Another possibility is to use a more conductive semiconductor (such as carbon, i.e. diamond, if you can figure out how to dope it). Dimond is VERY thermo-conductive.
The "Preposterous Scale Integration" rap involves a 6 foot diamond cube, with power and heatsinking on two opposite faces, completely covered by the ends of two water-cooled silver buss bars. I/O on the other four faces using a forest of optic fibers, and the whole thing running red hot (diamond is very stable) in an inert atmosphere (but it's flammible!) filled glass envelope.
(The idea is to create the visual impression of a component of a golden-age SF author's idea of a computer. Like something you'd find in the Skylark of Space. B-) )
The entire history of mainstream ICs has been optical lithography on flat silicon. And that technology is going to hit a wall between one and two decades out, because atoms have finite size.
Like I said: It will hit a wall ONLY if it stays one-dimentional.
Even with optical lithography you can go multi-layer - and the industry already has. The interconnects are obvious. But vertical profiles have been built by growing epitaxial layers and doping those - sometimes repeatedly.
Still, they've generally only used this for creating "buried layers" of a single plane of components.
But why stop at one layer of components? Grow some more semiconductor, make another layer. And another, and another, and another.
Yes, yield would be a problem. (But remind me to go into the "Preposterous Scale Integration" rap some time. There is a solution. It abandons lithography for ion and electron beams and per-circuit computer feedback, but allows you to correct imperfections on-the-fly and build to ridiculously large sizes.)
HP says silicon electronics will reach a dead end in 2012
Hardly. Even if a feature size limit IS reached.
They only hit the wall if they stay two-dimensional. Given that the circuits are getting to near molecular thickness you have a LOT of doublings available before the chips are as thick as they are wide and high - or a closer limit (like heat dissipation) is reached.
Does it matter with this? The "digital" quality of the sound has already largely disappeared via the "copy protection" scheme - you're hearing mosly interpolated bits, so nobody can even claim to produce a purely digital rip of this stuff...
Not really.
If they were smart, they'd only replace samples where the typical interpolation done by CD players would hit the original sample value on-the-nose, or close enough as not to matter. There should be PLENTY of those.
A higher dependency on Solar Cells and heat-energy-converting/trapping devices could also help lower the amount of heat we have to deal with.
(By "heat-energy-converting/trapping" devices I presume you mean solar water heaters and the like.)
Solar is NOT a solution to "global warming". That's because solar panels capture nearly all the incoming radiant energy. Solar cells convert a few percent to electricity (that ends up as heat when it's used) and the rest to heat at the panel. Solar heaters just capture it and turn it into local heat.
The problem is that they replace something that reflected a significant part of the incident light still as LIGHT - at frequencies that pass right back through the greenhouse gasses. Dropping the albedo at the surface means you've boosted the greenhouse effect BIG time, without even touching the gas concentration.
Fortunately, global warming is NOT a big issue. As it stands it's you're talking raising the temperature a couple degrees a century. That's like moving your farm 50 miles south or a couple hundred feet down the hill. Big deal. TINY compared to the ongoing climatic cycle.
But it's doubly not a problem because it's a RISE in temperature. It's TRIVIAL to drop the temperature by tens of degrees: Just inject a few tons of dust into the upper atmosphere - like volcanos do from timt to time. Inject as much dust as you need to get the cooling you want, and if you overdo it a little bit most of it will be down in a year or so - so just reduce the amount you're injecting until you get the temperature where you want it.
Now if it were global COOLING we were worried about - like the return of the ice ages - that might be an actual problem. (But we could just burn a LOT of fossil fuel and hold it off another few centuries, until we can put some moon-sized mirrors at L4 or L5 and solar-cook the planet as much as necessary.)
MA is interesting. I guess those laws allowed a senator from the state to get away with murder (or at least negligenct homicide), but prevent a common citizen from protecting himself from a authority figure abusing his/her power.
MA is indeed interesting. Once it was a major factor in the revolt against England and a hotbed of freedom-lovers - and the free press. But after the Potato Famine an influx of immigrants (not JUST from Ireland) turned its legal system upside-down, making it a hotbed of censorship. The ideological descandants of the revolutionaries headed west.
