High-bandwidth busses - VME320 is rated at 500+ megabytes per second. PCI-X533 is rated at 533 million transfers per second. (No description is given on how much actual data it can ferry.) Let me know when you can buy a home computer with either of these. Especially VME, as that can survive some hefty punishment.
The reason you don't see VME is cost. My students and I operate some data acquisition equipment that would have been straightforward to implement in VME. Straightforward, yes, but it would have increased the cost of our hardware by factors of the order five. We were able to find some PCI based equipment that works pretty well.
When you can buy a perfectly decent complete PC for a few hundred dollars, which will permit you access to basic documents, spreadsheets, and the Web, why on Earth would the Dells of the world waste more than a few microseconds worrying about the microcosm of Slashdotters who want some of the specialized performance that VME can offer.
In terms of customer satisfaction, it would be cheaper for Dell to help slashdotters get dates, than to provide a big fat cable interface to a VME crate.
It would be great to be able to put up 100-to-1000 kg payloads in LEO, inexpensively, reliably, frequently. The space science community would be overjoyed, but they would not use the ability to assemble big things from little things.
Suppose that the way you got a new automobile was by having it mailed to you in 50 pound packages. It is probably possible to get cars this way, but the cars would be very different from the cars we have now --- heavier, slower, less fuel efficient, leaving a rather larger trail of (packaging) debris, and a few specialized tools, that would be useful for nothing else. Such cars would probably have no welds for assembly, and manufacturing techniques which required hazardous chemicals or heat processing would simply not be available. The lesser performance would arise from the "design for assembly" which is rather different from "design for manufacturability." (If the 50 pound limit was enforced, there would be no monolithic engine blocks; this would probably imply the existence of many small engines coupled together with a complex, heavier, transmission.)
The upshot here is that, from the point of view of weight, energy efficiency, complexity reduction, maintainability, testability, there is practically no sense in which it would be advantageous to create a big thing by launching little things for remote assembly, unless it becomes far cheaper (cost per kilogram) to launch little things than big things.
I thought the animations of the Mars Rover landings were quite satisfactory, and were a genuine aid in conveying the critical and unusual sequence of events (unusual because of the beach ball landing scheme.). Of course, I wasn't expecting to be entertained by these videos; I was expecting to be educated. Entertainment is Lucas' job, and education is (part of) NASA's.
Consider the adjacent Slashdot article about Lucas's new studio, http://slashdot.org/article.pl?sid=05/06/26/133217 &tid=186&tid=101, which is nicely and expensively appointed to generate additional cinematic circuses. I'm sure that Lucas et al. could provide a splendid animation of the comet impact. But:
Would you pay $10 to see it once?
Do you expect NASA to produce it for "free"?
Do you expect NASA to subcontract the video to a "real" CG house?
The box office from the Star Wars movies, and related paraphernalia licensing, sufficed to pay for several Shuttle missions, or perhaps ten major satellite programs, or a century's worth of space science at NSF. It may be that these films have inspired a few people to go into science and engineering, But these films are, of course, pure fantasy in their depiction of space and space travel. I don't mean to diminish the splendid entertainment that Lucas offers, but I can't help the following comparison:
The Star Wars movies are, to the perception of space travel, what pornography is to the perception of sex.
Items 2 and 3 above will strongly impact NASA's budget; high quality CG added to a documentary structure could easily run in the mid seven figures for a single film. For a tenth that amount you can get Pretty Good results, and keep a hundred grad students in beer and chips for a year.
Those hundred grad students will get you to Mars in twenty years. Or, you could help George Lucas buy a spare yacht today.
In the case of Go, computers perform abysmally compared to humans.
While very high-end computer chess machines now play strong grandmaster chess, it takes relatively little practice to beat the best Go-playing computer.
In chess, "search" is part of the computer algorithm, and it is hard in chess because the tree of possibilities gets big in a hurry.
But in Go it is far worse.
In chess there are (I think) 16 first pawn moves + 4 first knight moves, and the same holds for black --- so that there are 400 possible positions after white and black have moved.
