Here it is, the official Billboard Top 10 for this week. This is what the RIAA wants you to listen to. These are the "hits" they are complaining aren't selling enough copies.
Confessions Part II, Usher
Slow Motion, Juvenile Featuring Soulja Slim
Burn, Usher
The Reason, Hoobastank
If I Ain't Got You, Alicia Keys
Move Ya Body, Nina Sky Featuring Jabba
Lean Back, Terror Squad
On Fire, Lloyd Banks
Turn Me On, Kevin Lyttle Featuring Spragga Benz
Freek-A-Leek, Petey Pablo
Now, is there anything on that list you're going to want to listen to next year?
Any questions?
What's really happening is that CD sales are up, but hits burn out faster than they used to. "Time on the chart" has shortened. That needs to be mentioned to rebut the RIAA's arguments.
The ADA runs into First Amendment issues in the US.
Requiring "accessability" implies "coerced speech". The US can't require that a book have a Braille edition to be published, for example.
Non-commercial, non-Government content thus has First Amendment protection.
Congress could require ads, which are "commercial speech" to be "accessable". Commercial speech can have coerced components (big type, disclamers, etc.), because that's part of "commerce", which Congress can regulate.
I'm always amused that you have to break out of Perl into C to parse HTML and XML. And even more amused that it doesn't bother the Perl fanatics. What's wrong with this picture?
I really should upload my XML/SGML/HTML parser to CPAN. It's all in Perl, and uses HTML::Element as the representation. It's written for robustness, not speed; it can parse ill-formed XML/SGML/HTML, producing good trees for the valid parts.
This is necessary to parse the material that appears in SEC filings.
That's not too impressive. They're not even close to active stabilization. All they're doing is rotating a weight that also moves in and out from the axis of rotation to change the moment of inertia. This gets them a little net angular motion. Big deal. I've seen wind-up toys do that.
Since it's a one-axis device, there's no need to test it in a zero-G environment. Hanging it from a string would work equally well.
There's useful work to be done on three-axis stablization algorithms, but this isn't it.
LISP is a good match for operating on HTML and XML, both of which are really tree structures. Operating on trees works very well in LISP. That's what it's good at. Perl, PHP, and Java don't do trees well. You have to hammer the tree into an object paradigm, which doesn't help all that much. Perl's representation of a tree is rather inefficient, too.
I do considerable parsing of large documents into trees in Perl. It works, it's portable, it's slow, and Perl is badly matched to the task. PHP is worse.
LISP's parentheses turn everybody off, including me, but the data structures really are a win for tree-like applications.
It's likely that some of these holes have already been quietly exploited. The script kiddies sending out viruses in bulk get all the press, but those aren't the real attackers. It's the people who quietly break into machines, find out what they want to know, and leave no trace who are the real threat.
DoD security people emphasize this. It's not the kids throwing rocks at the fence that are the real threat. It's the guy who finds out what tonight's target is and tells the enemy forces.
In the business world, you might never know if someone has seen the financial data you're releasing tomorrow and then shorted the stock.
That kind of info is far more valuable than credit card numbers.
Roland the Self-Promoting Blogger really fell for this one. I've seen better stuff advertised in spam. This guy is becoming a bottom-feeding George Gilder.
Then again, Slashdot's "editors" accepted the "story".
For a good laugh, read the developer page. "The main component of ACTIL OS is its System Functions Library, whose call syntax (i.e. function names and list of arguments - variables, parameters, and so on) and comments have the same format as the functions in Microsoft's VISUAL C++ library." Wow.
It's a per-registrar thing. I looked up domains registered via Melbourne IT and DomainSite, and they don't show a registrar. This is because their Whois output doesn't include the registrar name.
GoDaddy plugs itself quite a bit in its own Whois output, so you see GoDaddy.
Verisign isn't showing the registrar link and the registrar name from the common registry database.
Note what it doesn't mention - the registrar. The real Whois data has the name of and a link to the registrar. This Verisign thing totally hides the registrar. That, presumably, is the point.
You have to click on "underlying Whois data" to get the registrar info. At least you still can.
The military has a more realistic system called MILES, which they use for war games. It works much like laser tag, but their rules are tougher.
