Re:Not a very good idea - easily breakable
on
Inkblot Passwords
·
· Score: 1
The point is, the inkblot test relies on the fact that most people with "normal" brain function will look at an inkblot the same way.
I worked at MSR in the same group Adam (the author of the inkblot password work) did last summer. Most of his work on the project was on exactly the issues you bring up. He spent time in the University of Washington library reading decades-old psych papers on inkblot statistics, and he spent weeks getting other MSR employees to generate passwords by naming inkblots.
The result? Turns out you're wrong. Different people do see a few different items in each blot. Even if they see the same general item they may describe it different ways. "It's an alligator with a bat head!" "No, it's a batgator!" (Actual reactions during Adam's talk last summer.) Because your password is based on what you see AND how you name it, each blot gives you several bits of entropy. String 'em all together and you get a password with several dozen bits of entropy, easily better than user-selected 5-to-8-character passwords.
In fact, Adam had a chart with the amount of entopy likely to be in different authentication methods. The inkblots hit a sweet spot of "decent entropy, but still reasonably easy to remember."
Finally, inkblots are resistant to dictionary attacks. Everybody sees different blots, so to guess a password, you need to grab the person's blots from the (protected) password/shadow file. To build an attack dictionary, you need to find out all of the things that people might reasonably see in a blot. Even if it's only a dozen possibilities per blot, you still need to get a psychologically diverse group of humans to describe the blots to you. Not exactly something that 'L0phtcrack' can do on its own.
An innovative, potential useful idea coming from Microsoft?
In this case, no. Adam was an intern at Microsoft last year. I believe he already had this idea in mind when he walked in the door. He did do his inkblot experiments at MS, and most of the victims were MSR employees. But he did not dream up the idea while an MS employee.
Incidentally, don't expect to see inkblots in any Microsoft product anytime soon. It took them a year just to write this article.
Oh please....consumer choice doesn't have anything to do with this. A North Carolina company may get shut down, costing 1,200 jobs, which is why there is soon going to be a law protecting it.
I don't think Static Control would get shut down if they lost to Lexmark. Cloned printer cartridges are not their first or (IIRC) largest business. They make static bags and shields (thus the name) and electronic test equipment. I don't know what inspired them to venture into the ink cartridge market.
I do not doubt that Static Control asked for this law, but I don't think those 1200 jobs or the company's existence were at stake.
NC's passage of the "right to refill" law is in keeping with its passage two years ago of a UCITA bomb-shelter law. Both aim to limit the effect of bad technology laws in other jurisdictions on NC customers.
MIDI's note range is from 0-127, with 60 being middle C. I believe that gives it a high end of the G 5.5 octaves above middle C, which would be about 12.5KHz. So MIDI isn't representing any inaudible pest-repelling tones. Not even with pitch bending.
Sometimes you get something a little more advanced, like the ability to do simple FM synthesis or generate white noise. I strongly suspect most cell phones use this sort of approach. I've never noticed one producing a sampled sound; it's always that tweedly, one-voice bastardization of Beethoven's 5th or something.
FM synthesis is still going to be limited by the DAC's frequency range and by the response curve of the DAC and the speaker. If expensive desktop sound cards tail off at 24KHz, do you really expect cell phones to faithfully reproduce 40KHz?
Desktop computers these days all use sampled sounds, but lots of simpler devices (Palm OS for example) use a much simpler method
Hey, not all PalmOS devices are limited to crappy FM synth. My Clie SJ-33 has a 16-channel wavetable synth and 44.1KHz sampled sound. It plays MP3s -- rock on. But I still wouldn't expect it to handle 40KHz tones.
What format are ringtones stored in that they can represent tones beyond the range of human hearing? Most audio formats top out around 22 or 24 KHz tones (44.1 or 48 KHz sampling). My last cell phone only allowed the notes available on a piano, none of which is beyond human hearing.
Even if the ringtone format can represent tones that high, can the cellphone speaker reproduce them? Again, many speakers are only rated to about 20 KHz, because that's all that's useful for human beings.
And finally, couldn't you just make a device for about $5 that would actually do this right and last a whole lot longer on a set of batteries? Cell phones are not the right way to make a constant 40KHz (say) tone.