In Go, however, there are 361 possible first moves, and 360 possible second moves; divide by 8 if you wish for rotational symmetries, resulting in some 16,000 different possible positions after the first round of moves.
Humans have a spectacular ability to detect patterns which has yet to be duplicated in machines --- as anyone practicing speech recognition or image understanding can tell you. It is in fact quite remarkable how well the chess machines are doing.
Why cant the open source community get behind an open source college?
It's a fair question. And here are some answers.
http://en.wikibooks.org/wiki/Wikiversity is precisely such a thing. It is young, and parts of it that I have seen are quite lame, but the intent is good, and the structure is there. The content requires... effort.
accreditation would be an issue. If you think that's silly, then contemplate what lack of accreditation or safety standards would mean for air travel.
Quite a few universities have stepped up to the "online degree" watering hole, and they have discovered that:
it is relatively easy to put up crappy content that no one will buy, and which will sully the University's good name.
it is remarkably expensive to put up quality content which is actually worthy buying; the cost rivals or exceeds bricks & mortar conventional courses.
there is usually very little incentive for regular faculty to participate in online course delivery, unless the enrollment is very high (why? because it is a lot of extra work). So, if you're looking for that extra special course in kinetic theory of plasmas with application to incoherent scatter, well, don't hold your breath.
Likewise, universities have "entrepreneurial" units to go develop stuff for online course delivery. This inevitably begets warfare with the Department that "owns" the course. Example: my university offered a "certificate program" of four courses in electrical engineering. This is just dandy, until the Electrical Engineering Department discovered (by accident) that someone else was offering their courses. The "entrepreneurial" unit hadn't quite bothered to check with the home Department...
To expand upon a topic in the previous list, the authoring tools for WWW-based content delivery are... extremely poor, at least in relation to what you're trying to do. In a classroom, there is opportunity for detailed and remarkably complex interaction with a functioning expert system (the professor) as well as the other students. Just try to capture that functionality in some 'bot. Along those lines, see the recent James Fallows article explaining just how poor modern search engines are in answering questions. Google is wonderful! But it's also remarkably primitive compared to what we'd like to be able to do.
If you have ever wondered why there are so few really good WWW-based demos available, consider this: A really good, effective demonstration takes a minute or two to show to the class. However, it can easily take 12 hours of development time to prepare a quality demo that will be used one time, and fill 1 minute of lecture. It doesn't take long to realize that that development time is unjustifiable. (At my own university, there is the very real danger that the computer projection equipment will simply be out of order. There is no satisfaction in wasting 5 minutes of lecture time to show a 1 minute demo).
... and for all you l33t h4korz or however you spell it, there is more to a college education than learning how to program good (as Derek Zoolander might have put it.) The economic forces which create a faculty work force continue to develop a faculty which, however haphazardly, values breadth and experience and (yes) literature and history in addition to being able to log on and hack.
A college degree is not a commodity (yet); it is not like 87 octane gasoline dispensed at the pump. The college degree represents a period of time in which you study a lot of useless things in the hope that some of them will surprise you by being interesting; that the depressing or boring things will at least teach you how to wade through depressing or boring material for the rest of your life. It is a period of time when people stop being te
Although I understand zkn's sentiment, isn't it really equivalent to a regime in which nothing is checked?
Let me offer an example. Several years ago, a reporter named Byron Acohido wrote a series of articles about rudder problems in the Boeing 737 (http://flash.uoregon.edu/F97/acohido.html). In these articles, BA identified the rudder as the likely cause of two crashes (he's right), and he outlined his perception of slow response and stonewalling by both Boeing and the FAA. BA went on to win quite a few awards, including a Pulitzer, for these articles. In particular, BA chastised Boeing for not moving rapidly to correct the rudder problem.
But BA's articles missed some critical, absolutely critical analysis:
airplanes are stupendously complex. They are perhaps the quintessential example of "if it ain't broke, don't fix it."
the 737 is a safe airplane. When BA flew out to accept his Pulitzer, he could not have been safer on any plane other than a 737, even with the original rudder.
the 737 had a known, rare, failure mode, which flight crews were trained to deal with.
correcting this flaw hastily could easily have introduced new flaws that were not known or understood.
