So, explain to me again why this particular non-computable number is special?
One difference is, W_UTM a probability of a well-defined event. So it's a number that is "meaningful" in its own right; the number with digit n 0 or 1 depending as TM(n) halts or not is pretty cooked by comparison. Therefore it's arguably slightly more interesting that it should be random.
(Compare to Godel's unprovable sentence; it's not really a very "natural" number-theoretic statement. It is arguably more interesting to come up with a more usual-looking statement that is independent of the Peano axioms.)
There may be more to it than that, but Chaitin seems like such a blowhard in that talk transcript that I kind of hope not:).
By the way, I'd like to hear more about cruising the streets with a sniffer looking for open networks. How's that done, and what does one do to 'lock down' the network?
It depends on what wireless technology you're using, but here in my own private geek compound I run Orinoco (Lucent) Wavelan Gold wireless cards in 128-bit RC4 encryption mode.
This is quite easy to set up under Linux using the
wireless extensions to the standard
pcmcia services . You will have a switch branch in your wireless.opts file that looks something like
# Default Lucent Wavelan IEEE
# Note : wvlan_cs driver only,
# and version 1.0.4+ for encryption support
*,*,*,00:60:1D:*)
INFO="Wavelan IEEE ad-hoc"
ESSID="Secure Network"
MODE="Ad-hoc"
CHANNEL="3" #2.422GHz
RATE="auto"
KEY="1234-5648-9abc-def1-2345-6789-ab" ;;
(No, that's not my actual key:). And note it's not the full 128 bits... the version of the drivers I have won't permit that, for some reason that I don't understand. But 104 bits is pretty good.)
Anyway, you definitely want to "lock down" your network, unless you are into to providing a public access point. Without encryption, it would be like having a hub on my DSL modem that anybody driving by could plug in to...
There are 6 that are much brighter than the rest. It's not really obvious which is the seventh brightest. Look at the constellation through a telescope, or even a good pair of binoculars, sometime.
Entropia has already issued a press release announcing that they have contracted to do
just about the same thing (distributed web application performance monitoring).
There are several players in this space now that have some serious VC backing and it will be interesting to see how it shakes out. I'm no expert, but it seems to me that an Open Source / Free Software play here could really be interesting about now.
Right, Rambus's business model is primarily to own technology patents, and get a revenue stream from licensing them; only secondarily to actually produce any instances of the technology.
And arguably Qualcomm's patents were issued in the face of prior art as well; CDMA had been used in military communications for some time. The U.S. isn't about to "fix" this, since it's a place that U.S. companies have an edge.
Welcome to 21st century high-tech big business.
And by the way, Taco, "beg the question" doesn't mean "invite the question". If you had taken a philosophy course at Hope College you would know that:).
With all that massive programming talent in-house, why is it so many M$ products come out all lame and bloated?
Because M$ is essentially only competing with itself; that's the essence of monopoly.
And the massive programming talent (the largest computer science department of any institution in the world, right now) is hired to make sure it stays that way: that brainpower doesn't need to make anything for M$, as long as it is tied up not making anything for anybody else.
Of course, even if the published specifications are complete and consistent enough to let you write your own implementation (believe me, I haven't tried), being able to call the result "Java(tm)" is something else entirely.
Nice configuration there, and it seems to work nice too. Debian instead of FreeBSD might be a surprise, to someone who still doesn't get that VA is a Linux-only shop; but whatever.
My question is, why the big deal ArrowPoint switch? That's a $15,000 unit, and it looks like all you are doing with it is firewalling and load balancing a 100Mbit pipe across 6 web servers. Sure seems to me another VA Debian box could do that for a whole lot less.
Of course ArrowPoint is in the same building as Andover.net, and Exodus is a big ArrowPoint customer. But those aren't reasons to use an expensive, closed solution when an open one would do? Are they?
Scott McNealy is the best golfer among CEOs of publicly traded companies; or so they say. And that's saying a lot, considering how seriously those suits take golf. They're jocks, not nerds...
No, Roblimo, those images only took a long time to load when you were the only one in the western hemisphere loading them.
Now they are taking forever to load, since you posted the link to that poor unsuspecting site. Man, I wish you guys would just quit doing that without organizing some mirrors first.
Did I say consumer-level? I was speaking on principle.
Well, since we live in a physical universe, and you don't run your analog electronics at absolute zero, they are going to introduce noise. In principle. With a digital data stream, you can design the system so the noise has no effect. You can't do that with analog.