A significant fraction of them ended up in Oregon, which is now a hotbed of porn publishing. B-) Unfortunately, the east-coast transplanted-European-serf mindset followed in two stages - into California starting perhaps in the '60s, and is just now moving on Oregon, Washington, Arizona, and Nevada.
Perhaps the "hi tek crash" will slow it down - at least until the next economic half-cycle.
How many times has a cop car without its lights on blow by you doing 90MPH on the freeway? Have you ever seen one of these guys pulled over for speeding?
No, and you won't. Because sometimes it's proper and legal cop behavior.
When they are called to a crime in progress they are often ordered to approach dark and silent. This is to avoid alerting the perpetrator, which might lead to a hostage situation.
They are supposed to proceed lights-only (because sirens CARRY) until close enough that the perp or his lookout might see them, then go down to just running lights (to look like ordinary traffic) and perhaps go totally dark for final approach if it's safe.
If they're full-blast on the freeway without lights they're probably either on such a call or moving up on someone they observed doing something questionable or illegal and trying not to spook him into a chase (VERY dangerous to bystanders) until they're too close for him to think he has a chance to escape.
... the cop said I was weaving around in my lane. Well, the REASON I was weaving... I was looking in the rear-view mirror, trying to figure out why a car had closed in on me going 120-150 mph, then slowed down to my speed and started shadowing me
Something similar happened to me. I was in the right-but-one lane just after bar closing, approaching my exit. Car pulled up into my right blind-spot and sat there - where I couldn't tell whether it had my bumper hooked or not.
Sped up, it sped up. Slowed down, it slowed down. (Is it a drunk or a cop.) Repeat, with more extreme changes, until I finally hit the brakes max from 65 down to about 45 and they couldn't slow abruptly enough. So THEN they lit up and pulled me over. Two in car, one came out...
"Do you know why we pulled you over?"
"Why, no, officer. Could you tell me, please?"
"You seemed to be having difficulty controlling your speed."
(No I didn't explode in her face...)
After I told her I thought she might have been a drunk and I was trying to get her out of my blind spot (and her partner had a good laugh) she gave me a bunch of drunk tests looked through the junk in my back seat (for "junk"?) and finally sent me on my way.
The US was created BY amateurs FOR amateurs.
on
IANAL
·
· Score: 4
Why should they alone have the right to give legal advice?
Damn straight.
The US was created by people with essentially no "legal standing" - who just got approval from a bunch of like-minded people and went out and DID it (not necessarily in that order B-) ).
The legal system was INTENDED to be simple enough that anyone could understand the rules he must live by, to allow anyone to bring his own case when disputes arise, and to guarantee that he would be able to have advice by someone more knowledgable if he didn't feel confident to do it all himself. (Note that even supreme court justices do NOT have to be lawyers - and some were not. It "just happens" that lately they all are.)
But the government grew, and the legal code grew, and the precedents accumulated, and the spoken language drifted - causing the words whose precice meanings had been tested in court and thus frozen to become a technical jargon. And as the amount of law grew to mind-boggling proportions the schools gave up and/or deteriorated, so that people of legal adult age may have nothing but lore about the laws they are expected to live by.
Meanwhile the government decided to limit the people who would be allowed to charge for legal advice and "waste judges' time" to those it certified as having some expertese. As always, it cloaked its market-limiting move as consumer protection. But what about the guarantee of the right to council of your own chosing? Why, you still have that - just chose them from the licensed lawyers and pay the fee. Can't pay? That's OK - if you're a criminal defendant we'll draft a random lawyer and make him represent you. If you're really lucky he might actually know something about that branch of the law. Civil case? Tough luck!
I dont mean to bust anyone's ballon, but when was last time you saw an OSS project (other than the actual kernel) start from scratch, create a spec, follow it, implement it in a timely manner, and deliver a quality-end product?
Hmmm....
Bind
Telnet
Most of the other infrastructure of the Internet - especially in the early days.
vi
Sendmail
Netnews
Mosaic
BSD
UNIX itself - up through at least Version 6 and maybe 7.
I could go on.
Yes, they weren't explicitly "Open Source". But they WERE developed in environments where the source - either deliberately or defacto - was open or opened (for some value of "open") or otherwise not kept proprietary. Many of them were developed before there WAS an explicit open-source software model.
And yes, some of them didn't run straight from conception to execution according to the orignal design. (But what project does?)