In fact, the "Do Something!" imperative offered by BA's articles could, quite conceivably, have made the 737 more dangerous.
It may be the case that Boeing and the FAA could have behaved better --- but Acohido's articles revealed a tremendous lack of understanding of objective safety statistics. The public was ill served by the accolades that Acohido received for his articles.
(In case you wondered: in my day job I'm a professor in the Pacific Northwest, and I do ionospheric physics. I have no contact with the FAA. About a decade ago I had a small grant from Boeing, to do a project that I was spectacularly unqualifed to perform, but which needed a PI.)
Often heard, but not true. In fact, wheel re-invention is extremely useful.
"pretty good" programming languages were available decades ago, but people keep inventing new ones, and some of the new ones are pretty great.
Mathematicians and physicists frequently reinvent things, because better tools become available. example: proofs of Stokes' Theorem and Gauss's Law require a bit of effort in classical calculus, but both become special cases of a much more general theorem when you have the tools of differential forms available.
GNU/linux is pretty clearly the result of wheel reinvention. some of us think this has been a pretty useful activity.
wheel reinvention is obviously useful as a pedagogical tool. How many million times have students laid out some elementary circuit in VLSI, say, an eight bit adder? Would you hire someone to design a chip who had read all the literature, even memorized it, but had never actually laid out a single chip?
wheel reinvention is a critical (and underused) feature of modern science. In principle, peer-review is a kind of wheel reinvention, however it is usually in the form of checking the math, if you will (that's not even always possible http://www.google.com/search?q=four+color+theorem) . The best kind of peer review is duplication: can somebody else duplicate the experment? It is a real tragedy in modern Science that whoever was First gets all the credit, when the person who was Second should earn our deep gratitude for independently checking the result.
Wheel reinvention provides a critical opportunity for the advance of science and technology, by creating an opportunity to find a better way, and to detect previously undiscovered vulnerabilities.
The average engineer has a superior IQ, but can barely match his belt with his shoes.
But, but, I thought that a belt was for holding up my sliderule holster and my pants, and my shoes, well, I guess that I wear shoes because my mom doesn't let me leave the house without them.
Do you mean to say that they are supposed to match?
Thanks for playing, AC! but why not check some of the manifests for Shuttle flights; and whether the astronauts have security clearances; etc. The notion that NASA is "entirely civilian" is... quaint.
This is also not a problem. It doesn't matter *who* demonstrated the prior art. The only requirement is that it be prior.
If NASA today tried to patent "putting humans on the Moon" they would be denied the patent because of prior art... demonstrated by NASA.
Of course, NASA might then try to claim that the prior art was actually real Art, in the sense of "it was only a movie!" so that in fact putting humans on the Moon hadn't actually been demonstrated yet. The tinfoil hat crowd would immediately begin to suspect that the movies were faked, and that humans actually had landed on the Moon.
What would be useful here is a new term that would permit the distinction between "hunters" and "dickwads with big guns." Currently we lump the two groups together and heap derision upon both for the sins of the latter (and some don't like the former, either).
In my case, I *have* been deer hunting and goose hunting --- myself armed with a camera, and my companions with guns. I've had a bleeding deer carcase in my lap for 45 miles bouncing along in an open jeep in 25F weather,... and thought myself lucky to have had an interesting day.
I don't think I could pull the trigger, and there is that little issue that I'm a vegetarian. But I don't hate "hunters."
I do, however, hate dickwads with guns. In my day job, I put up scientific apparatus in remote places, and dickwads with guns use my antennas for target practice, chop up my coax, steal the guy lines, and generally remind us that the gene pool has a shallow end.
But if there is one group of people who should *really* loathe dickwads with guns, it is... the hunters, of course. It may be shallow to lump hunters together with dickwads with guns... but the hunters would not suffer so much abuse if the dickwads with guns went away forever.
