The Earth has a mass of 6x10^24 kg. A billion tonnes is 10^12 kg; this is about 2x10^-13 of Earth's mass. But the mass would carry off about 10^3 times its normal share of Earth's angular momentum -- so launching a billion tons would remove about 2x10^-10 of Earth's angular momentum (and rotational speed). There are about pix10^7 seconds in a year, so in one year you'd see a 5 millisecond drift between Earth time and your atomic clock.
That's no small potatoes, for ultraprecision wonks.
Put another way, if you launched a billion tons a year you would have about half of the effect on Earth's rotation as the Moon's measured tidal drag over the last 30 years.
...unless I was being a fscktard and doing something stupid.
That's no excuse. No OS should ever crash for any software-induced reason, ever. There's a famous story (perhaps in the Jargon File?) about a UNIX system that got half-blown-away by a misplaced "rm -rf/" and was recovered without rebooting. Now that's robust.
Perhaps you guys should start working on usability and driver coverage.
Spoken like a true flamethrower! IHBT, but I'll bite anyway. I just installed Mandrake 10 and I'm amazed at the usability -- it's really quite polished.
Drivers are slow to arrive mainly because nearly every single one requires someone to spend a month or two reverse-engineering some proprietary interface. But again, they're not really much of a problem anymore. There are still a few new-ish unsupported devices (the Centrino wireless cards are an example), but the windows compatibility layer takes care of that.
I recently bought CrossoverOffice to have access to Microsoft Weird and Internet Exploder (required by upper management to access company forms). I was delighted to find that it runs flawlessly under my bit-rotting Red Hat 9 system.
I've tried perhaps a dozen other apps (including YetiSports) and they all worked with no glitches.
The one feature that CrossoverOffice appears not to support well is trashing your whole system with malware -- you can grant your applications read-only access to your main file-system, or limited write access to particular directories.
I think the intended application is not bundling large numbers of NTSC quality movies; the application is shipping more pixels per frame of a single movie.
Neither one is feasible, for reasons beyond my control -- mostly having to do with cost.
Further, DLT is expensive and slow compared to platter technology -- the fundamental problem is that hard drives and computers have gotten better much faster than tape drives and the network have.
I'm in the middle of downloading 30 GB of data from one of the SOHO instruments; it will take 3 days to get it over our T1. The only advantage of doing the transfer over the net is that putting it on DVDs for mailing would require somebody on their end to monitor and swap out 6-7 DVDs as they're burned, and then somebody on my end to monitor and swap out those DVDs as they're read onto my hard drive. With a Blu-Ray disk they could burn a single medium then drop it in the mail. And I'd still get the data at the same time as my network transfer will finish.
To me, a rather damning part of the whole statement was that the data would be available in December -- basically, "You can have the data, but not if you plan to use it to investigate the candidates' integrity for this election".
Two of the axes are stable in the sense you're describing; the third (the intermediate axis) is definitely not -- it is at best metastable, because small deviations lead to large excursions of the axis, on the precessional timescale.
One might argue that the minimal axis isn't stable, in the sense that any artificially induced damping causes perturbations to grow, on generally longer timescales than precession. But the maximal axis is definitely stable, in the sense that any physical object isn't perfectly rigid and therefore will gradually damp toward the maximal axis.
If we put up a cylindrical satellite spinning around the maximum moment axis, it will stay that way (or close to it) forever, provided only that the satellite is even a little bit flexible. Interaction with tiny tidal forces and drag can perturb the absolute direction of the rotational axis, but they cannot prevent small sloshes and flexes from gradually working the body itself around so that the maximal axis points along the angular momentum vector.
But, er, none of this is particularly relevant to the SpaceShipOne, eh? It has three quite different moments of inertia, and doesn't stay up long enough for anything but the basic instability of tumbling to matter much. I'd invite you out for a beer, if you weren't probably somewhere else than me entirely.
Nah, only ideal blocks, ideal balls, and ideal disks can do that. Real ones are never perfectly symmetric.