MILES uses real weapons with blank rounds. The MILES laser transmitter clamps into the barrel, so if you do manage to load a live round, you destroy the transmitter and the weapon, but not your target. The "bang" of the blank round triggers the laser transmitter. So you have to lug ammo
and magazines around. All the real-world problems of jams and misfires occur, too.
If you're hit, it beeps. Loudly. Continously. And you can't turn it off. Only a referee can turn it off.
If you're hit, you're dead. You're carried off to the "dead" pen. Often, becoming "dead" means an extra 20-mile march or some similar unpleasant detail.
In the newer versions, beams are coded, and you can tell who hit whom. Soldiers who miss too much get sent to the rifle range for extra training.
Scores affect your real-life Army career. Why send losers to war?
The latest generation gear uses GPS and data links so that indirect fire weapons can be simulated. But you probably don't need that.
Col. (now Brig. Gen, retd.) Paul Tibbets is still alive. He's doing fine; he was flying until at least 1998. He has a web site., from which you can order a detailed model of the Hiroshima bomb personally signed by Gen Tibbets.
What we need is a registry of online merchants
on
An Online ID Registry
·
· Score: 2, Informative
This solves a problem we don't really have, which is why the last five or so attempts to solve it haven't gone anywhere.
What we need is a solid way to identify everyone who takes credit cards on the Internet, to help deal with spammers. It's a crime in many areas (California, for one) to run an anonymous business. California requires that the actual name and address of the business (not a P.O. box, unless you file some extra paperwork) be shown to the customer before the site accepts a credit card number. So it's not controversial to require this. It just needs a better implemention.
What we need is a banking regulation requirement that when a credit card merchant bank accepts a credit card transaction, there's a check at the bank's payment gateway of the web page from which the transaction came. The page must be SSL, of course. Its certificate information should be validated agains the ownership info for the merchant's bank account
The credit card transaction (merchant to bank) should be signed with the same key that signs the web page. Otherwise, the bank is required to reject the transaction.
This requires zero consumer-side changes. It makes it much easier to figure out who to blame for spam. Just get to the payment page and read the certificate. Right now, most SSL certificates don't guarantee anything. This forces accurate info into the site's certificate, or the transaction bounces.
It would be a pain for companies that rely on "affilate networks" and other marginal indirect payment schemes. But that's probably a good thing.
I use various names ending in "-replies" at "downside.com" for everything. The address I use on Slashdot, "slashdot-replies", is harvested frequently.
Just got one from OptInRealBig, clearly establishing that they use harvested E-mail addresses. "Opt-in", right.
That calculation is off by a factor of 5, but the numbers are still huge.
First, let's look at the Maxwell BCAP0010 Ultracapacitor. 2600F, 2.5V, 525g, 60mm diameter cylinder, 172mm length. Incidentally, of these can deliver 600A for 5 seconds, if you need that much power all at once. These aren't like those high-resistance supercapacitors used to keep computer clocks alive with a trickle. Ultracapacitors can deliver serious current. Six of these can start an auto engine.
This is about 5000 farads per kilogram, at 2.5V.
E = C*V*V*0.5, which is 15625 joules.
Average power for a capacitor is P = C*V*V*0.5/T, so if we need 12 gigawatts for
10 minutes, we get 12,000,000,000=C*2.5*2.5*0.5/600, or
C=2.304 terafarads. That's two billion capacitors.
You can get about 30 of those in a 4U enclosure (Siemens sells an ultracapacitor bank in that form),
maybe 300 in a rack, so this needs 700,000 racks of capacitors.
This isn't going to be a low-cost approach. Sure, it's modular, but you need a lot of modules.
That's an old paper, from 1972, republished by a company that (surprise!) makes tube audio gear.
This whole phenomenon is well understood today. You can buy a little "tube amp emulator", with emulations for famous tube amps. Choose your own harmonic distortion. There are product lines of amp modellers.
Most of the trouble in audio today is not tube vs. transistor vs. digital. It's from artifacts introduced during compression of the dynamic range. The real problem is the car audio listening environment, which is noisy.
Radio stations need to sound good in cars.
This led radio stations to compress their audio into a narrow dynamic range. People got used to this. Then, when cars got CD players, CD mixes began to be compressed like car audio.
("You don't want your record to be the softest one in the changer"). Now, most popular music is so compressed that musicians have totally lost the musical use of volume. You can't have a soft passage; it will be pumped up. Sharp attacks are clipped, so that tool has been taken away.