It's much more than just CO2. It's isotope ratios, dust deposits,
In soil and lakebed cores, a similar variety of information exists, and it really is hundreds of thousands of years.... and why wouldn't you put ice core data on par with weather station data? Do you know much about either one? What makes you think that surface weather station data are more informative than (say) snow/ice deposition in tracking climate? Weather signals have very large variance, whereas "average" signals such as ice deposition, silt deposition,... those integrated signals are far less subject to the enormous daily variability associated with weather.
... and this parent's comment is correct. The point is, if the proposal is funded, and doesn't promise dissemination of raw data, then dissemination of that raw data is not obligated. In my corner of NSF, dissemination of raw data is widely encouraged... and rarely funded. Our (enlightened, and broke) program managers encourage us to do what we can, w.r.t. dissemination of raw data, but do not expect miracles.
(Actually, though, this practice is unfortunately very common in academic sciences, largely as a way for universities to supplement their grant income.)
This statement reflects a misconception about "data" and "grants." I'm amused to be repeating an argument that I made nearly 20 years ago, while a grad student.
If I write a proposal to NSF to run an experiment and "get some data" then it is true in a sense that the US citizen/taxpayers "paid for that data." But that doesn't mean that they get to see it. You see, the point in these experiments is to "get data" and then "analyze it" and then "make some interesting conclusions." Typically the funded grant did not have a budget item "salary for a WWW-dude and disk space to distribute all this data to who ever wants it."
As it happens, many people these days (including me) work pretty hard to make our raw data available. But very very often the effort to do that is completely donated effort, done at midnight, on weekends,...
Believe me, there is nothing the science community would like more than money for new IT salary lines, which we desperately need. But as long as Congress keeps shafting NSF, it simply isn't going to happen.
The general public often has unrealistic notions of what "raw data" is anyway. Our instrument can generate 20 GBytes/day if we turn on all the bells and whistles. I know of instruments which generate several GBytes per second, and radio astronomers are thinking of ways to generate TBytes/second. For data rates like that, SneakerNet is still about the only viable distribution mechanism.
Actually that is not so. Tubes are the most linear amplifying devices. http://www.dogstar.dantimax.dk/tubestuf/miniblk1.h tm
Bzzzzt. This is absolutely false. In terms of dynamic range, the most linear amplifiers are *clearly* transistors. In my line of work, you'd be laughed out of the room if you dared to use vaccuum tube amps. For RF situations requiring very high dynamic range, vacuum tubes are *not* an option. Tubes are *not* used for truly high dynamic range applications.
Audiophiles do *not* have high dynamic range demands compared to some radar applications; and in these radar applications, there is no room for parlor quibbles about soft-clipping and warmer sounds and harmonics. There is only reality, in the form of whether or not you notice the incoming missile. So, please. I think it is a fine thing for audiophiles to blow kilobucks on tubes to make their guitars sound optimally crappy.
10 kW is definitely possible for transistors. I've seen a 50 kW transistor-based transmitter, for example, and I've got acquaintences working on a 1 MW transistor-based transmitter (for a phased array radar).
But if you want a lot of RF power for low cost, tubes are still the way to go.
Linus has moved to Portland, OR, which is a fine thing, and as others have noted, that puts him amusingly close to Redmond, WA.
I believe that this may provide a possible explanation for the recent eruption of a volcano (Mt. St. Helens) fairly close to the midpoint between Bill and Linus.
In some of your books, your action scenes are far detailed (and better informed) than are those of many authors, who gloss over the ways that actual physical objects, including people, interact at close range (including skateboarding, diving, fighting, and the awkwardness of in-car sex with Amy Shaftoe).
This reminds me of one of my favorite moments in Snow Crash, when Hiro takes off on a motorbike on the AlCan highway. Turning the page I wondered, "well what the hell is Stephenson going to occupy me with here?" only to have Hiro comment, "after that it's just a chase scene" and NS didn't even bother to fill in the details. Made me laugh, like so many other clever things in his books.
