Well, if you videotape it and put it on the Internet, then it becomes legal... (really, the only different between prostitution and porn is that in porn, you publicize the sex-for-money -- and guess which one is legal?)
I think a flaw in all of this is: if essentially everyone is unemployed, they will have no money to buy the products that are automatically produced.
That's right. Because those automatically produced products will cost so much. I mean, they have to be expensive, in order to pay the robots' salaries...
Oh wait, you don't have to pay robots.
Ah, you say, but you have to pay for the electricity to run them! Exactly, how is that? Why does electricity cost money? Because people have to work to mine the coal to run the power plants. Well, instead of paying people to mine coal now we'll have to pay the robots to mine it...
Oh wait, you don't have to pay robots.
Well, won't we still have to pay for food? Oh wait, produced by robots. How about new clothes? Wait, made by robots. How about new cars? Produced by robots. The metal to MAKE those cars? Mined by robots. Crap, but how does the metal get from the mine to the factory? Oh, robotic trains.
What about education? Robots can't be teachers, so we'll have to pay the professors, right?
WHY exactly do we need to pay them? What are they going to use the money for? Everything is free, produced by robots.
You suggest that the few human workers that remain will become rich. I think that's ridiculous. Why would anyone ever be paid for anything ever again, since you don't NEED money, since the stuff is FREE because it's produced by robots who don't know about or care about MONEY?
Money is just a way to ensure that everyone actually does some of the work required to maintain human civilization. If all that basic maintainance is done by robots, what exactly is the point of money anymore?
How on earth could it possibly be a bad thing for robots to do everything for us, so we can all sit around drinking fruit punch?! Oh no, we won't be employed! We'll all be sitting on tropical islands getting waited on hand and foot by machines that never get tired and don't expect to be paid! Oh, the horrors!
Why is it people are comfortable with companies sponsoring Yoga, a religious practice?
Hmm, maybe because they're corporations, not the government, and there's no imperative for them to stay out of religion?
Besides, get real. To 99.9% of Americans, "Yoga" means "stretching exercises." Arguing about this is equivalent to arguing about the distinction between "hacker" and "cracker." The public at large doesn't care about your little distinctions.
Suppose you have 10^11 brain cells (common estimate). There are (10^11-1)*(10^11-2)/2 = 5*10^21 possible links between different neurons. A link is either there, or not there. Therefore, a link could correspond to a bit, therefore the capacity would be about 5 billion terabits. How many possible brain configurations are there? Well, there's 5 billion trillion bits, therefore there are 2^(5 billion trillion) possible brains -- a number somewhere near 2*10^1505149978.
But of course, that estimate is just as bogus as the 1Tbit estimate, since nobody really knows how the brain stores data. But it's fun to calculate huge numbers:-)
Everyone in the group shared that thought. He's saying "four out of five testers prefer X to Y".
If that's all you want to determine, why not just ask the listener to do that -- put the codecs in order from best to worst?
I didn't say you can't do statistics on subjective data. My point is, one person's "3" is different than another's "3", so taking the average of a bunch of relative rankings is meaningless. It isn't the subjective nature of the perception that is the flaw, it's the subjective nature of the ranking.
Here a suggestion: ask 1000 people to rank the codecs in relative order -- don't assign numbers to anything, just say which is better than which. Now, compute an "average" ranking using the following two rules:
1. If the number of people who judged X better than Y is greater than the number who judged Y is better than X, then place X above Y in the final list.
2. If X is ranked above Y the same number of times as Y is ranked above X, break the tie in such a way that rule 1 is satisfied for the remaining codecs.
Question: is the above logically possible? Or could you construct a sane set of individual rankings such that there is no final ordering that obeys both rules 1 and 2?
Ever wondered how essay tests get graded by teams of people?
Unfairly? In all cases? Subjective quantification is useless. Rank the blueness of the sky today.
you can spot just the sort of glitches that perplex you so, and with a little work, even correct for these biases.