Your point is well taken -- symmetry can keep the tumbling from being rapid. But sooner or later, everything macroscopic and rigid will precess and/or tumble.
Even if you discount that argument, the intermediate axis (or any degenerate axis) isn't stable -- it's at best metastable.
Ah. Some of the coverage points out that there is a powered trim system for supersonic flight and a straight mechanical system for subsonic flight. Again, I'm speculating -- but I imagine that the control stick just shifts over to running the reaction jets when the air pressure gets low enough -- or maybe it actuates them all the time; they shouldn't affect in-atmosphere flight very much.
Hmmm... In this case, I think the "flight software" might all be resident in the pilot's wetware. (I haven't heard anyone talking about an attitude controller in the official releases...)
Nobody has been far enough from Earth to ignore its gravitational field, since the Apollo flights; and even for Apollo, the reason astronauts ignored Earth's large field is that they were in another large gravity well.
Low-Earth orbit is close enough to Earth's surface that you are "at the top of a parabolic trajectory" (actually, in an elliptical trajectory) all the time -- accelerating toward Earth's surface at like 0.99 gee.
Earth's field only gets negligible at distances of about a million miles -- that far out, the Sun's field dominates the local gravitational environment.
One might think that tumbling is easy to control -- after all, if the craft is spinning and you have cold gas thrusters, you can just fire the jets to oppose the spinning, right?
wrong.
Most objects do not spin cleanly about most axes. Rigid bodies (such as books, spaceships, rocks, lollipops, and bullets) have three "principal axes" that pass through the center of gravity and are determined by the mass distribution in the object. There's a "minimum" axis that minimizes the kinetic energy for a given angular momentum -- that's the axis around which the thing is the most clustered. For a screwdriver, the minimum axis generally points down the length of the scredriver shaft. There's also a "maximum" axis around which the thing is the most spread out of any direction. For a flat object like a book or a pancake, the maximum axis points directly out of the flat face. Those are the only two axes around which you can spin the object and have it stay stable.
Any other direction will give rise to precession and tumbling, even in vacuum! You can try it with a book -- most closed hardback books have the minimum axis pointing up through the top of the middle pages, and the maximum axis pointing out through the front of the cover. The third dimension -- pointing out through the spine -- is not stable. Tape a book shut and flip it in the air: if you flip it around the maximum or minimum moment axis it will do what you think -- just flip over before you catch it again. If you flip it around the intermediate axis (by, say, starting with the book facing you right-side up with the spine on the left, and pulling the bottom edge toward you as you throw it up in the air) then you might expect the spine to stay on your left side -- but it will flip back and forth, often ending up on your right side, as the book tumbles in the air. (Remember to tape the book closed before tossing it!).
Anyhow, that's a problem for stopping spin and tumbling, because it's not always obvious which way to fire the cold-gas jets to slow down your rotation: by the time you actually fire them you might have tumbled around so that they are speeding you up instead of slowing you down.
I guess that's why "carefree re-entry" is such a great feature of SpaceShipOne -- it's remarkable that they were able to land safely even without good attitude control at apogee.
They're using stuff that's more like 50% pure. If they could get 97% H2O2, they'd be in the running for the X-prize, instead of 8-10 months behind Rutan.
...with the marginally inferior audio quality they provide as opposed to LPs
I can't resist here. I've been ripping my LPs with a nice digitizer card and comparing to CDs. The LP sound may be nice and warm, and it's certainly slightly different than the CD from the same album -- but higher fidelity it ain't. To start with, the measured noise floor on my very best LPs is about -40dB. That's after highpass filtering to get rid of the rumble that comes from multiplexing the sound signal with the tracking error (remember, the needle tracks under the influence of small pushes from the record), and also after gating out the 60 Hz and harmonics that are picked up by my magnetic cartridge. And, yes, the system can handle better: making the same measurement with the turntable running and the stylus not touching the record (cue lever engaged) yields more than 10x lower noise floor -- so I'm pretty sure I'm measuring surface quality of the vinyl.