It's still just a guessing game. They could pick a number from 1 to 10 and ask you to guess it; one person in 10 would get it on the first try. Or maybe their favorite number is 6 and they gave a hint to that effect, so one person in 3 would get it on the first try.
Forgive me, but Google's puzzles were not at all "guessing games." The first (billboard) problem was completely well-posed. The second problem required some thought to find some sort of pattern or progression that four numbers satisfied, so that a fifth might be identified. Although the answer has been given, I'll still be coy and say that these numbers had the same hash value (calculated a certain, reasonable way); they were the first four numbers to achieve that hash, and the answer was the fifth available with that hash.
Good puzzles have objective solutions; any answer that meets the requirements inherent in the question is correct, and can be verified correct independently.
I agree. Remember, the first problem was completely well-posed.
The second problem was not well posed mathematically, but the pattern is very simple --- so simple that, in the Occam's Razor sense, I would be amazed if there is a second, equally simple rule, which leads to a valid (but incorrect) answer.
By analogy, many inverse problems are very ill-posed. The usual way to resolve them is by "regularization" which can be thought of as "make the answer sort of smooth" or with a bit more fancy talk, you use a Maximum Entropy constraint to restrict the choice of possible models. There are some very potent arguments in favor of Maximum Entropy as a rationale; and it is not so bizarre to assert that "Maximum Entropy" and "Occam's Razor" are, in fact, kin. Each basically suggests, "choose the simplest model which explains the available facts." If you follow this logic you will be led inexorably to the solutions which Google had in mind. Whether you call it Occam's Razor or Maximum Entropy, the answers really are effectively unique.
To assert otherwise is to defend solutions which are correct with probability zero. After all, we're talking about 10 digit sequences of a transcendental number; a "technically correct" but "google-y wrong" solution to the second problem has probability 1 part in 10^10. For the first problem, I claim there is exactly one solution.
Your mathematics professor should take a refresher course in logic. Bill Gates paid for most of the building. Donald Trump didn't. See the difference?
Although Bill Gates paid for more of Stanford's building than did Donald Trump, Gates did not pay for most of it. In fact, Gates wangled quite a deal --- for merely $7M out of about $60M he got his name on the bricks. Nice Trick! At least CMU is getting $20M,...
I'm campaigning to get some bricks named "Chestnut & Gibson" a couple of people who died while doing their jobs in DC. People like these never get things named for them. If Gates has the CMU building named for Chestnut & Gibson, then I'll be impressed.
Slashdot Mirror
The reason you don't see VME is cost. My students and I operate some data acquisition equipment that would have been straightforward to implement in VME. Straightforward, yes, but it would have increased the cost of our hardware by factors of the order five. We were able to find some PCI based equipment that works pretty well.
When you can buy a perfectly decent complete PC for a few hundred dollars, which will permit you access to basic documents, spreadsheets, and the Web, why on Earth would the Dells of the world waste more than a few microseconds worrying about the microcosm of Slashdotters who want some of the specialized performance that VME can offer.
In terms of customer satisfaction, it would be cheaper for Dell to help slashdotters get dates, than to provide a big fat cable interface to a VME crate.
It would be great to be able to put up 100-to-1000 kg payloads in LEO, inexpensively, reliably, frequently. The space science community would be overjoyed, but they would not use the ability to assemble big things from little things.
Suppose that the way you got a new automobile was by having it mailed to you in 50 pound packages. It is probably possible to get cars this way, but the cars would be very different from the cars we have now --- heavier, slower, less fuel efficient, leaving a rather larger trail of (packaging) debris, and a few specialized tools, that would be useful for nothing else. Such cars would probably have no welds for assembly, and manufacturing techniques which required hazardous chemicals or heat processing would simply not be available. The lesser performance would arise from the "design for assembly" which is rather different from "design for manufacturability." (If the 50 pound limit was enforced, there would be no monolithic engine blocks; this would probably imply the existence of many small engines coupled together with a complex, heavier, transmission.)
