"Even if making something elsewhere isn't a problem, sometimes there are laws or rules defining what you can or can't call a product based on where it was made, among other things. Bourbon whiskey for example."
So what? We were talking here about a *blind* test. You can't legally call a wine "Sherry fino" unless it's made using a specific method (soleras and criaderas under "flor"), using a specific grape variety (Palomino) *and* it comes from a protected designation of origin (the Sherry Triangle in Cadiz, Spain).
So, you see, one of the factors in a Sherry "fino" is purely administrative and impossible to stablish from a blind test: you can, at least theoretically, find a place with proper weather and soil type, use Palomino and the same specific method. The summelier should be able to detect that and tell if it's a "fino" instead of, say, an "amontillado" or a "palo cortado" or even a "manzanilla", but still there's nothing that would make tell him for sure that wine is, in fact, Sherry "fino" coming from Xerez: again, grapes come without a passport.
"Not much good in blind tests if there is no repeatability. Kinda like some tests of psychic powers out there, or homeopathy."
Not the same league, not even the same sports.
The problem is not that there are good wines and bad wines; the problem is not that sommeliers can't segregate good wines from bad wines both repeatably and in accordance to each other; the problem is not even that they can't segregate one wine from another. The problem is that they try to go too far: they take some a dozen of really excellent wines, which all of them already accord to be really excellent wines, which all of them already would be able to segregate from other "standard, out-of-the-shelf" wines, and then they try to strictly order them for their quality. No wonder each sommelier renders his own order, since at those already high levels all that rests is subjectivity, and no wonder even the same sommelier wouldn't be able to exactly repeat his own ordering since his separation abilities and taste memory is not so acute (that being said, there are other sommeliers contests where the taste memory is tested -recognizing wines by brand and year, and some people really get amazing results at that).
It is said that a normal human eye can distinguish roughly 16 levels of grey: imagine a team of "eye-sommeliers" trying to order a list of 256 levels of gray. Now, imagine them trying to order just 16 of them at the white side. The results would be like those of the tasting contest. Does that mean that there's no difference between black and white? Does that mean they can reliable and in accordance distinguish between blacker and whiter? Or does it mean that they just tried to go too far?
"You would think that if they were so good at it (the french judges) they could at least tell the difference between American and French grapes"
How they would? Grapes come with no passport. What they certainly can and will do is discerning what kind of grape the wine is made of.
Regarding the tasting itself is worth noting that there were almost no consensus among judges and that they would admit that the same tasting over the same wines from the same judges the next day would probably throw different results.
It is not that they can't distinguish between good and bad wines with a general accordance among all of them about what makes a wine good or bad but the stupid tendency to put everything into strictly ordered tabulated forms: it's not enough to say that Château Mouton-Rothschild '70 and Stag's Leap Wine Cellars '73 are both superb wines, you *have* to tell which one is better. Well, human perceptions don't work that way.
"tired of zombies, pirates, ninjas, and robots. Jesus"
Jesus won't help you here for He is Master or Zombies: He turned Lazarus into a zombie and He Himself was a zombie too (while it took him three days for the conversion).
"One big difference between American passenger trains (a-la-Acela) and the HSR trains in Madrid."
Those at Madrid were short-range trains for commuting, not high speed ones.
"For better or worse, American trains are basically rolling bank vaults legally required to be capable of surviving a head-on collision with a mile-long freight train."
That would change. On one hand you can't make those beasts to gain high speed, they need to be lighter; on the other is not needed: railways for high speed are exclusive for that service so you can't crash heads on with a freight train.
With regards of being targets, I'd say that the strongest value from plains is twofold: civil flights have the well-earned reputation of being the safest transport so taking one down has a stronger psycological effect; and then, a plane can be kind of a ballistic missile you can target against anything you want so nobody is safe, again a stronger psycological effect. So being the "terror factor" of a train more or less that of any other target (say, a sports center or a mcdonalds) its "cheaper" going for the low hanging fruit. Of course, terrorism is terrorism, so an attack to a train cannot be discounted as it can't be discounted for anything, and of course too, there's nothing preventing USA over-reacting on trains as it has done with plains. At least in the case of trains USA over-reacting won't have a global impact as it has been the case with planes (in Spain, even after the Madrid attack, high speed train security isn't too cumbersome for passengers).
"Show me ANY Linux where I can take a mix of totally random hardware thrown together and hand my 67 year old clueless dad the disc and have him install it PERFECTLY, without a SINGLE fuckup or hardware issue, and then we'll talk."