Take Unix, for instance. It was a spare-time hack done in the back room by a handfull of techies trying to turn an abandoned computer into something useful. When it first become part of an official project it was officially the underpinnings of a word-processor application for the legal department - under a strict injunction NOT to spend their time writing an OS. (The minimalization of the kernel proper - with important core functionality such as command interpreters as applications - may have arisen partly to provide plausible deniability: "That's not an OS. An OS does THIS and THAT and THIS OTHER THING, right? Well this doesn't!"). Its source was freely distributed to hundreds of educational institutions, for the cost of media and shipping, until upper management finally took notice when the kernel appeared in its entirety in a set of OS course notes that quickly became an underground classic.
There were major questions about the proprietary status of Unix. It was written before copyright had been revised to apply to software or the patent system hacked to apply, so the only protection it had was trade secret - which evaporates when the cat leaves the bag. It was written by a regulated monopoly, which was mandated to publish and make available anything it did that had applications outside of telephony. The entire EXISTENCE of SVRn may have occurred because a proprietary rewrite was needed to "close" the Unix source.
Perhaps there haven't been a lot of high-profile breakthrough applications out of open source just lately. But how many breakthroughs have you seen from CLOSED source lately? How long has it been since something as fundamental as the Internet, the Web, or spreadsheets came out of either open OR closed source projects, hmmm?
I talked with one of the programmers of the modem some time back. (We both used to go to a periodic social event.)
According to him there's a single configuration bit in the modems (at least the ones available several years ago) that switches it between being a portable leaf node and a base station. By flipping a user could donate bandwidth from his high-speed connection to his neighborhood. (Of course he'd have to run it through ipmasq on his firewall...)
Don't know the details. Perhaps somebody from the company could verify and expand, or deny, this story.
very interesting, but... if you were making use of the thing, how can you say it was obsolete?
I wasn't actually "making use" of it. They assigned me to writing a program to make sure all the parts were working.
But my personal definition of "obsolete" is "If you were going to buy something to do the same job would you consider buying this or is there something so much better/cheaper now that this wouldn't be considered."
I actually programmed a 606 back in my cutting-hacker-teeth days. A high-school summer job at a university lab - and it WAS obsolete at the time. The IBM 14xx and 709x and the Control Data 1604 (discrete transistors and diodes germanium mainly - on cards) were the state-of-the-art.
The 606 was a later, bigger box of the same pluggable modules. They had vacuum tubes at the end for the output amplifier/switch, and little baseless peanut-tube vacuum DIODES along the side of the module for the logic gates. (Think DTL, but with the "T" standing for "Triode" rather than "Transistor". At least I think the amplifer tube was a triode - perhaps a dual triode. (Two logic gates per plugin! Miniaturization!) Didn't get into the circuitry.)
606 had one K of memory - decimal K, addressed from 000 to 999. If I recall correctly, a word was ten four-bit digits plus sign. (I don't recall if it was BCD or bi-quinary.)
Ten program-readable registers - each consisting of a rotary switch for each digit plus a toggle for the sign - made up the middle part of the processor's end panel. Bottom part was a giant plugboard, top part a neon light display, with a neon lamp for each state of the program sequencer and a bunch more for an output register.
There was a knob (bakelite pointer on a pot) for adjusting the clock speed.
One I used had a printer/reader, separate punch, and a drum memory. UofMich guys had mad a plug board that turned it into a stored-program machine - booting a program from the console card reader. Printer was strictly numeric and took about a second per line to print. A row of heavy metal typebars would rise up, each one stopping at the correct height, and when enough time had elapsed for all of them to have reached maximum height the whole thing would "whack" forward through the ribbon onto the paper.
There was a prime-number sieve program that they used for basic checks. After about the first 13 primes or so it was taking longer than a printer cycle to compute them. B-)
the most common type of password attack comes in the form of "social engineering"
*cough*
Like giving your password to someone doing a study on passwords?
I figured someone would catch that. B-)
Which brings up the question of how many "cryptics", confonted with such an obvious piece of social engineering as asking you to disclose your password for a survey, would lie, masquerading as one of the other categories.