The upshot here is that, from the point of view of weight, energy efficiency, complexity reduction, maintainability, testability, there is practically no sense in which it would be advantageous to create a big thing by launching little things for remote assembly, unless it becomes far cheaper (cost per kilogram) to launch little things than big things.
Consider the adjacent Slashdot article about Lucas's new studio,
http://slashdot.org/article.pl?sid=05/06/26/13321
- Would you pay $10 to see it once?
- Do you expect NASA to produce it for "free"?
- Do you expect NASA to subcontract the video to a "real" CG house?
The box office from the Star Wars movies, and related paraphernalia licensing, sufficed to pay for several Shuttle missions, or perhaps ten major satellite programs, or a century's worth of space science at NSF. It may be that these films have inspired a few people to go into science and engineering, But these films are, of course, pure fantasy in their depiction of space and space travel. I don't mean to diminish the splendid entertainment that Lucas offers, but I can't help the following comparison:Items 2 and 3 above will strongly impact NASA's budget; high quality CG added to a documentary structure could easily run in the mid seven figures for a single film. For a tenth that amount you can get Pretty Good results, and keep a hundred grad students in beer and chips for a year.
Those hundred grad students will get you to Mars in twenty years. Or, you could help George Lucas buy a spare yacht today.
In the case of Go, computers perform abysmally compared to humans.
While very high-end computer chess machines now play strong grandmaster chess, it takes relatively little practice to beat the best Go-playing computer.
In chess, "search" is part of the computer algorithm, and it is hard in chess because the tree of possibilities gets big in a hurry.
But in Go it is far worse.
In chess there are (I think) 16 first pawn moves + 4 first knight moves, and the same holds for black --- so that there are 400 possible positions after white and black have moved.
In Go, however, there are 361 possible first moves, and 360 possible second moves; divide by 8 if you wish for rotational symmetries, resulting in some 16,000 different possible positions after the first round of moves.
Humans have a spectacular ability to detect patterns which has yet to be duplicated in machines --- as anyone practicing speech recognition or image understanding can tell you. It is in fact quite remarkable how well the chess machines are doing.
It's a fair question. And here are some answers.
A college degree is not a commodity (yet); it is not like 87 octane gasoline dispensed at the pump. The college degree represents a period of time in which you study a lot of useless things in the hope that some of them will surprise you by being interesting; that the depressing or boring things will at least teach you how to wade through depressing or boring material for the rest of your life. It is a period of time when people stop being te
Let me offer an example. Several years ago, a reporter named Byron Acohido wrote a series of articles about rudder problems in the Boeing 737 (http://flash.uoregon.edu/F97/acohido.html). In these articles, BA identified the rudder as the likely cause of two crashes (he's right), and he outlined his perception of slow response and stonewalling by both Boeing and the FAA. BA went on to win quite a few awards, including a Pulitzer, for these articles. In particular, BA chastised Boeing for not moving rapidly to correct the rudder problem.
But BA's articles missed some critical, absolutely critical analysis:
- airplanes are stupendously complex. They are perhaps the quintessential example of "if it ain't broke, don't fix it."
- the 737 is a safe airplane. When BA flew out to accept his Pulitzer, he could not have been safer on any plane other than a 737, even with the original rudder.
- the 737 had a known, rare, failure mode, which flight crews were trained to deal with.
- correcting this flaw hastily could easily have introduced new flaws that were not known or understood.
- In fact, the "Do Something!" imperative offered by BA's articles could, quite conceivably, have made the 737 more dangerous.
It may be the case that Boeing and the FAA could have behaved better --- but Acohido's articles revealed a tremendous lack of understanding of objective safety statistics. The public was ill served by the accolades that Acohido received for his articles.(In case you wondered: in my day job I'm a professor in the Pacific Northwest, and I do ionospheric physics. I have no contact with the FAA. About a decade ago I had a small grant from Boeing, to do a project that I was spectacularly unqualifed to perform, but which needed a PI.)
Often heard, but not true. In fact, wheel re-invention is extremely useful.
Wheel reinvention provides a critical opportunity for the advance of science and technology, by creating an opportunity to find a better way, and to detect previously undiscovered vulnerabilities.