I'm inclined to categorize this the same way I categorize stand-alone sonic pest-repelling devices: well-intentioned but useless. Incidentally, that's the category I put normal cell phones in as well.:)
It does not say that "The Program" means BOTH, but EITHER.
That's a strange reading. What the GPL actually means when it says "either the Program or any derivative work under copyright law" is that either one you choose to distribute -- the original or the derivative -- is still "the Program" under the license. It means you can distribute EITHER, and it will still have to be GPL.
Have you actually read the GPL?
Yes. Have you actually understood the GPL?
IANAL but I [snip]
But Eben Moglen, who wrote the GPL, is. This is the sort of language he's likely to get right, whether or not you personally understand it the same way a court would.
"Free" my butt. $5000 US is a LOT of money for a program, period, ESPECIALLY for one user.
You're wandering pretty far off-topic there. Just feel like ranting about free software today, or what? That $5000 is essentially a consulting fee. Once you have that custom CD, made just for your sorry butt, you can make as many copies or modifications as you please. Free software is about freedom -- to use, to copy, and to modify -- and not always about freeloading.
Of course, the source is, as always, on gnu.org. For free in both senses of the word.
What is Microsoft's true impression of Linux as both an OS and as a competitor?
They understand it, much better than you realize. We got our first glimpse of that with the Halloween memos. Now MS has an entire group devoted to fighting Linux. They have competitive strategy memos, presentations, and an eerily accurate understanding of why people develop and use Linux.
How clued in are the top-level people about the capabilities of Linux?
Very much so, I'm sure. How hard is it to get some underling -- an underling whose job it is to figure out how to compete with Linux anyway -- to download and install Red Hat Linux 9 and then tell you about it?
Will their strategy of ignoring it and spreading FUD change
Give them some credit here! The FUD is mostly gone: when's the last time you heard MS saying that "Linux is just a toy OS" or "There's no high-quality support available for Linux?" MS is competing now on features, on total cost of ownership, and (if illegally) on price. They are giving their software away cheaply or for free in colleges, grade schools, and underdeveloped markets. They're also (rightfully?) trying to quell any legislation that would mandate free software.
I can't help but say that most CS/IT majors need this. I've seen too many people write apps (simple ones even) that relied on that ethernet connection that the dorms give, 10Mbit between machines. "Scale down? Who has less than a fast cable modem these days?"
PlanetLab won't help much with that. Most of the PlanetLab nodes are pretty well connected, certainly better than modems. It lets you test latency pretty realistically, given that the nodes span the globe.
Modelnet might be a better bet for emulating modem-dominated networks.
I believe Bill Gates himself can get a free court-appointed attorney in criminal cases.
Nope. The Miranda statement is "If you cannot afford [an attorney], one will be appointed for you." You actually have to demonstrate your inability to afford an attorney, which means disclosing your finances to the court. At least, this is the case in North Carolina.
Public defenders are not known for their great track records. US jails (death row in particular) are packed with people who couldn't afford better lawyers.
Re:110VAC outlets available today
on
42-Volt Autos
·
· Score: 1
Check it out. Just $20 or so. Radioshack sells an adaptor but it's $99. Fucking rip-off artists. There are a couple of cars out today that have 110VAC outlets already.
The inverter you link to is only 100W and 2-prong. No wonder it's only $20. Spend the extra $5-$10 and get a Jazz-300 with two 3-prong outlets and 300-500W capacity.
Radioshack sells an adaptor but it's $99. Fucking rip-off artists. There are a couple of cars out today that have 110VAC outlets already.
It's a $100-or-so option on the Pontiac Vibe and Toyota Matrix. There's your rip-off artist. Just buy an inverter and install it yourself!
Unless they modified the source for your software, they don't have to distribute it.
Yes, they do. Section 3 of the GPL gives distributors three choices:
1) include the source code on whatever medium you're shipping the binary.
2) include a written offer with the binary letting people obtain the source for cost of media.
3) include whatever written offer the distributor got from his source. This third option is only available for non-commercial distribution, so it does not apply here.
In other words, every Linksys router should come with either all GPL source on CD, or a written offer telling me where I can obtain the GPL source from Linksys. They cannot point me at someone else's source distribution center, because Linksys is a commercial user.