The end result is popular music that has no texture. Background music.
Didn't Roland the Plogger post this already?
on
Data Mining Goes 3D
·
· Score: 1
Adding what passes for "machine learning" to a user interface usually results in something that does the right thing some of the time, the wrong thing some of the time, and you can't figure out why.
Bayesian spam filtering is becoming like that. At first it worked, but it's breaking down under the rising percentage spam.
Laser launch is a nice idea, but the power requirements are huge. The current altitude record is 71 meters (not kilometers), with a 51 gram (not Kg) craft and a 10 kilowatt laser.
Kare, who's been plugging this idea for decades, writes
"A rule of thumb for
laser launchers is that the unit payload is 1 kg per MW of laser power." The Apollo lunar module (all the stuff that went to the moon) massed about 6500 Kg, of which 2500Kg made the round trip. So we're going to need several gigawatts of laser power for a moon shot.
Kare is talking about using continuous diode lasers in the 1KW range. These don't exist, but 60W units are available, so this isn't totally unreasonable. Kare proposes to use maybe 150 of these future 1KW units in a prototype. That only launches a 150g craft.
Launching something the size of the Apollo lunar module would take six million such units, and about 12 gigawatts of electrical power for several minutes. This is twice the power output of Grand Coulee Dam, the biggest single power source in the US.
The power storage problem might be overcome using ultracapacitors. You can get 2600 farad capacitors (not ufd, farads) at 2.5V today, and you can take current out fast. Auto engines can be started with six of these things, weighing a total of about 3Kg. With a big budget, a laser launch system could have enough energy storage to do the job.
Six million lasers, though, is a bit much. The prototype doesn't put enough mass in orbit to be useful, and the real version is too big.
If you want to launch a microsat, you call Orbital Sciences Corporation, and they launch a Pegasus rocket from a L-1011 for you. The X-prize guys get all the press, but Orbital actually puts stuff in orbit. They've launched 45 payloads so far. Click here for their user manual.
"Thank you for calling Friendly Senior Services. Your call is important to us. All attendants are currently busy helping other callers. Please stay on the line, and an attendant will be with you shortly. If this is an emergency, hang up and dial 911".
The real problem was that programming on 16-bit x86 machines always sucked. UNIX, OS/2, and Microsoft Windows were all crammed into that wierd addressing space.
The really amazing thing is that Microsoft managed to make the horror of 16-bit Windows work as well as it did.
The reason why MS is potentially a big deck of cards is that they consistently shove things down peoples' throats and therefore never get to see what they *would* choose if they had the choice.
One can say that of Windows and IE. But Office, where Microsoft makes its money, won out in a crowded field. Gates once said, of how Office began, "We asked developers to develop for Windows, and they said no. So we asked Microsoft's Application Division, and they didn't have that option." Many of Microsoft's competitors in office-type programs stayed with DOS too long. Lotus (of Lotus 1-2-3, not Notes) was bigger than Microsoft until the early 1990s.
Today, Office is where Microsoft makes its big money. Windows makes some money, and everything else (XBox, MSN, tools. etc. loses money). The real threat to Microsoft is not Linux. It's OpenOffice.
Why would anyone want to migrate to the Inanium at this late date? Now that Intel finally caved and cloned AMD's 64-bit machines, the Itanium is clearly on the way out. That's not where you want to be three years from now. The Itanium is headed for the Intel scrap heap of wierd processors, along with the i860, the i960, and the iapx432. All of which were architecturally better than x86.
Slashdot Mirror
Now, is there anything on that list you're going to want to listen to next year?
Any questions?
What's really happening is that CD sales are up, but hits burn out faster than they used to. "Time on the chart" has shortened. That needs to be mentioned to rebut the RIAA's arguments.
Non-commercial, non-Government content thus has First Amendment protection. Congress could require ads, which are "commercial speech" to be "accessable". Commercial speech can have coerced components (big type, disclamers, etc.), because that's part of "commerce", which Congress can regulate.
I really should upload my XML/SGML/HTML parser to CPAN. It's all in Perl, and uses HTML::Element as the representation. It's written for robustness, not speed; it can parse ill-formed XML/SGML/HTML, producing good trees for the valid parts. This is necessary to parse the material that appears in SEC filings.
Since it's a one-axis device, there's no need to test it in a zero-G environment. Hanging it from a string would work equally well.