To calculate the "true" ranking? There's no such thing as a true subjective judgement. That's sort of the definition of "subjective." If you had some way to find the "true" ranking, why even bother asking people? Just compute the true ranking!
If you want to do statistics on subjective judgments, you need to limit the number of possible options. Ask people binary questions -- then you can actually say something meaningful, like "82% of listeners heard clicking in this codec. Only 16% of listeners heard it in this other codec." That tells me something. Telling me that "the average clickiness of this codec is 2.8" tells me nothing.
Are you saying that you can't perform statistical analysis of subjective data?
Yep. Here's my 1-5 scale:
1: Base ranking
2: 1.668 times better than 1
3: 5.881291 times better than 2
4: 2.5877 times better than 3
5: 100000 times better than 4
Except for when they play that certain song I like, in which case my entire ranking system is subconsciously skewed. In fact, if they play Metallica, I will rank it 2 no matter how subjectively good it sounds. Because Metallica just sounds like shit.
And here's Johnny's scale:
1: Base ranking
2: 2.4 times better than 1
3: 1000 times better than 2
4: 16.8523 times better than 3
5: 1.0000001 times better than 4
Would you like to see Suzie's scale? Do you see why this is bullshit yet?
No, most statisticians would have never done something so ridiculous as compute standard deviations based on subjective rankings. The entire study became total bullshit the moment they tried to do any analysis of the results.
Had they just plotted a histogram of the rankings, this might have been a respectable attempt. But they just made themselves look like morons.
What they should have done is asked a series of binary questions: "Is there perceptable pre-ringing?" "Did you perceive any audio clicks?" "Would you say this is CD quality audio?" etc. Then use at least a thousand listeners. Then they might have been able to compute something meaningful.
They asked people to rank things on a subjective scale, and then they did statistics with the results? I'm sorry, but that's a scientific faux-pas. The moment they tried to compute a standard deviation, they became pseudoscientists. Sorry, the "hard" stats are worthless. The only even possibly meaningful thing would be a histogram of the rankings.
Just to make it clear why you can't do this: There is a certain incremental difference in subjective quality, that will cause listener A to rank something 4 instead of 3, for example. Will that same incremental difference in subjective quality case listener B to also rank something 4 instead of 3? We don't even know if the scale is linear, much less whether the different increments are even the same for different listeners.
Example: on the Kuro5hin blog site, if I really like a comment I will rate it 5. If I hate it, I will rate it 1. I never use the ratings in between. Thus, sometimes a 1 is bad, and sometimes a 1 is really bad. The same could happen here. Even if different listeners perceive similar subjective changes in quality, they might assign different changes in rank to those changes in quality.
God, this entire thing is bullshit, now that I think about it.
I think they might be assuming a 12-bit salt added to the hash. This would make the hash dictionary 4096 times larger, since they would have to compute the hash of each password 4096 times (for each possible 12-bit salt value).
If this is the case, it implies that Windows password hashes do not use salts. Now, I'm not claiming that salting makes the process secure (it doesn't), but it does make it orders of magnitude more intensive to compute a complete hash dictionary. At the expense of 12 bits per password (hell, use more if you want!) it seems worth it to use salts.
Just so long as they're kicked out from 8km or higher.
I assume that's to ensure that they die a slow, wheezing, oxygen-deprived death, instead of a painless sudden-impact death? (8 km is considered the beginning of the "death zone" for high altitude mountaineering.)
"Software IP" includes DSP firmware do decode Ogg vorbis and the CPU firmware for overall system control.
Basically, it's designed how it should be designed: seperate CPU and DSP cores, and both are independently programmable. It would be incredibly stupid to design a "pure" hardware solution (decoder in silico) since everyone admits that Vorbis is going to evolve and change, especially right now, during its "adolescent" period.
It's a de facto ban because no shipping company is going to be willing to pay for employee training and certification just to ship a small quantity of rocket motors. As a result nobody will be able to acquire the motors unless they live close enough to the factory to go over and pick them up themselves. Since there aren't many such people, the number of motors sold will plummet, and the industry will probably go out of business.