Anything else processing the sound in between, digital or analog, is just going to add noise. Right. And when you're talking about going through the number of stages required to record on commercially available media, digital can easily be designed to introduce less noise than analog. Most audiophiles don't care about the poor signal to noise ratio of analog, though. They rather seem to like it. Never developed a taste for it myself.
If the original sound source has two signal components of 60kHz and 65kHz, there will be a tertiary tone of 5kHz as a result of the other two being superimposed. Is that 5kHz tone sampled successfully with a 44.1kHz sampling rate?
The 5kHz beat would be a result of nonlinear mixing of the components, for example in the cochlea or in neural processing. Tranmission of sound through air is very linear at normal SPLs. If you have a linear analog system with rolloff well below 60kHz (i.e. a good vinyl recording and playback system) the components won't be there in the output and you won't hear the 5kHz beat. If you are doing digital at 44.1kHz sampling, you will want to rolloff the signal above 22kHz, so the components won't be in the output there either.
If you had a multiplicative nonlinearity in the input stage, you could generate the 5kHz beat there, and record it. Some people might even like that.
Think about what "analog" means. You store a signal on one medium that is a direct analog of the recorded signal.
What universe do you live in? In mine, you can't do that. The best you can do is store a signal that is an analog of the source, plus unavoidable noise and distortion.
However, with digital, you're taking repeated samples, and approximating each sample to the nearest quantized level determined by the bit depth. So you lose some quality converting to digital.
And this quantization noise can easily be less than the noise introduced in going to an analog medium. Easily. What consumer-available analog medium has better SNR than you get with 16-bit quantization?
That doesn't change the fact that you can't make a true 10khz sine wave on a CD (roughly 4 sample points per cycle, and you actually have a sawtooth wave that phases in and out w/ the sample rate).
Please, learn Nyquist's theorem. Audio is bandlimited. You can reconstruct it perfectly from discrete samples. No "sawtooth wave" at all.
As has been pointed out elsewhere... If these coordinated DoS attacks use compromised boxes, they don't care about IP spoofing. The boxes don't belong to the perps anyway.
You don't even need to root the boxes to do this (user space compromise is enough). Spoofing packets is beside the point.
Why is it cool to have only computer purchases included in these tax holidays?
The nice thing about an across-the-board sales tax is that it doesn't distort the economy; it just applies a small uniform drag to everything. Now the Governor of Pennsylvania is going to artificially assist the retail PC industry (relative to everything else). As if computer-related industries hadn't got enough artificial assistance in the last few years.
That suggests to me that the lawyers for the defendants are doing a poor job.
Right, and I hope that someone from the EFF, or someone who knows the details of the EFF efforts here (Bruce Perens?) can explain to us how this can happen.
We've been hearing about the crack team of lawyers the EFF had on the case, and how we should all donate to the EFF, etc. And then they show what appears to be a remarkable level of incompetence in this important case.
So my question is: Why should we continue to donate the EFF? Or, to whom should we donate? Some of us with checkbooks but no time would like to know the answer...
When those networks sell advertising, it's with the understanding from their customers that they will have 30 seconds of captive audience.
30 seconds times how many viewers?
That's what advertisers care about.
And is there any reason to think that people are going to watch iCraveTV instead of watching the same thing on their TV? I sure wouldn't! (I don't have a TV:)
So the advertisers are getting more than what they paid for. So they should be happy. So the broadcasters should be happy.
You can't run a site that sponges off other people's (very very large) investments for your own personal gain without expecting the legitimate owners of that signal to get a little peeved.
But making extra money off their stuff is not at all the same thing as taking money they would have made off their stuff. I really don't see that iCraveTV is hurting the sources of the material in any way. So I conclude they are peeved for no good reason. --Seen
If you look at his official bio pic you'll see he's not totally bald. He has those little tufts of hair on the sides over the ears. Yes, friends, he just needs to grow them out... make them more, you know, pointy! Then we'll be all set.
Actually, wait another minute. How can you pipe 1.5Mb/s over AM frequencies, anyways? Don't you need 2x frequecy to send x b/s? Doesn't Nyquist theorem state this?
No, because in general bits/s != samples/s. Bits per second equals cycles per second TIMES BITS PER SAMPLE. You can typically get a lot more than one bit per sample, depending on the resolution of your A/D converter and the effective signal to noise ratio.
Put it another way: The Nyquist theorem is about discrete sampling of an analog waveform: the samples themselves are analog values. When you go further and digitize the samples, now you can start talking about bits per second.