It's just like someone asking for information about whether you own a gun, what kind, where you keep it, etc. in a situation where the person giving the answer can be identified. Even if you are otherwise scrupulously honest, the canonical thing to do is to lie. No one else has a right to that information, revealing it reduces your security, anyone asking is suspect, and refusing to answer leaks part of the information you want hidden (because people who DON'T own one generally won't refuse).
Pollsters asking how you voted/will vote is a similar situation. B-)
Each telecom laid plenty of extra dark fiber, to be sure, but they also assumed that a much larger portion of what they laid would be utilized...
Not true.
If you dig a trench from California to New York, are you going to lay in a couple of fibers and when you need more dig another trench, or are you going to lay enough fibers that you don't need to dig another trench for a hundred years? Especially when essentially ALL the cost is digging the trench.
The tellcos did NOT expect all the fiber to be lit up soon. And they do NOT buy equipment to light it up until they NEED to light it up - because Moore's Law applies to the equipment. (Are you going to buy equipment today to light fibers that won't be needed for two years? Or are you going to put the money to earning more money for a couple years then buy a box that can light twice as much bandwidth and have bux left over?)
So with the economic slowdown the fibers aren't getting lit up as fast as some people expected. That's no skin off the carriers' noses - they had to dig the trench and lay the fiber to have a business at all. The box makers are having a hard time because the carriers aren't buying many boxes, so the box makers' stocks have tanked.
But the box makers' stocks are bottoming out. (The carriers ARE still installing boxes, and before all the ones they already bought are in service the carriers will have to order some more.) And some market players would like them to tank even more, so they can make more money by anticipating the move.
So suddently the press "discovers" the dark fiber. And you see scare articles that make it sound like it's a disaster, a major failure of planning on the carriers' part. "Oh, dear! Only 5% of the fiber is being used. (wring hands) The SKY is falling!" As if the carriers were in trouble because only 5% of their investment was generating revenue, instead of all but a small delta of it - just like they intended.
And the stocks tank for a couple more days, letting the players pick up more at fire-sale prices before they recover.
Disaster my aching tail. The only thing that would be a disaster for a carrier would be if the dark fiber ran out and the carrier had to dig again.
Yes the slowdown is hard on the box makers, who'd like to sell more boxes than they are. But it's a slowing of growth, not a retreat. So the boxes still sell - and when things pick up (or when the carriers exhaust their pre-bought inventory even before a recovery) the boxes will be selling a lot faster.
Re:Difference between "adjusted" and "reported"?
on
Red Hat In The Black
·
· Score: 2
So let me get this straight.
They pay $30m for a company worth $2m, don't include the $28m loss and everyone is ok with that?
No.
They pay $30m for a company that is worth $30m. But $2m of that is desks and chairs and computers and office supplies, and $28m is that it's Cygnus Support, a money-making (or potentially money-making) company, which is worth a lot more than the desks and chairs in its office.
Now they didn't really "pay" anything but printing-press stock (which dilutes the stock in their shareholders hands, so it's not really nothing). But the combined pie got bigger, so it's appropriate that each piece of stock is a smaller part of the bigger pie.
But they paid more stock for it than if they'd gone to the market and somehow sold stock at the Red Had market price (without driving that into the ground) and bought the stock of Cygnus Support (without blasting that into the stratosphere). So the difference is treated as an expense over the next several years. That means the part of the company profits that support the valuation of the stock only gets taxed once (when the former Cygnus Support shareholders sell their Red Had stock), not twice (also when Red Hat makes the money in the first place).
If what you're interested in is whether the basic business of Red Hat + Cygnus is actually making a profit, ignoring the financial noise from the stock-certificate printing press, you need to look at the "adjusted" numbers, not the "reported" ones.
You mean diff. between reported and pro-forma?
on
Red Hat In The Black
·
· Score: 2
Just because a cost is a one time expense doesn't mean it is irrelevant.
Yes. But I think we're talking about the difference between "pro-forma" and "reported" here.
Pro-forma excludes several things, in order to get a measure of how the underlying business is actually doing.
Some of them are actual one-time costs that you're concerned about. But the main components are "amortized good-will" from merger accounting and "deferred compensation" from stock options. Now IANAnAccountant, but if I understand them correctly, those last two are taking advantage of provisions in the tax law to avoid the shareholders being double-taxed on a couple things.
Here's my understanding of this - which may be flawed, so you HAVE been warned.