Show me FIRST the Windows where I can take a mix of totally random hardware thrown together and hand my 67 year old clueless dad the disc and have him install it PERFECTLY, without a SINGLE fuckup or hardware issue.
"If Oracle for whatever reason decides to stop investing in BTRFS, the likely outcome AFAICS is not that BTRFS dies, but rather that Chris Mason and his team jump shop to Red Hat, Novell, Google, IBM or some other Linux contributor"
Are you sure their contracts allow for even thinking to do that?
"Not to mention that, while ZFS may not become a universal file system, it could well dominate in NAS appliances, and other proprietary closed-box products running OpenSolaris."
I don't think so.
Not because any technical problem with ZFS but because that other beast called "patents".
And this is not an opinion but a fact: Coraid already tried to build up a NAS device based on ZFS but NetApp made them think it twice by patent violation threatening.
"They paper over the fact that if this guy had been hit by a bus, his employer, the City of San Francisco, would well and truly have been up a creek without a paddle."
Which is a management issue, not a technical one, so the one to blame must be a manager. Was Childs in a manager-level position or in a "mere" technical one?
"However harsh the sentence may have been, the fact is that Childs was a shitty IT manager."
Truly so. But was he in a managerial position to start with? All I can find about him is that he was a "network administrator", a "network engineer" or an "IT administrator", never a manager, so he was not the one to say how the passwords should have to be managed nor the one to deal with policy violations. In fact, as per this reference (http://blogs.sfweekly.com/thesnitch/2010/08/terry_childs_sentenced_hacker.php) it seems clear that upper SF management agree this being a case of bad management: both Terry's direct manager and the security manager were displaced (they are not fired -yet, probably not to ashame that very SF upper management).
"You seriously think Microsoft will embrace an LGPL browser engine?"
Why not? They are against GPL (being communist and all that song) but they are quite in favour of the BSD (as long as they can take advantage of others' code, not that they distribute so much under the BSD themselves) and LGPL is second to the BSD in that they can take others' code to their advantage without giving back so much.
"On the other side the german economy has the biggest growth in europe with over 2%, looks we do get something right.
I would argue that copying the "American school" of economics and printing money to stimulate growth isn't really growth at all."
You know that Germans are within the eurozone so they don't print their money at all, do you?
"If you have something more complex than a flat file, then use relational databases. Even Access databases are better than a collection of text files."
Ever tried to access from Access (pun intended) big blob fields?
The guy has a lot of data. Therefore he needs a data-base. He already has one with a hierarchical storage (the filesystem) that probably conveys the way data is generated (hierarchically, if only by date).
Then he needs to access the data by different criteria. Are they relational? Then a relational data-base is probably what's needed. Access is still hierarchical, tree-based...?
My point is that (quite of obviously) data storage and data mining/retrival are different beasts so they can (and probably should) be managed in different ways/by different tools.
One first thing to do is "freezing" the data storage deployment. He says he has problems because dangling symlinks. That wouldn't happen if he didn't find the need to move the "original" data around. OK: don't do that and the problem will go away. Just let the "original" data go into a plain or almost plain structure: as long as there're no name collisions anything could do, from a single directory to a somehow optimized tree structure (like a first level directory list ordered by date or alphabetically or anything that fits and adding subdirs as needed to the point that any subdirectory only holds files on the hundreds or lower thousands).
The only important thing to remember is that once a file gets stored it never moves from its place (at least logically: with time maybe newer data can go to faster filesystems while older/less used data can go to second tier storage, etc.).
On top of that you add searching/datamining tools. Since your data storage is fixed by convention you can add as many and as unrelated searching tools as you see fit. It can be a forest of symlinks for different ordering criteria, it can be a RDBM, it can be a search engine, it can be accesable only by means of command line tools, a web interface... whatever and you can use all or part of them as need arises.
A practical example: Linux distributions' sources of packages like apt or yum. Packages are stored on an (almost) flat alphabetically ordered filesystem and then an in-parallel structure handles access (ordered by architecture, usage subset, etc.) by different means.
"I've seen entire departments laid off where they spends on the upwards of hundreds of thousands of dollars to train the department and still came to conclusion it was cheaper to fire them all and retrain people overseas."
That's why I carefully avoided talking about any company but I said "can any *country*...?" Maybe your company's stakeholders were glad with the quarterly benefits but will your country be a better place fifty years from now which such practices?