Slashdot Mirror
One difference is, W_UTM a probability of a well-defined event. So it's a number that is "meaningful" in its own right; the number with digit n 0 or 1 depending as TM(n) halts or not is pretty cooked by comparison. Therefore it's arguably slightly more interesting that it should be random.
(Compare to Godel's unprovable sentence; it's not really a very "natural" number-theoretic statement. It is arguably more interesting to come up with a more usual-looking statement that is independent of the Peano axioms.)
There may be more to it than that, but Chaitin seems like such a blowhard in that talk transcript that I kind of hope not :).
--Seen
For your bedtime reading, y'all.
--Seen
But if you do a Google search on "ginger dean kamen", you get nothing. Not even any wack rumors. Deja doesn't turn up anything either.
So, it's got me curious, which is a pretty good PR trick if nothing else.
--Seen
It depends on what wireless technology you're using, but here in my own private geek compound I run Orinoco (Lucent) Wavelan Gold wireless cards in 128-bit RC4 encryption mode.
This is quite easy to set up under Linux using the wireless extensions to the standard pcmcia services . You will have a switch branch in your wireless.opts file that looks something like
;; :). And note it's not the full 128 bits... the version of the drivers I have won't permit that, for some reason that I don't understand. But 104 bits is pretty good.)
# Default Lucent Wavelan IEEE
# Note : wvlan_cs driver only,
# and version 1.0.4+ for encryption support
*,*,*,00:60:1D:*)
INFO="Wavelan IEEE ad-hoc"
ESSID="Secure Network"
MODE="Ad-hoc"
CHANNEL="3" #2.422GHz
RATE="auto"
KEY="1234-5648-9abc-def1-2345-6789-ab"
(No, that's not my actual key
Anyway, you definitely want to "lock down" your network, unless you are into to providing a public access point. Without encryption, it would be like having a hub on my DSL modem that anybody driving by could plug in to...
--Seen
--Seen
There are several players in this space now that have some serious VC backing and it will be interesting to see how it shakes out. I'm no expert, but it seems to me that an Open Source / Free Software play here could really be interesting about now.
--Seen
They're not the only one. For example, Qualcomm is doing the same thing, with considerable success.
And arguably Qualcomm's patents were issued in the face of prior art as well; CDMA had been used in military communications for some time. The U.S. isn't about to "fix" this, since it's a place that U.S. companies have an edge. Welcome to 21st century high-tech big business.
And by the way, Taco, "beg the question" doesn't mean "invite the question". If you had taken a philosophy course at Hope College you would know that :).
--Seen
Because M$ is essentially only competing with itself; that's the essence of monopoly.
And the massive programming talent (the largest computer science department of any institution in the world, right now) is hired to make sure it stays that way: that brainpower doesn't need to make anything for M$, as long as it is tied up not making anything for anybody else.
--Seen
Of the 40, only 5 dropped out or didn't go to college. Michael Jordan was one of the 5.
--Seen
OK, well this is for Hemos's birthday, so the mind boggles. How about a 10/100 Ethernet option?
Or the Beowulf clustering option...
--Seen
Obsidian knives have been around for thousands of years. Not quite as long as meteorites, but hey.
It's probably where Stephenson got the idea... it fits with the picture of Raven as master of pre-metaverse aboriginal traditions.
--Seen
Of course, even if the published specifications are complete and consistent enough to let you write your own implementation (believe me, I haven't tried), being able to call the result "Java(tm)" is something else entirely.
--Seen
My question is, why the big deal ArrowPoint switch? That's a $15,000 unit, and it looks like all you are doing with it is firewalling and load balancing a 100Mbit pipe across 6 web servers. Sure seems to me another VA Debian box could do that for a whole lot less.
Of course ArrowPoint is in the same building as Andover.net, and Exodus is a big ArrowPoint customer. But those aren't reasons to use an expensive, closed solution when an open one would do? Are they?
No troll here, just curious.
--Seen
--Seen
Now they are taking forever to load, since you posted the link to that poor unsuspecting site. Man, I wish you guys would just quit doing that without organizing some mirrors first.
What the heck, here's another link (don't want to call it a mirror, since it's the original): http://www.nasa.gov/newsinfo/srtm_images.html
--Seen
Well, since we live in a physical universe, and you don't run your analog electronics at absolute zero, they are going to introduce noise. In principle. With a digital data stream, you can design the system so the noise has no effect. You can't do that with analog.