But, but, I thought that a belt was for holding up my sliderule holster and my pants, and my shoes, well, I guess that I wear shoes because my mom doesn't let me leave the house without them.
Do you mean to say that they are supposed to match?
Thanks for playing, AC! but why not check some of the manifests for Shuttle flights; and whether the astronauts have security clearances; etc. The notion that NASA is "entirely civilian" is
This is also not a problem. It doesn't matter *who* demonstrated the prior art. The only requirement is that it be prior.
... demonstrated by NASA.
If NASA today tried to patent "putting humans on the Moon" they would be denied the patent because of prior art
Of course, NASA might then try to claim that the prior art was actually real Art, in the sense of "it was only a movie!" so that in fact putting humans on the Moon hadn't actually been demonstrated yet. The tinfoil hat crowd would immediately begin to suspect that the movies were faked, and that humans actually had landed on the Moon.
What would be useful here is a new term that would permit the distinction between "hunters" and "dickwads with big guns." Currently we lump the two groups together and heap derision upon both for the sins of the latter (and some don't like the former, either).
... and thought myself lucky to have had an interesting day.
... the hunters, of course. It may be shallow to lump hunters together with dickwads with guns ... but the hunters would not suffer so much abuse if the dickwads with guns went away forever.
In my case, I *have* been deer hunting and goose hunting --- myself armed with a camera, and my companions with guns. I've had a bleeding deer carcase in my lap for 45 miles bouncing along in an open jeep in 25F weather,
I don't think I could pull the trigger, and there is that little issue that I'm a vegetarian. But I don't hate "hunters."
I do, however, hate dickwads with guns. In my day job, I put up scientific apparatus in remote places, and dickwads with guns use my antennas for target practice, chop up my coax, steal the guy lines, and generally remind us that the gene pool has a shallow end.
But if there is one group of people who should *really* loathe dickwads with guns, it is
Yikes! I married a Caltech girl! Or ... at least that's what I thought I did.
It's much more than just CO2. It's isotope ratios, dust deposits,
... and why wouldn't you put ice core data on par with weather station data? Do you know much about either one? What makes you think that surface weather station data are more informative than (say) snow/ice deposition in tracking climate? Weather signals have very large variance, whereas "average" signals such as ice deposition, silt deposition, ... those integrated signals are far less subject to the enormous daily variability associated with weather.
In soil and lakebed cores, a similar variety of information exists, and it really is hundreds of thousands of years.
... and this parent's comment is correct. The point is, if the proposal is funded, and doesn't promise dissemination of raw data, then dissemination of that raw data is not obligated. In my corner of NSF, dissemination of raw data is widely encouraged ... and rarely funded. Our (enlightened, and broke) program managers encourage us to do what we can, w.r.t. dissemination of raw data, but do not expect miracles.
This statement reflects a misconception about "data" and "grants." I'm amused to be repeating an argument that I made nearly 20 years ago, while a grad student.
If I write a proposal to NSF to run an experiment and "get some data" then it is true in a sense that the US citizen/taxpayers "paid for that data." But that doesn't mean that they get to see it. You see, the point in these experiments is to "get data" and then "analyze it" and then "make some interesting conclusions." Typically the funded grant did not have a budget item "salary for a WWW-dude and disk space to distribute all this data to who ever wants it."
As it happens, many people these days (including me) work pretty hard to make our raw data available. But very very often the effort to do that is completely donated effort, done at midnight, on weekends,
Believe me, there is nothing the science community would like more than money for new IT salary lines, which we desperately need. But as long as Congress keeps shafting NSF, it simply isn't going to happen.
The general public often has unrealistic notions of what "raw data" is anyway. Our instrument can generate 20 GBytes/day if we turn on all the bells and whistles. I know of instruments which generate several GBytes per second, and radio astronomers are thinking of ways to generate TBytes/second. For data rates like that, SneakerNet is still about the only viable distribution mechanism.