"With no intent to fault him it was me who did the engineering and design he [an Engineering PhD Professor] just took the credit."
Maybe that was the case or maybe the PhD professor was able to validate your design, even accepting he didn't have the brains to come with it. Who knows.
"Class rooms and labs are nice tools for learning, but the fact of the new economics, that people will settle for less if it costs less."
Maybe. But then you'll end up with scholars of lessen value then. *Specially* at the face of the new economics can any country allow for worse knowledge workers? I don't think so.
"History, social sciences, psychology, most of philosophy and a lot of math subjects can be taught entirely over the web."
Only if you can socialize enterily over the web.
Even disregarding labs, the best point of dictated education is that it allows for "no holes" knowledge: at your own pace and growing from ignorance you don't even know what you ignore so probably you will end up "mastering" a narrow subset of an issue and lacking lots of basic knowledge. On the other hand, as an average and due to our psycological constrains it's much more difficult to find the will to go through the hard parts without day-to-day physical interaction. Not undoable but much more difficult. I.e.: in Spain there's an official "university-at-a-distance" and has been for years. Usually students take more years to gain their degrees, the abandon rate is higher and the percieved value for such degrees is lower than those from "brick-and-mortar" universities.
But this works both ways: the student can and do drive the professor too: an after the lecture chatting will make your lecturer to modify a bit future lectures because of the feedback he recieves from students, or will give you specific assignments at the sight of your interests and capabilities, and surely a face to face chat with a lecturer will easierly give you the motivation to gain or stregthen effort on the issue.
Slashdot Mirror
"Even if making something elsewhere isn't a problem, sometimes there are laws or rules defining what you can or can't call a product based on where it was made, among other things. Bourbon whiskey for example."
So what? We were talking here about a *blind* test. You can't legally call a wine "Sherry fino" unless it's made using a specific method (soleras and criaderas under "flor"), using a specific grape variety (Palomino) *and* it comes from a protected designation of origin (the Sherry Triangle in Cadiz, Spain).
So, you see, one of the factors in a Sherry "fino" is purely administrative and impossible to stablish from a blind test: you can, at least theoretically, find a place with proper weather and soil type, use Palomino and the same specific method. The summelier should be able to detect that and tell if it's a "fino" instead of, say, an "amontillado" or a "palo cortado" or even a "manzanilla", but still there's nothing that would make tell him for sure that wine is, in fact, Sherry "fino" coming from Xerez: again, grapes come without a passport.
"Not much good in blind tests if there is no repeatability.
Kinda like some tests of psychic powers out there, or homeopathy."
Not the same league, not even the same sports.
The problem is not that there are good wines and bad wines; the problem is not that sommeliers can't segregate good wines from bad wines both repeatably and in accordance to each other; the problem is not even that they can't segregate one wine from another. The problem is that they try to go too far: they take some a dozen of really excellent wines, which all of them already accord to be really excellent wines, which all of them already would be able to segregate from other "standard, out-of-the-shelf" wines, and then they try to strictly order them for their quality. No wonder each sommelier renders his own order, since at those already high levels all that rests is subjectivity, and no wonder even the same sommelier wouldn't be able to exactly repeat his own ordering since his separation abilities and taste memory is not so acute (that being said, there are other sommeliers contests where the taste memory is tested -recognizing wines by brand and year, and some people really get amazing results at that).
It is said that a normal human eye can distinguish roughly 16 levels of grey: imagine a team of "eye-sommeliers" trying to order a list of 256 levels of gray. Now, imagine them trying to order just 16 of them at the white side. The results would be like those of the tasting contest. Does that mean that there's no difference between black and white? Does that mean they can reliable and in accordance distinguish between blacker and whiter? Or does it mean that they just tried to go too far?
"You would think that if they were so good at it (the french judges) they could at least tell the difference between American and French grapes"
How they would? Grapes come with no passport. What they certainly can and will do is discerning what kind of grape the wine is made of.
Regarding the tasting itself is worth noting that there were almost no consensus among judges and that they would admit that the same tasting over the same wines from the same judges the next day would probably throw different results.
It is not that they can't distinguish between good and bad wines with a general accordance among all of them about what makes a wine good or bad but the stupid tendency to put everything into strictly ordered tabulated forms: it's not enough to say that Château Mouton-Rothschild '70 and Stag's Leap Wine Cellars '73 are both superb wines, you *have* to tell which one is better. Well, human perceptions don't work that way.