The sounds that come out of the turntable are indeed a little "warmer" than most CDs. For example, I like the smooth, jetstream sounds of Abbey Road side 2 better off the LP than off the CD of the same album. But the CD is definitely cleaner.
Comparatively pure sounds (like flutes) come out measurably distorted: my LPs seem to have harmonic distortion at about the -30dB (0.1%) level in the best cases, based on deconvolving the LP Fourier spectrum using the appropriate CD sample as a comparison (same sound, two media). Some albums are much worse. It's not clear whether the harmonic distortion is due to the records I'm using or due to my turntable, as I can't be bothered to repeat the experiment with someone else's equipment. There's also stereo crosstalk at -15 - -20 dB.
Admittedly, it's no audiophile deck -- I "only" have a belt drive with 2kg turntable and about a $100 cartridge -- but it certainly qualifies as "hifi".
In short -- "good sound" it is, but "higher fidelity" it aint. It's telling that one can't tell the difference between hearing the LP directly off the album or replayed (uncompressed) from my sound card: the 44kHz, 16bit sampling is plenty good to capture whatever's coming out of the turntable.
...because he knows up-front that others will try to duplicate and build on his work. In other words, if you falsify data you will be found out. That is to say, in addition to wasting your time and others, in the long (or even medium) run falsification cannot possibly succeed.
The example I like to think of is a NASA scientist who worked with my advisor when I was a graduate student.
Our project was a suborbital rocket payload that would pop up above the atmosphere just long enough to snap about 300 pictures of the Sun in extreme ultraviolet (which doesn't penetrate the atmosphere); in some ways it was a prototype for the TRACE spacecraft that is in orbit today. Part of the project was finding a photometric calibration for our photographic film. We would expose our film to extreme ultraviolet that we made at a local synchrotron facility under controlled circumstances, then later measure how dark the film got.
This fellow, call him Mr. Electrolux, developed all the film that we calibrated that way -- but he seems to have failed to write down which processes were used on which piece of film. At the end of the day (we think) he simply wrote down a bunch of processes in a plausible order.
The result? Although he didn't get ridiculed at the moment, once we analyzed the data (three to six months later) we realized that we'd flushed nearly $100,000 and six months of our lives down the toilet because the calibration data didn't make sense and we had to redo the work.
I often marvel at the chutzpah/stupidity that goes with such a move: scientists are in the business of figuring out ground truth, and if you lie it will be uncovered.
The point of the 90/90 rule is that complex software development frequently takes nearly twice as long as you think it will. Even if you have a good handle on the vast majority of the things you must implement (and hence can schedule well), you will get tripped up by the remaining few items, which will swell to take as much or more time as the whole rest of the project.
The way to avoid the 90/90 rule is to always pull a Scotty on your bosses -- keep nearly a factor-of-two time margin between what you think you can do and what you commit to.
I suppose we are in technical agreement -- I just don't think that a chip with planned dead zones on it is a lower quality product: it might take slightly more real estate than a "perfect" chip, but the size of the CPU itself is not a major headache -- the package size is determined by the pin count. So long as the circuits that I'm actually using are up to spec, the product is "perfect". But then, I'm no overclocker...
Solar Physics is free to authors but quite expensive to subscribe to. ApJ is expensive to publish in, but is quite cheap to subscribe to (at least for AAS members).
Perhaps in part because of the funding structure, Europeans seem to prefer publishing in Solar Physics while many Americans seem to prefer ApJ. It may have something to do with how science is funded: in the U.S. most of us are on soft money and budget page charges into our grants and/or overhead rates, while in Europe most folks are on fixed departmental budgets. But it's hard to say, because Solar Physics is published in Europe while ApJ is published in North America -- so it may just be the home team advantage in each case.