There's useful work to be done on three-axis stablization algorithms, but this isn't it.
Even NASA didn't go for this one.
With a few slight mods to the screen formats, the Online Herd Management System should be applicable to schools.
LISP's parentheses turn everybody off, including me, but the data structures really are a win for tree-like applications.
DoD security people emphasize this. It's not the kids throwing rocks at the fence that are the real threat. It's the guy who finds out what tonight's target is and tells the enemy forces.
In the business world, you might never know if someone has seen the financial data you're releasing tomorrow and then shorted the stock. That kind of info is far more valuable than credit card numbers.
Then again, Slashdot's "editors" accepted the "story".
For a good laugh, read the developer page. "The main component of ACTIL OS is its System Functions Library, whose call syntax (i.e. function names and list of arguments - variables, parameters, and so on) and comments have the same format as the functions in Microsoft's VISUAL C++ library." Wow.
Verisign isn't showing the registrar link and the registrar name from the common registry database.
You have to click on "underlying Whois data" to get the registrar info. At least you still can.
Registrars are going to be annoyed about this.
-
MILES uses real weapons with blank rounds. The MILES laser transmitter clamps into the barrel, so if you do manage to load a live round, you destroy the transmitter and the weapon, but not your target. The "bang" of the blank round triggers the laser transmitter. So you have to lug ammo
and magazines around. All the real-world problems of jams and misfires occur, too.
-
If you're hit, it beeps. Loudly. Continously. And you can't turn it off. Only a referee can turn it off.
-
If you're hit, you're dead. You're carried off to the "dead" pen. Often, becoming "dead" means an extra 20-mile march or some similar unpleasant detail.
-
In the newer versions, beams are coded, and you can tell who hit whom. Soldiers who miss too much get sent to the rifle range for extra training.
-
Scores affect your real-life Army career. Why send losers to war?
The latest generation gear uses GPS and data links so that indirect fire weapons can be simulated. But you probably don't need that.Col. (now Brig. Gen, retd.) Paul Tibbets is still alive. He's doing fine; he was flying until at least 1998. He has a web site., from which you can order a detailed model of the Hiroshima bomb personally signed by Gen Tibbets.
What we need is a solid way to identify everyone who takes credit cards on the Internet, to help deal with spammers. It's a crime in many areas (California, for one) to run an anonymous business. California requires that the actual name and address of the business (not a P.O. box, unless you file some extra paperwork) be shown to the customer before the site accepts a credit card number. So it's not controversial to require this. It just needs a better implemention.
What we need is a banking regulation requirement that when a credit card merchant bank accepts a credit card transaction, there's a check at the bank's payment gateway of the web page from which the transaction came. The page must be SSL, of course. Its certificate information should be validated agains the ownership info for the merchant's bank account The credit card transaction (merchant to bank) should be signed with the same key that signs the web page. Otherwise, the bank is required to reject the transaction.
This requires zero consumer-side changes. It makes it much easier to figure out who to blame for spam. Just get to the payment page and read the certificate. Right now, most SSL certificates don't guarantee anything. This forces accurate info into the site's certificate, or the transaction bounces.
It would be a pain for companies that rely on "affilate networks" and other marginal indirect payment schemes. But that's probably a good thing.
I use various names ending in "-replies" at "downside.com" for everything. The address I use on Slashdot, "slashdot-replies", is harvested frequently. Just got one from OptInRealBig, clearly establishing that they use harvested E-mail addresses. "Opt-in", right.
First, let's look at the Maxwell BCAP0010 Ultracapacitor. 2600F, 2.5V, 525g, 60mm diameter cylinder, 172mm length. Incidentally, of these can deliver 600A for 5 seconds, if you need that much power all at once. These aren't like those high-resistance supercapacitors used to keep computer clocks alive with a trickle. Ultracapacitors can deliver serious current. Six of these can start an auto engine.
This is about 5000 farads per kilogram, at 2.5V. E = C*V*V*0.5, which is 15625 joules.
Average power for a capacitor is P = C*V*V*0.5/T, so if we need 12 gigawatts for 10 minutes, we get 12,000,000,000=C*2.5*2.5*0.5/600, or C=2.304 terafarads. That's two billion capacitors. You can get about 30 of those in a 4U enclosure (Siemens sells an ultracapacitor bank in that form), maybe 300 in a rack, so this needs 700,000 racks of capacitors.