"tired of zombies, pirates, ninjas, and robots. Jesus"
Jesus won't help you here for He is Master or Zombies: He turned Lazarus into a zombie and He Himself was a zombie too (while it took him three days for the conversion).
"One big difference between American passenger trains (a-la-Acela) and the HSR trains in Madrid."
Those at Madrid were short-range trains for commuting, not high speed ones.
"For better or worse, American trains are basically rolling bank vaults legally required to be capable of surviving a head-on collision with a mile-long freight train."
That would change. On one hand you can't make those beasts to gain high speed, they need to be lighter; on the other is not needed: railways for high speed are exclusive for that service so you can't crash heads on with a freight train.
With regards of being targets, I'd say that the strongest value from plains is twofold: civil flights have the well-earned reputation of being the safest transport so taking one down has a stronger psycological effect; and then, a plane can be kind of a ballistic missile you can target against anything you want so nobody is safe, again a stronger psycological effect. So being the "terror factor" of a train more or less that of any other target (say, a sports center or a mcdonalds) its "cheaper" going for the low hanging fruit. Of course, terrorism is terrorism, so an attack to a train cannot be discounted as it can't be discounted for anything, and of course too, there's nothing preventing USA over-reacting on trains as it has done with plains. At least in the case of trains USA over-reacting won't have a global impact as it has been the case with planes (in Spain, even after the Madrid attack, high speed train security isn't too cumbersome for passengers).
"Paying a million people a year say, $50000 (to cover their wage plus the usual overhead) is $50 BILLION"
And that's, what? about 10% of the defense budget? There you have your money.
"Rail is the most efficient form of long-distance travel other than water, assuming that there is no value accorded to speed of transport."
Door to door over 200~400 miles times for high speed train are quite comparable to that of plane.
Coast to coast serviceability is quite comparable too (you can travel by night).
"If it was really cost effective some private company would have already built it."
Maybe in the XIX century... Hell, they did it!!!
With current corporate culture you won't get a gigantic project returning costs in 15-25 years starting from private hands.
"What merit does topology hiding have in your opinion?"
You'll know when ISPs start enforcing their "this connection is just for one computer" policies.
"If you can't afford $40/month, you are not the kinda guy that the ladies on e-harmony are looking for."
Can *you* afford it?
Hummm... we are talking here for dates 550 M.y.o, at $40/month, that makes... well, I know I won't pay that.
"Show me ANY Linux where I can take a mix of totally random hardware thrown together and hand my 67 year old clueless dad the disc and have him install it PERFECTLY, without a SINGLE fuckup or hardware issue, and then we'll talk."
Show me FIRST the Windows where I can take a mix of totally random hardware thrown together and hand my 67 year old clueless dad the disc and have him install it PERFECTLY, without a SINGLE fuckup or hardware issue.
"If Oracle for whatever reason decides to stop investing in BTRFS, the likely outcome AFAICS is not that BTRFS dies, but rather that Chris Mason and his team jump shop to Red Hat, Novell, Google, IBM or some other Linux contributor"
Are you sure their contracts allow for even thinking to do that?
"Btrfs is GPL licensed. Who started it is irrelevant. If Oracle drops it then others will pick it up."
Correction: others *may* pick it up. ...or may not.
"Not to mention that, while ZFS may not become a universal file system, it could well dominate in NAS appliances, and other proprietary closed-box products running OpenSolaris."
I don't think so.
Not because any technical problem with ZFS but because that other beast called "patents".
And this is not an opinion but a fact: Coraid already tried to build up a NAS device based on ZFS but NetApp made them think it twice by patent violation threatening.
"They paper over the fact that if this guy had been hit by a bus, his employer, the City of San Francisco, would well and truly have been up a creek without a paddle."
Which is a management issue, not a technical one, so the one to blame must be a manager. Was Childs in a manager-level position or in a "mere" technical one?
"However harsh the sentence may have been, the fact is that Childs was a shitty IT manager."
Truly so. But was he in a managerial position to start with? All I can find about him is that he was a "network administrator", a "network engineer" or an "IT administrator", never a manager, so he was not the one to say how the passwords should have to be managed nor the one to deal with policy violations. In fact, as per this reference (http://blogs.sfweekly.com/thesnitch/2010/08/terry_childs_sentenced_hacker.php) it seems clear that upper SF management agree this being a case of bad management: both Terry's direct manager and the security manager were displaced (they are not fired -yet, probably not to ashame that very SF upper management).
"You seriously think Microsoft will embrace an LGPL browser engine?"