Uh, dude, this isn't an episode of Transformers, it's a CPU. AMD and Intel already resolved this issue by building very strong chips that don't fail. Even if physically modifying the chip to lop off the bad parts is possible, I can only see it leading to a reduction in quality of chips produced, with manufacturers knowing that worst case, if it fails, it'll just lop itself to pieces.
That is actually an acknowledged way to boost chip yields. Yields of "good" chips are down around the 10-30% level for the most complex chips (e.g. CPUs) -- the manufacturer throws away something like 3-10 chips for every saleable one they make. That number gets worse as you add transistors: with around 200 million transistors on a Pentium 4, a one-in-a-million failure rate would put 200 bad gates on each chip! One way around that problem is to add redundant copies of key systems and then tell the chip itself to choose the more reliable copy. Then yield goes up because you can
tolerate a few bad or weak devices on each chip.
Slashdot Mirror
C'mon, dude, get with the program.
Everyone knows it's
And the next few digits are so don't even try to trot out that old "I just rounded badly" excuse.We get a few laughs out of it, but I suppose we could run a pretty good scam if we wanted to.
That's no small potatoes, for ultraprecision wonks.
Put another way, if you launched a billion tons a year you would have about half of the effect on Earth's rotation as the Moon's measured tidal drag over the last 30 years.
Nah. It only takes one crash to blow away a 10% performance advantage. I'll gladly take the safety nets and the performance hit.
That's no excuse. No OS should ever crash for any software-induced reason, ever. There's a famous story (perhaps in the Jargon File?) about a UNIX system that got half-blown-away by a misplaced "rm -rf /" and was recovered without rebooting. Now that's robust.
Spoken like a true flamethrower! IHBT, but I'll bite anyway. I just installed Mandrake 10 and I'm amazed at the usability -- it's really quite polished.
Drivers are slow to arrive mainly because nearly every single one requires someone to spend a month or two reverse-engineering some proprietary interface. But again, they're not really much of a problem anymore. There are still a few new-ish unsupported devices (the Centrino wireless cards are an example), but the windows compatibility layer takes care of that.
I've tried perhaps a dozen other apps (including YetiSports) and they all worked with no glitches.
The one feature that CrossoverOffice appears not to support well is trashing your whole system with malware -- you can grant your applications read-only access to your main file-system, or limited write access to particular directories.
I think the intended application is not bundling large numbers of NTSC quality movies; the application is shipping more pixels per frame of a single movie.
Further, DLT is expensive and slow compared to platter technology -- the fundamental problem is that hard drives and computers have gotten better much faster than tape drives and the network have.
I'm in the middle of downloading 30 GB of data from one of the SOHO instruments; it will take 3 days to get it over our T1. The only advantage of doing the transfer over the net is that putting it on DVDs for mailing would require somebody on their end to monitor and swap out 6-7 DVDs as they're burned, and then somebody on my end to monitor and swap out those DVDs as they're read onto my hard drive. With a Blu-Ray disk they could burn a single medium then drop it in the mail. And I'd still get the data at the same time as my network transfer will finish.
Hmmm....
Heh. This is kind of fun...
Two of the axes are stable in the sense you're describing; the third (the intermediate axis) is definitely not -- it is at best metastable, because small deviations lead to large excursions of the axis, on the precessional timescale.
One might argue that the minimal axis isn't stable, in the sense that any artificially induced damping causes perturbations to grow, on generally longer timescales than precession. But the maximal axis is definitely stable, in the sense that any physical object isn't perfectly rigid and therefore will gradually damp toward the maximal axis.
If we put up a cylindrical satellite spinning around the maximum moment axis, it will stay that way (or close to it) forever, provided only that the satellite is even a little bit flexible. Interaction with tiny tidal forces and drag can perturb the absolute direction of the rotational axis, but they cannot prevent small sloshes and flexes from gradually working the body itself around so that the maximal axis points along the angular momentum vector.
But, er, none of this is particularly relevant to the SpaceShipOne, eh? It has three quite different moments of inertia, and doesn't stay up long enough for anything but the basic instability of tumbling to matter much. I'd invite you out for a beer, if you weren't probably somewhere else than me entirely.
