a disproportionate share of cpu time for the content they're delivering. Some form of cpu usage indicator per tab would be helpful, similar in vein to the speaker icon on tabs that produce sound.
Although to me this is clearly a case of A.I., parent has a point that not all types of problems lend themselves to be treated as reinforcement learning tasks, which seems to have been the key to success in this case.
PageRank, at its core, prioritizes popularity.
Yeah, but I think it's obvious that popularity of information in Google's ranking algorithm is used as a proxy for truthfulness/correctness (or in a wider sense: utility), and by and large that approach seems to work well. However if it turns out that in certain (predictable) circumstances popularity is not a good proxy, fixing that is an improvement.
If I query Google for information, I am not interested in falsehoods just because they are popular.
If you are afraid that Google won't serve sites telling you that earth is flat, Obama is a muslim, or the Las Vegas shooter was a f*ng n*r, you can add an extra term "fake" to your query, just like you have to add "xxx" if you want to see pussies.
I spent a day in 1998 getting my mouse to work on Red Hat 5.2...
The frequent stops of buses at predetermined locations may be an opportunity to recharge the battery at these locations (half a minute of charging every five minutes amounts to 6 minutes of charging per hour), perhaps eliminating the need to make a long stop for a full charge.
I like how you regard the rejection of net neutrality as profanity.
Why not put them slightly to the right of the center?
However, he added that HMS Queen Elizabeth is due to be given a computer refit within a decade.
What's the fuss about? In 2027 this warship will be up-to-date with bleeding edge Windows 10. Oh wait...
Is that really the market value of scientific knowledge? Who needs traditional facts anyway, now that we have alternative and renewable facts?