Whether intended or not this is going to destroy model rocketry.
For a moment I thought this article was about how new regulations threaten to ban the sport of model rocketry. It would be good for that issue to get a little more airplay...
That's the standard "academic" publication format, at least in the field of computer science. It looks just like any other university-produced research publication I've ever seen...
Look at the spectrum of light reflecting from the planet's atmosphere. Different isotopes of various gases in the atmosphere have slightly different spectra. By looking at the ratio of isotope abundances you can tell how long the radioactive isotopes have been decaying since the initial formation of the planet. Under certain assumptions about how stars produce new elements and the ratios in which they are produced, you can compute how long it has been since the planetary material was spewed out of a star -- a pretty good definition of the "birthday" of that planet.
Couldn't I get a slightly wider patent, namely a patent on an AI system that includes logic to safeguard against human injury? Given the current state of the patent office, could I possibly patent such a broad concept? "An algorithm which uses a series of computational instructions to give rise to internal states which are unconducive to human injury," or other similarly opaque language.
Then, in the future, when AI actually takes off, a law is passed requiring all AI devices to have this human safeguarding logic -- seems like a law people would probably want to pass. Since I got the patent on this way back before people understood the ramifications of allowing such a patent, I'm now the only person authorized to make AI systems.
No, but the latency is different. This is precisely the point I'm trying to make. It doesn't matter how long the trucks take to get there -- that's latency, not bandwidth. The bandwidth is how much data can pass a given point in a second. It has nothing to do with how long it ultimately takes to get there.
If the trucks were driving to the moon, which is 300000 miles away, the bandwidth would still be 40-something TB per second, but the latency would suck -- 208 days to make the journey at 60 MPH.
That figure is per tape, the actual shipment has 1,139 tapes, I think. 10.175GB * 1,139 = ~11.6TB. That *is* impressive bandwith.
Well in theory, that's not really "bandwidth," it's just a number of bytes. The bandwidth would be the maximum sustained throughput. Essentially, how much data could be delivered per second, if there were a constant stream of trucks pulling in, each carrying 11.6TB. Assume the trucks drive bumper-to-bumper, at 60 MPH. Assume each truck is what, 25' long. At 60 MPH it takes about 0.28 second to travel one truck-length. Therefore, the actual bandwidth is 11.8/0.28 = 42.1 TB per second.
How can the RIAA shut down webcasters who stream music the RIAA has no copyright over? Do they have the patent on music, or something? Or are they trying to get rights to webcast RIAA-corporation-owned music without paying the royalties? If the latter, I don't see how they can possibly expect to get that right, any more than I could sue Microsoft to let me host Windows XP ISOs on my website.
I hate to say it, but I gotta go with the RIAA on this one. If they want unencumbered rights to distribute music, they should get truly independent artists, i.e. artists that have never had anything to do with any RIAA company.
Slashdot Mirror
Well, if you videotape it and put it on the Internet, then it becomes legal... (really, the only different between prostitution and porn is that in porn, you publicize the sex-for-money -- and guess which one is legal?)
That's right. Because those automatically produced products will cost so much. I mean, they have to be expensive, in order to pay the robots' salaries...
Oh wait, you don't have to pay robots.
Ah, you say, but you have to pay for the electricity to run them! Exactly, how is that? Why does electricity cost money? Because people have to work to mine the coal to run the power plants. Well, instead of paying people to mine coal now we'll have to pay the robots to mine it...
Oh wait, you don't have to pay robots.
Well, won't we still have to pay for food? Oh wait, produced by robots. How about new clothes? Wait, made by robots. How about new cars? Produced by robots. The metal to MAKE those cars? Mined by robots. Crap, but how does the metal get from the mine to the factory? Oh, robotic trains.
What about education? Robots can't be teachers, so we'll have to pay the professors, right?
WHY exactly do we need to pay them? What are they going to use the money for? Everything is free, produced by robots.