Nah, only ideal blocks, ideal balls, and ideal disks can do that. Real ones are never perfectly symmetric.
:-)
Your point is well taken -- symmetry can keep the tumbling from being rapid. But sooner or later, everything macroscopic and rigid will precess and/or tumble.
Even if you discount that argument, the intermediate axis (or any degenerate axis) isn't stable -- it's at best metastable.
Hey, pedantry for pedantry!
Ah. Some of the coverage points out that there is a powered trim system for supersonic flight and a straight mechanical system for subsonic flight. Again, I'm speculating -- but I imagine that the control stick just shifts over to running the reaction jets when the air pressure gets low enough -- or maybe it actuates them all the time; they shouldn't affect in-atmosphere flight very much.
Hmmm... In this case, I think the "flight software" might all be resident in the pilot's wetware. (I haven't heard anyone talking about an attitude controller in the official releases...)
Low-Earth orbit is close enough to Earth's surface that you are "at the top of a parabolic trajectory" (actually, in an elliptical trajectory) all the time -- accelerating toward Earth's surface at like 0.99 gee.
Earth's field only gets negligible at distances of about a million miles -- that far out, the Sun's field dominates the local gravitational environment.
wrong.
Most objects do not spin cleanly about most axes. Rigid bodies (such as books, spaceships, rocks, lollipops, and bullets) have three "principal axes" that pass through the center of gravity and are determined by the mass distribution in the object. There's a "minimum" axis that minimizes the kinetic energy for a given angular momentum -- that's the axis around which the thing is the most clustered. For a screwdriver, the minimum axis generally points down the length of the scredriver shaft. There's also a "maximum" axis around which the thing is the most spread out of any direction. For a flat object like a book or a pancake, the maximum axis points directly out of the flat face. Those are the only two axes around which you can spin the object and have it stay stable.
Any other direction will give rise to precession and tumbling, even in vacuum! You can try it with a book -- most closed hardback books have the minimum axis pointing up through the top of the middle pages, and the maximum axis pointing out through the front of the cover. The third dimension -- pointing out through the spine -- is not stable. Tape a book shut and flip it in the air: if you flip it around the maximum or minimum moment axis it will do what you think -- just flip over before you catch it again. If you flip it around the intermediate axis (by, say, starting with the book facing you right-side up with the spine on the left, and pulling the bottom edge toward you as you throw it up in the air) then you might expect the spine to stay on your left side -- but it will flip back and forth, often ending up on your right side, as the book tumbles in the air. (Remember to tape the book closed before tossing it!).
Anyhow, that's a problem for stopping spin and tumbling, because it's not always obvious which way to fire the cold-gas jets to slow down your rotation: by the time you actually fire them you might have tumbled around so that they are speeding you up instead of slowing you down.
I guess that's why "carefree re-entry" is such a great feature of SpaceShipOne -- it's remarkable that they were able to land safely even without good attitude control at apogee.
They're using stuff that's more like 50% pure. If they could get 97% H2O2, they'd be in the running for the X-prize, instead of 8-10 months behind Rutan.
I can't resist here. I've been ripping my LPs with a nice digitizer card and comparing to CDs. The LP sound may be nice and warm, and it's certainly slightly different than the CD from the same album -- but higher fidelity it ain't. To start with, the measured noise floor on my very best LPs is about -40dB. That's after highpass filtering to get rid of the rumble that comes from multiplexing the sound signal with the tracking error (remember, the needle tracks under the influence of small pushes from the record), and also after gating out the 60 Hz and harmonics that are picked up by my magnetic cartridge. And, yes, the system can handle better: making the same measurement with the turntable running and the stylus not touching the record (cue lever engaged) yields more than 10x lower noise floor -- so I'm pretty sure I'm measuring surface quality of the vinyl.