The section you quoted simply clarifies the first possibility. If you're distributing binaries from a site, you can distribute source code separately on that same site. You don't have to force every user of the binary to download both.
How big a house do you need, exactly, to start needing three base stations? I have one, a LinkSys, and it covers my entire house, upstairs and down, my entire yard, and out into two different streets. My base station fears no microwave and no closed door.
Why spend $650 on three Apple Airports when $100 on a Linksys will suffice? If you need three base stations, get a smaller house!
Senator Russ Feingold, from Wisconsin. I believe he's on record opposing the DMCA, but now I can't find the interview. He's been in the news more recently opposing TIA and promoting civil liberties in the face of the new American "homeland security" culture.
If the talks are revolving around using this technology to distribute full resolution movies to digitally equipped movie theaters
That's a thought. I think of Disney and Microsoft as consumer companies, so I thought this was an "improving broadband" sort of proposal.
Still, though. We're talking about gigabit links with 100-200ms latencies. A connection for downloading movies is more likely to be 20-50mbit/s and 50ms latency, for which TCP FAST is definitely unnecessary. An entirely uncompressed 2-hour movie (1280x1024, 48bpp, 24fps) is 1.3TB. Downloading that over the course of a week (how many new movies do theaters get each week) requires only 17mbit/s. Why deploy a gbit link when 1/3 of a T3 will suffice?
That said, if you have that much data to transfer, it's much-much-much cheaper to do it by FedExing someone six hard disks.* Movie studios are already balking at the notion of helping to pay for digital projectors. Can you imagine how they'd react if asked to buy each theater a T-3 (let alone an OC-48)?
So I don't think that even uncompressed movie distribution would require this sort of wizardry.
* credit to Andy Tanenbaum: "Never underestimate the bandwidth of a station wagon full of tapes [disks are denser these days] hurtling down the highway."
Still, those numbers don't look right. AFAIK TCP has 5-15% overhead, so they must have been using a high-bandwidth, really-high-latency line to get that much improvement. Really high.
Exactly right. TCP's overhead is normally 5-15% because of its sawtooth "send a little too much, then back way off" nature. When the bandwidth-latency product is enormous, though, there is an absolute cap on data rate. They used a gigabit link from Silicon Valley (Sunnyvale, CA) to CERN (in Switzerland). A GB link spanning 8000 miles is... unusual.
Normal TCP gets close enough to 100% throughput. Their "6000x improvement over broadband" puffery is an apples-to-oranges comparison. They improved the performance of intercontinental gigabit TCP flows by a factor of not-quite-four, but the same improvements applied to my broadband connection would do essentially nothing. Thus, the talks with Disney are stupid.
Note: this won't even improve the performance of normal gigabit backbone links significantly. It only improves the performance if a) it's many hundreds of megabits or more, b) the latency is huge, like from the US to Europe, and c) it's just one TCP flow. A normal backbone carries many TCP flows in aggregate, where standard TCP Reno or TCP Vegas will probably perform as efficiently, and more predictably, than this TCP FAST thing.
I haven't really paid that much attention to the Ximian Connector. Does it require Outlook Web Access running?
Yes. It connects via webdav, which is the same way that OWA connects. It doesn't connect with the native protocol that Outlook itself does.
I use Linux at work and connect to our OWA without having the Connector, so I don't get what this statement is about.
Connector requires that OWA be enabled, but OWA doesn't require Ximian Connector! Ximian Connector just makes all of the OWA features available in Evolution instead of in an ugly web interface.
A watt is one joule per second. It is already expressed as per-unit-time, so there's no need to say "18.2GW/year." It's just 18.2GW.
or about 36 watts per person
Eh? China has somewhat over one billion people. 18.2 billion watts / 1.2 billion people = 15.2 watts per person.
By comparison, an American household uses around 1000 watts. So this dam alone could provide power for 18 million American homes -- nearly 20% of American residential power consumption.
It may well become a cesspool the size of lake superior.
The US had, at one point, a cesspool the size of Lake Erie. It was... Lake Erie. It's quite a bit better now.