Why not? They are against GPL (being communist and all that song) but they are quite in favour of the BSD (as long as they can take advantage of others' code, not that they distribute so much under the BSD themselves) and LGPL is second to the BSD in that they can take others' code to their advantage without giving back so much.
"it would apply to the geographical meaning of Europe, not the political meaning."
Do you really think "Europe" has any kind of geographical meaning?
"Quick, somebody should... Wait, german citizens already go to university for free..."
And no speed limit on their autobahn, so why would they want a simulator?
"On the other side the german economy has the biggest growth in europe with over 2%, looks we do get something right.
I would argue that copying the "American school" of economics and printing money to stimulate growth isn't really growth at all."
You know that Germans are within the eurozone so they don't print their money at all, do you?
"If you have something more complex than a flat file, then use relational databases. Even Access databases are better than a collection of text files."
Ever tried to access from Access (pun intended) big blob fields?
The guy has a lot of data. Therefore he needs a data-base. He already has one with a hierarchical storage (the filesystem) that probably conveys the way data is generated (hierarchically, if only by date).
Then he needs to access the data by different criteria. Are they relational? Then a relational data-base is probably what's needed. Access is still hierarchical, tree-based...?
My point is that (quite of obviously) data storage and data mining/retrival are different beasts so they can (and probably should) be managed in different ways/by different tools.
One first thing to do is "freezing" the data storage deployment. He says he has problems because dangling symlinks. That wouldn't happen if he didn't find the need to move the "original" data around. OK: don't do that and the problem will go away. Just let the "original" data go into a plain or almost plain structure: as long as there're no name collisions anything could do, from a single directory to a somehow optimized tree structure (like a first level directory list ordered by date or alphabetically or anything that fits and adding subdirs as needed to the point that any subdirectory only holds files on the hundreds or lower thousands).
The only important thing to remember is that once a file gets stored it never moves from its place (at least logically: with time maybe newer data can go to faster filesystems while older/less used data can go to second tier storage, etc.).
On top of that you add searching/datamining tools. Since your data storage is fixed by convention you can add as many and as unrelated searching tools as you see fit. It can be a forest of symlinks for different ordering criteria, it can be a RDBM, it can be a search engine, it can be accesable only by means of command line tools, a web interface... whatever and you can use all or part of them as need arises.
A practical example: Linux distributions' sources of packages like apt or yum. Packages are stored on an (almost) flat alphabetically ordered filesystem and then an in-parallel structure handles access (ordered by architecture, usage subset, etc.) by different means.
"I've seen entire departments laid off where they spends on the upwards of hundreds of thousands of dollars to train the department and still came to conclusion it was cheaper to fire them all and retrain people overseas."
That's why I carefully avoided talking about any company but I said "can any *country*...?" Maybe your company's stakeholders were glad with the quarterly benefits but will your country be a better place fifty years from now which such practices?
"There's so little taught in a university course that I couldn't read off a public library."
Except which contents you should read, of course.
"With no intent to fault him it was me who did the engineering and design he [an Engineering PhD Professor] just took the credit."
Maybe that was the case or maybe the PhD professor was able to validate your design, even accepting he didn't have the brains to come with it. Who knows.
"Class rooms and labs are nice tools for learning, but the fact of the new economics, that people will settle for less if it costs less."
Maybe. But then you'll end up with scholars of lessen value then. *Specially* at the face of the new economics can any country allow for worse knowledge workers? I don't think so.
"History, social sciences, psychology, most of philosophy and a lot of math subjects can be taught entirely over the web."
Only if you can socialize enterily over the web.
Even disregarding labs, the best point of dictated education is that it allows for "no holes" knowledge: at your own pace and growing from ignorance you don't even know what you ignore so probably you will end up "mastering" a narrow subset of an issue and lacking lots of basic knowledge. On the other hand, as an average and due to our psycological constrains it's much more difficult to find the will to go through the hard parts without day-to-day physical interaction. Not undoable but much more difficult. I.e.: in Spain there's an official "university-at-a-distance" and has been for years. Usually students take more years to gain their degrees, the abandon rate is higher and the percieved value for such degrees is lower than those from "brick-and-mortar" universities.
But this works both ways: the student can and do drive the professor too: an after the lecture chatting will make your lecturer to modify a bit future lectures because of the feedback he recieves from students, or will give you specific assignments at the sight of your interests and capabilities, and surely a face to face chat with a lecturer will easierly give you the motivation to gain or stregthen effort on the issue.