The sounds that come out of the turntable are indeed a little "warmer" than most CDs. For example, I like the smooth, jetstream sounds of Abbey Road side 2 better off the LP than off the CD of the same album. But the CD is definitely cleaner.
Comparatively pure sounds (like flutes) come out measurably distorted: my LPs seem to have harmonic distortion at about the -30dB (0.1%) level in the best cases, based on deconvolving the LP Fourier spectrum using the appropriate CD sample as a comparison (same sound, two media). Some albums are much worse. It's not clear whether the harmonic distortion is due to the records I'm using or due to my turntable, as I can't be bothered to repeat the experiment with someone else's equipment. There's also stereo crosstalk at -15 - -20 dB.
Admittedly, it's no audiophile deck -- I "only" have a belt drive with 2kg turntable and about a $100 cartridge -- but it certainly qualifies as "hifi".
In short -- "good sound" it is, but "higher fidelity" it aint. It's telling that one can't tell the difference between hearing the LP directly off the album or replayed (uncompressed) from my sound card: the 44kHz, 16bit sampling is plenty good to capture whatever's coming out of the turntable.
The example I like to think of is a NASA scientist who worked with my advisor when I was a graduate student.
Our project was a suborbital rocket payload that would pop up above the atmosphere just long enough to snap about 300 pictures of the Sun in extreme ultraviolet (which doesn't penetrate the atmosphere); in some ways it was a prototype for the TRACE spacecraft that is in orbit today. Part of the project was finding a photometric calibration for our photographic film. We would expose our film to extreme ultraviolet that we made at a local synchrotron facility under controlled circumstances, then later measure how dark the film got.
This fellow, call him Mr. Electrolux, developed all the film that we calibrated that way -- but he seems to have failed to write down which processes were used on which piece of film. At the end of the day (we think) he simply wrote down a bunch of processes in a plausible order.
The result? Although he didn't get ridiculed at the moment, once we analyzed the data (three to six months later) we realized that we'd flushed nearly $100,000 and six months of our lives down the toilet because the calibration data didn't make sense and we had to redo the work.
I often marvel at the chutzpah/stupidity that goes with such a move: scientists are in the business of figuring out ground truth, and if you lie it will be uncovered.
The way to avoid the 90/90 rule is to always pull a Scotty on your bosses -- keep nearly a factor-of-two time margin between what you think you can do and what you commit to.
Wouldn't know -- I don't have online access to Nature.
I suppose we are in technical agreement -- I just don't think that a chip with planned dead zones on it is a lower quality product: it might take slightly more real estate than a "perfect" chip, but
the size of the CPU itself is not a major headache -- the package size is determined by the pin count. So long as the circuits that I'm actually using are up to spec, the product is "perfect". But then, I'm no overclocker...
the Astrophysical Journal, published by the University of Chicago Press for the American Astronomical Society. Both of them have
respected peer review systems.
Solar Physics is free to authors but quite expensive to subscribe to. ApJ is expensive to publish in, but is quite cheap to subscribe to (at least for AAS members).
Perhaps in part because of the funding structure, Europeans seem to prefer publishing in Solar Physics while many Americans seem to prefer ApJ. It may have something to do with how science is funded: in the U.S. most of us are on soft money and budget page charges into our grants and/or overhead rates, while in Europe most folks are on fixed departmental budgets. But it's hard to say, because Solar Physics is published in Europe while ApJ is published in North America -- so it may just be the home team advantage in each case.
I tend to alternate between the two.
That is actually an acknowledged way to boost chip yields. Yields of "good" chips are down around the 10-30% level for the most complex chips (e.g. CPUs) -- the manufacturer throws away something like 3-10 chips for every saleable one they make. That number gets worse as you add transistors: with around 200 million transistors on a Pentium 4, a one-in-a-million failure rate would put 200 bad gates on each chip! One way around that problem is to add redundant copies of key systems and then tell the chip itself to choose the more reliable copy. Then yield goes up because you can tolerate a few bad or weak devices on each chip.