Re:Journaling File System: for those who don't kno
on
Looking at Longhorn
·
· Score: 1
First of all, the file system is not consistent after a crash: journaling file systems need to replay the journal in order to make it consistent. This is no different in principle from non-journaling file systems (which, traditionally, also have incorporated various features to permit recovery), it just happens to be faster.
This absolutely is different in principle from non-journaled filesystems. A journaled filesystem can just wind backward to the last consistent state. A non-journaled filesystem may have no consistent state and may have to recreate one by fsck'ing the whole disk. An instructive example: when you delete a file, it removed the file's directory entry and puts the files blocks back on the free list. If your computer crashes during this time, one of those operations may happen without the other: your file exists, but its blocks have been marked for reuse, or the file is gone but the blocks still can't be reused. Only an fsck-walk of every file can discover this, which is what takes so damn long when booting non-journaled filesystems after a crash. With a journal, either neither or both of the operations would have committed, leaving you in a consistent state very quickly.
Put another way: a non-journaled filesystem must be checked after a crash. A journaled filesystem only needs to have any partial operations played or rewound; everything not mentioned in the journal is guaranteed to be consistent. You can still, of course, lose any dirty buffers in the block cache.
They are prohibiting Dini from blacklisting tudents who are in fact qualified, as a result of their beliefs, irrelevant to their qualification.
Do you even know what blacklisting is? Dini is being asked to make a positive statement that a given student is qualified. He wishes to make no statement at all. To qualify as "blacklisting" a candidate, he would have to take active steps to prohibit the student's advancement.
Note that Dini is not on an admissions committee, where he actually could exercise some sort of veto or blacklist. All he's doing is refusing to recommend candidates he doesn't think are qualified.
Dini is a modern Torquemada.
No, he's not. Torquemada could actively punish the people he disagreed with. The worst Dini can do is say, "I will make no recommendation." The students can always get a recommendation from another, less stringent professor.
Those of you who, perhaps rightly, point out that a belief in Divine Creation and a medical degree need not be mutually exclusive are still missing the point.
The point is this: students are asking Prof Dini to make a recommendation. A recommendation is a personal statement that Dr. Dini believes the student in question is qualified for... something. If he doesn't believe the student is qualified, he shouldn't write the recommendation -- regardless of his reason.
To put it another way, the Department of Justice is essentially compelling Dr. Dini to write recommendations for students that Dr. Dini does not feel are qualified. Our Benevolent Government is making someone misstate his own personal opinions. The fact that it's a recommendation for medical school and the fact that Dr. Dini teaches at a public university are red herrings; We The People are still asking him to lie about his evaluation of a student's qualifications.
I already own a 48GX and 49G, so I had no moral delimmas with installing those rom images.
HP supports the 49 and 48G/GX emulation efforts, and has released the ROMs freely for non-commercial use. AFAIK, you're welcome to use them even if you don't already own a real 48 or 49.
MPEG4 provides the same quality as DVDs (MPEG2)in a tiny fraction of the space.
Oh, absolutely not. MPEG4 is marginally acceptable at a rate of about 400 MB/hr. Certainly better than MPEG2 would be at the same bitrate. But there is a clearly noticeable difference. MPEG4 files show banding, blocking, and smearing artefacts that just aren't present on a DVD.
VP3 provides better quality than MPEG4, and (like Vorbis) is completely free of patents
It's highly unlikely that VP3, or any other relatively modern video codec, is free of patents. It's just a matter of which patents companies choose to enforce now and which patents they hide so that they can submarine you when your format becomes popular.
Slashdot Mirror
I worked at MSR in the same group Adam (the author of the inkblot password work) did last summer. Most of his work on the project was on exactly the issues you bring up. He spent time in the University of Washington library reading decades-old psych papers on inkblot statistics, and he spent weeks getting other MSR employees to generate passwords by naming inkblots.
The result? Turns out you're wrong. Different people do see a few different items in each blot. Even if they see the same general item they may describe it different ways. "It's an alligator with a bat head!" "No, it's a batgator!" (Actual reactions during Adam's talk last summer.) Because your password is based on what you see AND how you name it, each blot gives you several bits of entropy. String 'em all together and you get a password with several dozen bits of entropy, easily better than user-selected 5-to-8-character passwords.