Anything else processing the sound in between, digital or analog, is just going to add noise. Right. And when you're talking about going through the number of stages required to record on commercially available media, digital can easily be designed to introduce less noise than analog. Most audiophiles don't care about the poor signal to noise ratio of analog, though. They rather seem to like it. Never developed a taste for it myself.
If the original sound source has two signal components of 60kHz and 65kHz, there will be a tertiary tone of 5kHz as a result of the other two being superimposed. Is that 5kHz tone sampled successfully with a 44.1kHz sampling rate?
The 5kHz beat would be a result of nonlinear mixing of the components, for example in the cochlea or in neural processing. Tranmission of sound through air is very linear at normal SPLs. If you have a linear analog system with rolloff well below 60kHz (i.e. a good vinyl recording and playback system) the components won't be there in the output and you won't hear the 5kHz beat. If you are doing digital at 44.1kHz sampling, you will want to rolloff the signal above 22kHz, so the components won't be in the output there either.
If you had a multiplicative nonlinearity in the input stage, you could generate the 5kHz beat there, and record it. Some people might even like that.
--Seen
What universe do you live in? In mine, you can't do that. The best you can do is store a signal that is an analog of the source, plus unavoidable noise and distortion.
However, with digital, you're taking repeated samples, and approximating each sample to the nearest quantized level determined by the bit depth. So you lose some quality converting to digital.
And this quantization noise can easily be less than the noise introduced in going to an analog medium. Easily. What consumer-available analog medium has better SNR than you get with 16-bit quantization?
That doesn't change the fact that you can't make a true 10khz sine wave on a CD (roughly 4 sample points per cycle, and you actually have a sawtooth wave that phases in and out w/ the sample rate).
Please, learn Nyquist's theorem. Audio is bandlimited. You can reconstruct it perfectly from discrete samples. No "sawtooth wave" at all.
--Seen
You don't even need to root the boxes to do this (user space compromise is enough). Spoofing packets is beside the point.
--Seen
The nice thing about an across-the-board sales tax is that it doesn't distort the economy; it just applies a small uniform drag to everything. Now the Governor of Pennsylvania is going to artificially assist the retail PC industry (relative to everything else). As if computer-related industries hadn't got enough artificial assistance in the last few years.
--Seen
Right, and I hope that someone from the EFF, or someone who knows the details of the EFF efforts here (Bruce Perens?) can explain to us how this can happen.
We've been hearing about the crack team of lawyers the EFF had on the case, and how we should all donate to the EFF, etc. And then they show what appears to be a remarkable level of incompetence in this important case.
So my question is: Why should we continue to donate the EFF? Or, to whom should we donate? Some of us with checkbooks but no time would like to know the answer...
--Seen
30 seconds times how many viewers?
That's what advertisers care about.
And is there any reason to think that people are going to watch iCraveTV instead of watching the same thing on their TV? I sure wouldn't! (I don't have a TV :)
So the advertisers are getting more than what they paid for. So they should be happy. So the broadcasters should be happy.
You can't run a site that sponges off other people's (very very large) investments for your own personal gain without expecting the legitimate owners of that signal to get a little peeved.
But making extra money off their stuff is not at all the same thing as taking money they would have made off their stuff. I really don't see that iCraveTV is hurting the sources of the material in any way. So I conclude they are peeved for no good reason.
--Seen
But you can't get both Java2 and Linux together.
(At least, that I can see; but I find the VisualAge site a pain to navigate around so maybe I missed something.)
The "Early Adopters" Java2 stuff seems to just be for Windows at this point.
If you have a URL for a Linux Java2 download from IBM, please post it! Profuse thanks.
--Seen
--Seen
- Thinks parents are "lusers on both sides": check.
- Listens to Frank Zappa, but doesn't think he's talking about US: check.
- Wants uncompromising change: check.
- Idealistically thinks writing software (or acting locally, thinking globally -- or whatever) will actually do this: check.
Sounds like the 60's to me, and I just want to say: I think that's great.--Seen
No, because in general bits/s != samples/s. Bits per second equals cycles per second TIMES BITS PER SAMPLE. You can typically get a lot more than one bit per sample, depending on the resolution of your A/D converter and the effective signal to noise ratio.
Put it another way: The Nyquist theorem is about discrete sampling of an analog waveform: the samples themselves are analog values. When you go further and digitize the samples, now you can start talking about bits per second.
--Seen