You suggest that the few human workers that remain will become rich. I think that's ridiculous. Why would anyone ever be paid for anything ever again, since you don't NEED money, since the stuff is FREE because it's produced by robots who don't know about or care about MONEY?
Money is just a way to ensure that everyone actually does some of the work required to maintain human civilization. If all that basic maintainance is done by robots, what exactly is the point of money anymore?
You aren't seeing the bigger picture here.
How on earth could it possibly be a bad thing for robots to do everything for us, so we can all sit around drinking fruit punch?! Oh no, we won't be employed! We'll all be sitting on tropical islands getting waited on hand and foot by machines that never get tired and don't expect to be paid! Oh, the horrors!
Hmm, maybe because they're corporations, not the government, and there's no imperative for them to stay out of religion?
Besides, get real. To 99.9% of Americans, "Yoga" means "stretching exercises." Arguing about this is equivalent to arguing about the distinction between "hacker" and "cracker." The public at large doesn't care about your little distinctions.
Suppose you have 10^11 brain cells (common estimate). There are (10^11-1)*(10^11-2)/2 = 5*10^21 possible links between different neurons. A link is either there, or not there. Therefore, a link could correspond to a bit, therefore the capacity would be about 5 billion terabits. How many possible brain configurations are there? Well, there's 5 billion trillion bits, therefore there are 2^(5 billion trillion) possible brains -- a number somewhere near 2*10^1505149978.
But of course, that estimate is just as bogus as the 1Tbit estimate, since nobody really knows how the brain stores data. But it's fun to calculate huge numbers :-)
Oh, and packs of rabid nutria maraud the streets at night!
Don't come to Oregon!
If that's all you want to determine, why not just ask the listener to do that -- put the codecs in order from best to worst?
I didn't say you can't do statistics on subjective data. My point is, one person's "3" is different than another's "3", so taking the average of a bunch of relative rankings is meaningless. It isn't the subjective nature of the perception that is the flaw, it's the subjective nature of the ranking.
Here a suggestion: ask 1000 people to rank the codecs in relative order -- don't assign numbers to anything, just say which is better than which. Now, compute an "average" ranking using the following two rules:
1. If the number of people who judged X better than Y is greater than the number who judged Y is better than X, then place X above Y in the final list.
2. If X is ranked above Y the same number of times as Y is ranked above X, break the tie in such a way that rule 1 is satisfied for the remaining codecs.
Question: is the above logically possible? Or could you construct a sane set of individual rankings such that there is no final ordering that obeys both rules 1 and 2?
Unfairly? In all cases? Subjective quantification is useless. Rank the blueness of the sky today.
you can spot just the sort of glitches that perplex you so, and with a little work, even correct for these biases.
To calculate the "true" ranking? There's no such thing as a true subjective judgement. That's sort of the definition of "subjective." If you had some way to find the "true" ranking, why even bother asking people? Just compute the true ranking!
If you want to do statistics on subjective judgments, you need to limit the number of possible options. Ask people binary questions -- then you can actually say something meaningful, like "82% of listeners heard clicking in this codec. Only 16% of listeners heard it in this other codec." That tells me something. Telling me that "the average clickiness of this codec is 2.8" tells me nothing.
Yep. Here's my 1-5 scale:
1: Base ranking
2: 1.668 times better than 1
3: 5.881291 times better than 2
4: 2.5877 times better than 3
5: 100000 times better than 4
Except for when they play that certain song I like, in which case my entire ranking system is subconsciously skewed. In fact, if they play Metallica, I will rank it 2 no matter how subjectively good it sounds. Because Metallica just sounds like shit.
And here's Johnny's scale:
1: Base ranking
2: 2.4 times better than 1
3: 1000 times better than 2
4: 16.8523 times better than 3
5: 1.0000001 times better than 4
Would you like to see Suzie's scale? Do you see why this is bullshit yet?
Had they just plotted a histogram of the rankings, this might have been a respectable attempt. But they just made themselves look like morons.