Amortized good-will avoids double-taxing the shareholders of the acquired company (typically people who got their stock near the founding when it was almost free, then busted their butts and risked their houses during the startup period) for the appreciation of their stock from the acquisition.
When a merger is accounted as an acquisition, extra stock in the "surviving" company is printed to replace the retired stock of the "acquired" company. What really happens is the two pies are merged and everybody gets a proportional piece of the bigger pie. But it's accounted as if the company that printed the stock actually sold it, then spent the money to buy the other company's stock. The amount they paid will be at a premium over what the other company was worth on the open market. So the difference is treated as if the acquired company had an asset called "good will" which makes up the difference. Uncle lets the acquiring company act as if they "spent the money" to "buy the good-will", and then treat like any other capital expense and "amortize" it - deducting the cost in little chunks over a number of years. (Uncle eventually gets his cut - but only ONCE - when the former holders of the stock in the "absorbed" company finally sell it. Their stock price went up by the premium paid - the value of the good-will asset - as a result of the merger.)
"Deferred compensation" is similar, but relates to stock options. What actually happens is that the company offered the employees some stock when it was cheap. The employees don't actually get the shares into their hands and have the right to sell it until it's "vested" - once they've been around for a while. (And of course they might not actually bother to exercise the option - especially if the stock goes down.) When they sell it they'll be taxed on the difference between what they paid for it and what it sold for.
But it's accounted roughly as if the company sold the stock on the open market at the exercise price, then had to buy it back at the higher price to give to the employee once it's vested. So the company gets to write off the difference against profits on ITS taxes. If you think of it the way it's accounted that's what happened - the company had an expense and deducted it. But if you think of it as the company selling the stock to the employee at the lower price, the company got credited for the taxes the option-holder paid (in return for not making the money from the stock's price change for itself).
Pro-forma accounting treats the stock as being sold at the exercise price (what you and I would think of as having happened), but the reported income treats it as if the company paid the employee the difference. The latter is another valid way of looking at it: The company was trying to pay the guy extra by letting him assume some of the risk in return for sharing the rewards, and if it had actually paid cash instead it could have deducted THAT and everybody would have been in the same tax situation. But if you're evaluating the health of the company's business you want to look at things the pro-forma way, not the reported income way.
... GPL explicitly denies the software itself as a source of revenue...
No, it doesn't. You can sell GPLed software. You just can't do so EXCLUSIVELY - somebody else could start selling it, or giving it away, once they have a copy.
That doesn't mean they WILL. And it doesn't mean that, if they do, your market will dry up entirely. Think: Did you buy the distribution of Linux you last loaded, or did you download it from the net?
Yes the financial model pays you for the service you provide rather than the software itself. But even with proprietary software a very large component of the price is for the service rather than the underlying code. People buy software for what it will do, not what it is, and when they pay they expect support.
Microsoft wants to (or already does) use GPL'ed code but does not want to release what they did with it.
If so, I'd bet it was incorporated by low-level workers, in violation of the company's policies and the wishes of the upper management. Too much of Microsoft's business model is built on keeping the source to themselves for an exec to risk having to give it all away for a few extra features - or even a lot of very powerful features.
Especially since the GPL doesn't stop you from reverse-engineering the code and writing your own equivalent to create the feature! It even allows a single person to do this - though a large company would want to "clean-room" it, with one team doing the analysis and another the coding, to avoid risk of conatmination with enough code snippets to cross the boundary between a genre member with fair-use quotes and a derived work.
GPL and the other open-source licenses are built on copyright - which protects an expression - not on patent - which protects an idea. (Despite the way some companies are trying to stretch copyright into a super-patent.)
And the open-source social contract (not to be confused with the Social Contract license B-) ) is this:
- Here's what I did. Some nice ideas, and a lot of drudgework to make it run.
- You like it? Use it.
- You tweak it and keep it to yourself? That's fine.
- You tweak it and sell it, or give it away?
- Don't keep the tweaks to yourself,
- don't keep ME from using your tweaks, and
- make sure everybody else who gets it does the same.
- (You want to take the ideas and do your OWN drudgework to make another version run? And sell that? And NOT share the guts? I can't stop you. Just don't use the fruit of MY drudgework in YOUR version.)
Cygnus doesn't even exist anymore. It was acquired by RedHat a long time ago.