In fact, Adam had a chart with the amount of entopy likely to be in different authentication methods. The inkblots hit a sweet spot of "decent entropy, but still reasonably easy to remember."
Finally, inkblots are resistant to dictionary attacks. Everybody sees different blots, so to guess a password, you need to grab the person's blots from the (protected) password/shadow file. To build an attack dictionary, you need to find out all of the things that people might reasonably see in a blot. Even if it's only a dozen possibilities per blot, you still need to get a psychologically diverse group of humans to describe the blots to you. Not exactly something that 'L0phtcrack' can do on its own.
In this case, no. Adam was an intern at Microsoft last year. I believe he already had this idea in mind when he walked in the door. He did do his inkblot experiments at MS, and most of the victims were MSR employees. But he did not dream up the idea while an MS employee.
Incidentally, don't expect to see inkblots in any Microsoft product anytime soon. It took them a year just to write this article.
I don't think Static Control would get shut down if they lost to Lexmark. Cloned printer cartridges are not their first or (IIRC) largest business. They make static bags and shields (thus the name) and electronic test equipment. I don't know what inspired them to venture into the ink cartridge market.
I do not doubt that Static Control asked for this law, but I don't think those 1200 jobs or the company's existence were at stake.
NC's passage of the "right to refill" law is in keeping with its passage two years ago of a UCITA bomb-shelter law. Both aim to limit the effect of bad technology laws in other jurisdictions on NC customers.
MIDI's note range is from 0-127, with 60 being middle C. I believe that gives it a high end of the G 5.5 octaves above middle C, which would be about 12.5KHz. So MIDI isn't representing any inaudible pest-repelling tones. Not even with pitch bending.
FM synthesis is still going to be limited by the DAC's frequency range and by the response curve of the DAC and the speaker. If expensive desktop sound cards tail off at 24KHz, do you really expect cell phones to faithfully reproduce 40KHz?
Desktop computers these days all use sampled sounds, but lots of simpler devices (Palm OS for example) use a much simpler method
Hey, not all PalmOS devices are limited to crappy FM synth. My Clie SJ-33 has a 16-channel wavetable synth and 44.1KHz sampled sound. It plays MP3s -- rock on. But I still wouldn't expect it to handle 40KHz tones.
Even if the ringtone format can represent tones that high, can the cellphone speaker reproduce them? Again, many speakers are only rated to about 20 KHz, because that's all that's useful for human beings.
And finally, couldn't you just make a device for about $5 that would actually do this right and last a whole lot longer on a set of batteries? Cell phones are not the right way to make a constant 40KHz (say) tone.
I'm inclined to categorize this the same way I categorize stand-alone sonic pest-repelling devices: well-intentioned but useless. Incidentally, that's the category I put normal cell phones in as well. :)
That's a strange reading. What the GPL actually means when it says "either the Program or any derivative work under copyright law" is that either one you choose to distribute -- the original or the derivative -- is still "the Program" under the license. It means you can distribute EITHER, and it will still have to be GPL.
Have you actually read the GPL?
Yes. Have you actually understood the GPL?
IANAL but I [snip]
But Eben Moglen, who wrote the GPL, is. This is the sort of language he's likely to get right, whether or not you personally understand it the same way a court would.
"Free" my butt. $5000 US is a LOT of money for a program, period, ESPECIALLY for one user.
You're wandering pretty far off-topic there. Just feel like ranting about free software today, or what? That $5000 is essentially a consulting fee. Once you have that custom CD, made just for your sorry butt, you can make as many copies or modifications as you please. Free software is about freedom -- to use, to copy, and to modify -- and not always about freeloading.
Of course, the source is, as always, on gnu.org. For free in both senses of the word.
They understand it, much better than you realize. We got our first glimpse of that with the Halloween memos. Now MS has an entire group devoted to fighting Linux. They have competitive strategy memos, presentations, and an eerily accurate understanding of why people develop and use Linux.
How clued in are the top-level people about the capabilities of Linux?
Very much so, I'm sure. How hard is it to get some underling -- an underling whose job it is to figure out how to compete with Linux anyway -- to download and install Red Hat Linux 9 and then tell you about it?