This isn't going to be a low-cost approach. Sure, it's modular, but you need a lot of modules.
This whole phenomenon is well understood today. You can buy a little "tube amp emulator", with emulations for famous tube amps. Choose your own harmonic distortion. There are product lines of amp modellers.
Most of the trouble in audio today is not tube vs. transistor vs. digital. It's from artifacts introduced during compression of the dynamic range. The real problem is the car audio listening environment, which is noisy. Radio stations need to sound good in cars. This led radio stations to compress their audio into a narrow dynamic range. People got used to this. Then, when cars got CD players, CD mixes began to be compressed like car audio. ("You don't want your record to be the softest one in the changer"). Now, most popular music is so compressed that musicians have totally lost the musical use of volume. You can't have a soft passage; it will be pumped up. Sharp attacks are clipped, so that tool has been taken away. The end result is popular music that has no texture. Background music.
Didn't we see this article before?
Adding what passes for "machine learning" to a user interface usually results in something that does the right thing some of the time, the wrong thing some of the time, and you can't figure out why.
Bayesian spam filtering is becoming like that. At first it worked, but it's breaking down under the rising percentage spam.
Kare, who's been plugging this idea for decades, writes "A rule of thumb for laser launchers is that the unit payload is 1 kg per MW of laser power." The Apollo lunar module (all the stuff that went to the moon) massed about 6500 Kg, of which 2500Kg made the round trip. So we're going to need several gigawatts of laser power for a moon shot.
Kare is talking about using continuous diode lasers in the 1KW range. These don't exist, but 60W units are available, so this isn't totally unreasonable. Kare proposes to use maybe 150 of these future 1KW units in a prototype. That only launches a 150g craft.
Launching something the size of the Apollo lunar module would take six million such units, and about 12 gigawatts of electrical power for several minutes. This is twice the power output of Grand Coulee Dam, the biggest single power source in the US.
The power storage problem might be overcome using ultracapacitors. You can get 2600 farad capacitors (not ufd, farads) at 2.5V today, and you can take current out fast. Auto engines can be started with six of these things, weighing a total of about 3Kg. With a big budget, a laser launch system could have enough energy storage to do the job.
Six million lasers, though, is a bit much. The prototype doesn't put enough mass in orbit to be useful, and the real version is too big.
If you want to launch a microsat, you call Orbital Sciences Corporation, and they launch a Pegasus rocket from a L-1011 for you. The X-prize guys get all the press, but Orbital actually puts stuff in orbit. They've launched 45 payloads so far. Click here for their user manual.
"Thank you for calling Friendly Senior Services. Your call is important to us. All attendants are currently busy helping other callers. Please stay on the line, and an attendant will be with you shortly. If this is an emergency, hang up and dial 911".
The real problem was that programming on 16-bit x86 machines always sucked. UNIX, OS/2, and Microsoft Windows were all crammed into that wierd addressing space.
The really amazing thing is that Microsoft managed to make the horror of 16-bit Windows work as well as it did.
One can say that of Windows and IE. But Office, where Microsoft makes its money, won out in a crowded field. Gates once said, of how Office began, "We asked developers to develop for Windows, and they said no. So we asked Microsoft's Application Division, and they didn't have that option." Many of Microsoft's competitors in office-type programs stayed with DOS too long. Lotus (of Lotus 1-2-3, not Notes) was bigger than Microsoft until the early 1990s.
Today, Office is where Microsoft makes its big money. Windows makes some money, and everything else (XBox, MSN, tools. etc. loses money). The real threat to Microsoft is not Linux. It's OpenOffice.
Why would anyone want to migrate to the Inanium at this late date? Now that Intel finally caved and cloned AMD's 64-bit machines, the Itanium is clearly on the way out. That's not where you want to be three years from now. The Itanium is headed for the Intel scrap heap of wierd processors, along with the i860, the i960, and the iapx432. All of which were architecturally better than x86.
Hello? Browsers must not execute arbitrary programs on client machines. Is there anybody who doesn't get this yet?
And why aren't we running browsers in jails yet, anyway?
200 MPH winds at 0.01 atmosphere aren't that big a deal. There's some dust. Here are Viking lander pictures taken during a Martian dust storm in 1976.