While Cygnus has been absorbed by Red Had, my understanding is that they still provide open-source support contracts for a fee. It's just that they're now part of Red Hat (and doubling as Red Hat's support department), rather than an independent entity.
If that has changed I (and a lot of other people) would like to know about it. B-)
But as far as actually HIRING them I have indeed been out of the loop for a while - since I don't have to buy open-source support in my current situation. So if Red Hat DID break them I would only know about it if somebody complained - like here on Slashdot.
I've been reading/. since before the acquisition and haven't seen anybody complain. So I'm assuming that they're still in business, doing what Gilmore and crew created them to do.
And that includes spiking the perenial "Who will support me?" objection to open source applications and system components.
Whenever I try to get my employer to try some open source app, the main thing they scream about is upport. If it breaks, they want someone to blame/fix it, and they are willing to pay big bucks for that.
Then contract with Cygnus Support!
That's what they DO!
(Besides upgrade many of the Gnu tools, cut distributions, and maintain archives for them - with money from the support contracts.)
Think of Cygnus as a software company that happens to put its products under the GPL. Or think of it as a service organization specializing in GPLed software tools.
Why should the professors be in fear of being sued now (since they are doing this pre-emptive "I'm going to sue to make sure you can't" thing). It is already known that they have cracked or found a way to circumvent the watermark, so the damage has been done. The watermark is useless.
Because the company in question has already threatened to sue them if they publish HOW the watermark is defective and how to circumvent it - and the potential penalties are draconian.
The company probably threatened suit because it would like to do a somewhat tightened-up version of the watermarking scheme and sell THAT. Publication of the research would give the readers (researchers in the field):
a detailed list of what the researchers found
powerful tools to investigate related schemes
a strong insight into the ways the company's designers were hiding the watermark information
This would compromise all related schemes, effectively destroying not just the original set of schemes, but everything the company might develop in the future.
If the company can sue and win, it means major financial damage for the researchers, their college, their peer-reviewed journal, and the specialty's academic organization. This amounts to the bulk of the academics and professionals that are currently working in the field. Such a result would have the proverbial "chilling effect" on further research into weaknesses of copy-protection and watermarking systems. (Chilling as in liquid nitrogen.)
The problem with DSL is that telcos are trying to do it over ANCIENT POTS lines. Much of it untwisted, oxidizing, all connected at rats nest type junction boxes, alligator clips, etc. Good enough for voice. But no more than that.
And this whole "how many feet are you from the switch" contingency only proves that DSL is a HUGE KLUGE.
It's not just the quality of the old lines. The distance limit is inherent in the use of the copper pairs.
A copper pair has stray capacatance, inductance, and resistance. The stray capacatance and inductance combine to form a transmission line. If it's balanced well enough that the audio isn't crufty, it's balanced well enough that the first hundred or two megahertz is pretty OK, too.
Deviations from a smooth twised pair (like the stuff you see in junction boxes) distort the smoothness of the transmission line, producing some reflections and other pathologies. But the amount of havoc they produce is dependent on their size relative to the wavelength of the signal - and the wavelength of one megahertz is pushing a kilometer. (That's why T1 (and Primary Rate ISDN), with a carrier at 1.544/2 MHz and harmonics several times higher, can travel over ordinary pair, junction boxes and all.) Just don't splice in a branch with a few hundred feet of wire to give options on what house to feed. (Or cut the branch off when somebody wants to use the line for DSL.)
But the wire also has resistance. And that makes the transmission line a LOSSY transmission line.
The distance limit comes from the fact that the stray resistance is in SERIES and the stray capacatance is in PARALLEL. That forms a low-pass filter. The farther you go down the wire, the more AC is attenuated, and the higher frequencies are attenuated more than lower frequencies. They are also shifted in phase as you go farther down the wire, and the combination of the two effects distorts waveforms as well as losing power.
After a few miles, even audio becomes muffled. (Then phone companies may add "loading coils" - lump inductors which level out the high frequencies a bit up to the resonance - a couple KHz - but eat everything above that. Which is why you can forget DSL, base-rate ISDN, and 56k modems in rural areas.)