Will their strategy of ignoring it and spreading FUD change
Give them some credit here! The FUD is mostly gone: when's the last time you heard MS saying that "Linux is just a toy OS" or "There's no high-quality support available for Linux?" MS is competing now on features, on total cost of ownership, and (if illegally) on price. They are giving their software away cheaply or for free in colleges, grade schools, and underdeveloped markets. They're also (rightfully?) trying to quell any legislation that would mandate free software.
PlanetLab won't help much with that. Most of the PlanetLab nodes are pretty well connected, certainly better than modems. It lets you test latency pretty realistically, given that the nodes span the globe.
Modelnet might be a better bet for emulating modem-dominated networks.
Nope. The Miranda statement is "If you cannot afford [an attorney], one will be appointed for you." You actually have to demonstrate your inability to afford an attorney, which means disclosing your finances to the court. At least, this is the case in North Carolina.
Public defenders are not known for their great track records. US jails (death row in particular) are packed with people who couldn't afford better lawyers.
The inverter you link to is only 100W and 2-prong. No wonder it's only $20. Spend the extra $5-$10 and get a Jazz-300 with two 3-prong outlets and 300-500W capacity.
Radioshack sells an adaptor but it's $99. Fucking rip-off artists. There are a couple of cars out today that have 110VAC outlets already.
It's a $100-or-so option on the Pontiac Vibe and Toyota Matrix. There's your rip-off artist. Just buy an inverter and install it yourself!
Yes, they do. Section 3 of the GPL gives distributors three choices:
1) include the source code on whatever medium you're shipping the binary.
2) include a written offer with the binary letting people obtain the source for cost of media.
3) include whatever written offer the distributor got from his source. This third option is only available for non-commercial distribution, so it does not apply here.
In other words, every Linksys router should come with either all GPL source on CD, or a written offer telling me where I can obtain the GPL source from Linksys. They cannot point me at someone else's source distribution center, because Linksys is a commercial user.
The section you quoted simply clarifies the first possibility. If you're distributing binaries from a site, you can distribute source code separately on that same site. You don't have to force every user of the binary to download both.
And neither do you, Mister Mid-Four Digits.
Why spend $650 on three Apple Airports when $100 on a Linksys will suffice? If you need three base stations, get a smaller house!
Senator Russ Feingold, from Wisconsin. I believe he's on record opposing the DMCA, but now I can't find the interview. He's been in the news more recently opposing TIA and promoting civil liberties in the face of the new American "homeland security" culture.
That's a thought. I think of Disney and Microsoft as consumer companies, so I thought this was an "improving broadband" sort of proposal.
Still, though. We're talking about gigabit links with 100-200ms latencies. A connection for downloading movies is more likely to be 20-50mbit/s and 50ms latency, for which TCP FAST is definitely unnecessary. An entirely uncompressed 2-hour movie (1280x1024, 48bpp, 24fps) is 1.3TB. Downloading that over the course of a week (how many new movies do theaters get each week) requires only 17mbit/s. Why deploy a gbit link when 1/3 of a T3 will suffice?
That said, if you have that much data to transfer, it's much-much-much cheaper to do it by FedExing someone six hard disks.* Movie studios are already balking at the notion of helping to pay for digital projectors. Can you imagine how they'd react if asked to buy each theater a T-3 (let alone an OC-48)?
So I don't think that even uncompressed movie distribution would require this sort of wizardry.
* credit to Andy Tanenbaum: "Never underestimate the bandwidth of a station wagon full of tapes [disks are denser these days] hurtling down the highway."
Exactly right. TCP's overhead is normally 5-15% because of its sawtooth "send a little too much, then back way off" nature. When the bandwidth-latency product is enormous, though, there is an absolute cap on data rate. They used a gigabit link from Silicon Valley (Sunnyvale, CA) to CERN (in Switzerland). A GB link spanning 8000 miles is... unusual.
Normal TCP gets close enough to 100% throughput. Their "6000x improvement over broadband" puffery is an apples-to-oranges comparison. They improved the performance of intercontinental gigabit TCP flows by a factor of not-quite-four, but the same improvements applied to my broadband connection would do essentially nothing. Thus, the talks with Disney are stupid.