Actually that is not so. Tubes are the most linear amplifying devices. http://www.dogstar.dantimax.dk/tubestuf/miniblk1.h tm
Bzzzzt. This is absolutely false. In terms of dynamic range, the most linear amplifiers are *clearly* transistors. In my line of work, you'd be laughed out of the room if you dared to use vaccuum tube amps. For RF situations requiring very high dynamic range, vacuum tubes are *not* an option. Tubes are *not* used for truly high dynamic range applications.
Audiophiles do *not* have high dynamic range demands compared to some radar applications; and in these radar applications, there is no room for parlor quibbles about soft-clipping and warmer sounds and harmonics. There is only reality, in the form of whether or not you notice the incoming missile. So, please. I think it is a fine thing for audiophiles to blow kilobucks on tubes to make their guitars sound optimally crappy.
Everyone needs a hobby.
heh. That's one of the finest summaries of the US defense industry I have ever read.
An understandable, but incorrect assumption: the radar in question is not for defense purpose.
10 kW is definitely possible for transistors. I've seen a 50 kW transistor-based transmitter, for example, and I've got acquaintences working on a 1 MW transistor-based transmitter (for a phased array radar).
But if you want a lot of RF power for low cost, tubes are still the way to go.
Can you name a more widely used application of tubes now days?
... although that status is now eroding rapidly.
Practically all high power radio transmitters use vacuum tubes.
All your atom smashers use klystrons and their kin to goose those particles along.
As others have pointed out, most computer monitors are *still* vacuum tube devices
Linus has moved to Portland, OR, which is a fine thing, and as others have noted, that puts him amusingly close to Redmond, WA.
I believe that this may provide a possible explanation for the recent eruption of a volcano (Mt. St. Helens) fairly close to the midpoint between Bill and Linus.
This reminds me of one of my favorite moments in Snow Crash, when Hiro takes off on a motorbike on the AlCan highway. Turning the page I wondered, "well what the hell is Stephenson going to occupy me with here?" only to have Hiro comment, "after that it's just a chase scene" and NS didn't even bother to fill in the details. Made me laugh, like so many other clever things in his books.
Forgive me, but Google's puzzles were not at all "guessing games." The first (billboard) problem was completely well-posed. The second problem required some thought to find some sort of pattern or progression that four numbers satisfied, so that a fifth might be identified. Although the answer has been given, I'll still be coy and say that these numbers had the same hash value (calculated a certain, reasonable way); they were the first four numbers to achieve that hash, and the answer was the fifth available with that hash.
I agree. Remember, the first problem was completely well-posed.
The second problem was not well posed mathematically, but the pattern is very simple --- so simple that, in the Occam's Razor sense, I would be amazed if there is a second, equally simple rule, which leads to a valid (but incorrect) answer.
By analogy, many inverse problems are very ill-posed. The usual way to resolve them is by "regularization" which can be thought of as "make the answer sort of smooth" or with a bit more fancy talk, you use a Maximum Entropy constraint to restrict the choice of possible models. There are some very potent arguments in favor of Maximum Entropy as a rationale; and it is not so bizarre to assert that "Maximum Entropy" and "Occam's Razor" are, in fact, kin. Each basically suggests, "choose the simplest model which explains the available facts." If you follow this logic you will be led inexorably to the solutions which Google had in mind. Whether you call it Occam's Razor or Maximum Entropy, the answers really are effectively unique.
To assert otherwise is to defend solutions which are correct with probability zero. After all, we're talking about 10 digit sequences of a transcendental number; a "technically correct" but "google-y wrong" solution to the second problem has probability 1 part in 10^10. For the first problem, I claim there is exactly one solution.
Although Bill Gates paid for more of Stanford's building than did Donald Trump, Gates did not pay for most of it. In fact, Gates wangled quite a deal --- for merely $7M out of about $60M he got his name on the bricks. Nice Trick! At least CMU is getting $20M,
I'm campaigning to get some bricks named "Chestnut & Gibson" a couple of people who died while doing their jobs in DC. People like these never get things named for them. If Gates has the CMU building named for Chestnut & Gibson, then I'll be impressed.