A DSL modem acts like a bundle of narrow-band modems operating at a comb of frequencies. Each modem handles only a narrow range of frequencies, so the phase distortion within its band does not produce significant waveform distortion of its portion of the signal. The modems are independent, so phase differences between the separate carriers can be ignored. And the modems can independently adjust their gain, so the selective fading of the higher frequencies can be compensated for - up to a point.
Eventually the signal in the higher bands is too weak to reliably extract from line and modem internal noise. So the modem drops off the higher carriers and runs at a lower data rate. And the farther out, the more you lose, until there's really no point to it...
Want to go farther? Use thicker wires, spaced farther apart. (Lower resistance and lower capacatance.) Want to go still farther? Use an amplifier/filter combination (like the cable systems do) or a repeater (like some extended-DSL systems). Want higher data rate, ditto - and also use a different insulation. (Insulation can attenuate REALLY high frequencies by "inductively coupling" a "stray resistance" into the "stray inductance" of the line.)
But if you replaced the copper with a type-II superconductor and the plastic with vacuum insulation, and didn't foul up the smooth twist too much, you could run your DSL line to the other side of the continent.
The main thing that "category >3" lines are doing is twisting the copper pair more often and more evenly, so frequencies in the hundreds of megahertz don't couple to nearby wires and get interfered with, distorted, or lost. For the first couple megahertz, phone lines are OK except for the selective attenuation issue. Cat 5 still has the same selective attanuation as Cat 3, which is why you can only run it around in a building rather than all over the city.
Just because it not included in the distibution does not mean you cant use it. You can still download the package from http://coombs.anu.edu.au/~avalon/ and compile it yourself.
I think you missed a point...
The license (as "clarified") bans modified versions.
The version used with OpenBSD is modified.
So they had to take it down.
Now you have two choices:
Download and install a copy of the modified version. This violates the license.
Download and install a copy of the UNmodified version. This means you don't have the OpenBSD modifications.
Now since the whole POINT of OpenBSD is that it has been heavily vetted for security bugs, do you REALLY want to install the UNmodified version? Of the FIREWALL code?
Slashdot Mirror
Another possibility is to use a more conductive semiconductor (such as carbon, i.e. diamond, if you can figure out how to dope it). Dimond is VERY thermo-conductive.
The "Preposterous Scale Integration" rap involves a 6 foot diamond cube, with power and heatsinking on two opposite faces, completely covered by the ends of two water-cooled silver buss bars. I/O on the other four faces using a forest of optic fibers, and the whole thing running red hot (diamond is very stable) in an inert atmosphere (but it's flammible!) filled glass envelope.
(The idea is to create the visual impression of a component of a golden-age SF author's idea of a computer. Like something you'd find in the Skylark of Space. B-) )
The entire history of mainstream ICs has been optical lithography on flat silicon. And that technology is going to hit a wall between one and two decades out, because atoms have finite size.
Like I said: It will hit a wall ONLY if it stays one-dimentional.
Even with optical lithography you can go multi-layer - and the industry already has. The interconnects are obvious. But vertical profiles have been built by growing epitaxial layers and doping those - sometimes repeatedly.
Still, they've generally only used this for creating "buried layers" of a single plane of components.
But why stop at one layer of components? Grow some more semiconductor, make another layer. And another, and another, and another.
Yes, yield would be a problem. (But remind me to go into the "Preposterous Scale Integration" rap some time. There is a solution. It abandons lithography for ion and electron beams and per-circuit computer feedback, but allows you to correct imperfections on-the-fly and build to ridiculously large sizes.)
HP says silicon electronics will reach a dead end in 2012
Hardly. Even if a feature size limit IS reached.
They only hit the wall if they stay two-dimensional. Given that the circuits are getting to near molecular thickness you have a LOT of doublings available before the chips are as thick as they are wide and high - or a closer limit (like heat dissipation) is reached.
I thought there were costs built into blank CDs to offset some of thus. Does this mean the prices of blanks will decrease?
It might be interesting to institute a suit to block any company producing copy-protected CDs from receiving their share of the "tax" money. B-)
Does it matter with this? The "digital" quality of the sound has already largely disappeared via the "copy protection" scheme - you're hearing mosly interpolated bits, so nobody can even claim to produce a purely digital rip of this stuff...
Not really.
If they were smart, they'd only replace samples where the typical interpolation done by CD players would hit the original sample value on-the-nose, or close enough as not to matter. There should be PLENTY of those.