Note: this won't even improve the performance of normal gigabit backbone links significantly. It only improves the performance if a) it's many hundreds of megabits or more, b) the latency is huge, like from the US to Europe, and c) it's just one TCP flow. A normal backbone carries many TCP flows in aggregate, where standard TCP Reno or TCP Vegas will probably perform as efficiently, and more predictably, than this TCP FAST thing.
Yes. It connects via webdav, which is the same way that OWA connects. It doesn't connect with the native protocol that Outlook itself does.
I use Linux at work and connect to our OWA without having the Connector, so I don't get what this statement is about.
Connector requires that OWA be enabled, but OWA doesn't require Ximian Connector! Ximian Connector just makes all of the OWA features available in Evolution instead of in an ugly web interface.
A watt is one joule per second. It is already expressed as per-unit-time, so there's no need to say "18.2GW/year." It's just 18.2GW.
or about 36 watts per person
Eh? China has somewhat over one billion people. 18.2 billion watts / 1.2 billion people = 15.2 watts per person.
By comparison, an American household uses around 1000 watts. So this dam alone could provide power for 18 million American homes -- nearly 20% of American residential power consumption.
It may well become a cesspool the size of lake superior.
The US had, at one point, a cesspool the size of Lake Erie. It was... Lake Erie. It's quite a bit better now.
This absolutely is different in principle from non-journaled filesystems. A journaled filesystem can just wind backward to the last consistent state. A non-journaled filesystem may have no consistent state and may have to recreate one by fsck'ing the whole disk. An instructive example: when you delete a file, it removed the file's directory entry and puts the files blocks back on the free list. If your computer crashes during this time, one of those operations may happen without the other: your file exists, but its blocks have been marked for reuse, or the file is gone but the blocks still can't be reused. Only an fsck-walk of every file can discover this, which is what takes so damn long when booting non-journaled filesystems after a crash. With a journal, either neither or both of the operations would have committed, leaving you in a consistent state very quickly.
Put another way: a non-journaled filesystem must be checked after a crash. A journaled filesystem only needs to have any partial operations played or rewound; everything not mentioned in the journal is guaranteed to be consistent. You can still, of course, lose any dirty buffers in the block cache.
Do you even know what blacklisting is? Dini is being asked to make a positive statement that a given student is qualified. He wishes to make no statement at all. To qualify as "blacklisting" a candidate, he would have to take active steps to prohibit the student's advancement.
Note that Dini is not on an admissions committee, where he actually could exercise some sort of veto or blacklist. All he's doing is refusing to recommend candidates he doesn't think are qualified.
Dini is a modern Torquemada.
No, he's not. Torquemada could actively punish the people he disagreed with. The worst Dini can do is say, "I will make no recommendation." The students can always get a recommendation from another, less stringent professor.
The point is this: students are asking Prof Dini to make a recommendation. A recommendation is a personal statement that Dr. Dini believes the student in question is qualified for... something. If he doesn't believe the student is qualified, he shouldn't write the recommendation -- regardless of his reason.
To put it another way, the Department of Justice is essentially compelling Dr. Dini to write recommendations for students that Dr. Dini does not feel are qualified. Our Benevolent Government is making someone misstate his own personal opinions. The fact that it's a recommendation for medical school and the fact that Dr. Dini teaches at a public university are red herrings; We The People are still asking him to lie about his evaluation of a student's qualifications.
Compelled speech isn't free speech.
HP supports the 49 and 48G/GX emulation efforts, and has released the ROMs freely for non-commercial use. AFAIK, you're welcome to use them even if you don't already own a real 48 or 49.
Oh, absolutely not. MPEG4 is marginally acceptable at a rate of about 400 MB/hr. Certainly better than MPEG2 would be at the same bitrate. But there is a clearly noticeable difference. MPEG4 files show banding, blocking, and smearing artefacts that just aren't present on a DVD.
VP3 provides better quality than MPEG4, and (like Vorbis) is completely free of patents
It's highly unlikely that VP3, or any other relatively modern video codec, is free of patents. It's just a matter of which patents companies choose to enforce now and which patents they hide so that they can submarine you when your format becomes popular.
After the fourth posting of this same story in 24 hours, you'd hope he could get the spelling and grammar right. You'd hope.