What they should have done is asked a series of binary questions: "Is there perceptable pre-ringing?" "Did you perceive any audio clicks?" "Would you say this is CD quality audio?" etc. Then use at least a thousand listeners. Then they might have been able to compute something meaningful.
Just to make it clear why you can't do this: There is a certain incremental difference in subjective quality, that will cause listener A to rank something 4 instead of 3, for example. Will that same incremental difference in subjective quality case listener B to also rank something 4 instead of 3? We don't even know if the scale is linear, much less whether the different increments are even the same for different listeners.
Example: on the Kuro5hin blog site, if I really like a comment I will rate it 5. If I hate it, I will rate it 1. I never use the ratings in between. Thus, sometimes a 1 is bad, and sometimes a 1 is really bad. The same could happen here. Even if different listeners perceive similar subjective changes in quality, they might assign different changes in rank to those changes in quality.
God, this entire thing is bullshit, now that I think about it.
If this is the case, it implies that Windows password hashes do not use salts. Now, I'm not claiming that salting makes the process secure (it doesn't), but it does make it orders of magnitude more intensive to compute a complete hash dictionary. At the expense of 12 bits per password (hell, use more if you want!) it seems worth it to use salts.
I assume that's to ensure that they die a slow, wheezing, oxygen-deprived death, instead of a painless sudden-impact death? (8 km is considered the beginning of the "death zone" for high altitude mountaineering.)
"Software IP" includes DSP firmware do decode Ogg vorbis and the CPU firmware for overall system control.
Basically, it's designed how it should be designed: seperate CPU and DSP cores, and both are independently programmable. It would be incredibly stupid to design a "pure" hardware solution (decoder in silico) since everyone admits that Vorbis is going to evolve and change, especially right now, during its "adolescent" period.
Don't worry, they've done the right thing :-)
Uh. BitTorrent has many of the features you are describing. Sheesh, keep up with the times, man! :-)
Whether intended or not this is going to destroy model rocketry.
For a moment I thought this article was about how new regulations threaten to ban the sport of model rocketry. It would be good for that issue to get a little more airplay...
That's the standard "academic" publication format, at least in the field of computer science. It looks just like any other university-produced research publication I've ever seen...
Look at the spectrum of light reflecting from the planet's atmosphere. Different isotopes of various gases in the atmosphere have slightly different spectra. By looking at the ratio of isotope abundances you can tell how long the radioactive isotopes have been decaying since the initial formation of the planet. Under certain assumptions about how stars produce new elements and the ratios in which they are produced, you can compute how long it has been since the planetary material was spewed out of a star -- a pretty good definition of the "birthday" of that planet.
Then, in the future, when AI actually takes off, a law is passed requiring all AI devices to have this human safeguarding logic -- seems like a law people would probably want to pass. Since I got the patent on this way back before people understood the ramifications of allowing such a patent, I'm now the only person authorized to make AI systems.
Can this be possible?
If the trucks were driving to the moon, which is 300000 miles away, the bandwidth would still be 40-something TB per second, but the latency would suck -- 208 days to make the journey at 60 MPH.
Well, I was assuming the trucks would drive into one of about 50 million different parallel unloading docks, but you know, I was just estimating...
Well in theory, that's not really "bandwidth," it's just a number of bytes. The bandwidth would be the maximum sustained throughput. Essentially, how much data could be delivered per second, if there were a constant stream of trucks pulling in, each carrying 11.6TB. Assume the trucks drive bumper-to-bumper, at 60 MPH. Assume each truck is what, 25' long. At 60 MPH it takes about 0.28 second to travel one truck-length. Therefore, the actual bandwidth is 11.8/0.28 = 42.1 TB per second.
That's stupid. Why couldn't you just look for the blank spot on the map and assume that whatever is there must be important?
I hate to say it, but I gotta go with the RIAA on this one. If they want unencumbered rights to distribute music, they should get truly independent artists, i.e. artists that have never had anything to do with any RIAA company.