A higher dependency on Solar Cells and heat-energy-converting/trapping devices could also help lower the amount of heat we have to deal with.
(By "heat-energy-converting/trapping" devices I presume you mean solar water heaters and the like.)
Solar is NOT a solution to "global warming". That's because solar panels capture nearly all the incoming radiant energy. Solar cells convert a few percent to electricity (that ends up as heat when it's used) and the rest to heat at the panel. Solar heaters just capture it and turn it into local heat.
The problem is that they replace something that reflected a significant part of the incident light still as LIGHT - at frequencies that pass right back through the greenhouse gasses. Dropping the albedo at the surface means you've boosted the greenhouse effect BIG time, without even touching the gas concentration.
Fortunately, global warming is NOT a big issue. As it stands it's you're talking raising the temperature a couple degrees a century. That's like moving your farm 50 miles south or a couple hundred feet down the hill. Big deal. TINY compared to the ongoing climatic cycle.
But it's doubly not a problem because it's a RISE in temperature. It's TRIVIAL to drop the temperature by tens of degrees: Just inject a few tons of dust into the upper atmosphere - like volcanos do from timt to time. Inject as much dust as you need to get the cooling you want, and if you overdo it a little bit most of it will be down in a year or so - so just reduce the amount you're injecting until you get the temperature where you want it.
Now if it were global COOLING we were worried about - like the return of the ice ages - that might be an actual problem. (But we could just burn a LOT of fossil fuel and hold it off another few centuries, until we can put some moon-sized mirrors at L4 or L5 and solar-cook the planet as much as necessary.)
MA is interesting. I guess those laws allowed a senator from the state to get away with murder (or at least negligenct homicide), but prevent a common citizen from protecting himself from a authority figure abusing his/her power.
MA is indeed interesting. Once it was a major factor in the revolt against England and a hotbed of freedom-lovers - and the free press. But after the Potato Famine an influx of immigrants (not JUST from Ireland) turned its legal system upside-down, making it a hotbed of censorship. The ideological descandants of the revolutionaries headed west.
A significant fraction of them ended up in Oregon, which is now a hotbed of porn publishing. B-) Unfortunately, the east-coast transplanted-European-serf mindset followed in two stages - into California starting perhaps in the '60s, and is just now moving on Oregon, Washington, Arizona, and Nevada.
Perhaps the "hi tek crash" will slow it down - at least until the next economic half-cycle.
How many times has a cop car without its lights on blow by you doing 90MPH on the freeway? Have you ever seen one of these guys pulled over for speeding?
No, and you won't. Because sometimes it's proper and legal cop behavior.
When they are called to a crime in progress they are often ordered to approach dark and silent. This is to avoid alerting the perpetrator, which might lead to a hostage situation.
They are supposed to proceed lights-only (because sirens CARRY) until close enough that the perp or his lookout might see them, then go down to just running lights (to look like ordinary traffic) and perhaps go totally dark for final approach if it's safe.
If they're full-blast on the freeway without lights they're probably either on such a call or moving up on someone they observed doing something questionable or illegal and trying not to spook him into a chase (VERY dangerous to bystanders) until they're too close for him to think he has a chance to escape.
... the cop said I was weaving around in my lane. Well, the REASON I was weaving ... I was looking in the rear-view mirror, trying to figure out why a car had closed in on me going 120-150 mph, then slowed down to my speed and started shadowing me
Something similar happened to me. I was in the right-but-one lane just after bar closing, approaching my exit. Car pulled up into my right blind-spot and sat there - where I couldn't tell whether it had my bumper hooked or not.
Sped up, it sped up. Slowed down, it slowed down. (Is it a drunk or a cop.) Repeat, with more extreme changes, until I finally hit the brakes max from 65 down to about 45 and they couldn't slow abruptly enough. So THEN they lit up and pulled me over. Two in car, one came out...
"Do you know why we pulled you over?"
"Why, no, officer. Could you tell me, please?"
"You seemed to be having difficulty controlling your speed."
(No I didn't explode in her face...)
After I told her I thought she might have been a drunk and I was trying to get her out of my blind spot (and her partner had a good laugh) she gave me a bunch of drunk tests looked through the junk in my back seat (for "junk"?) and finally sent me on my way.
Why should they alone have the right to give legal advice?
Damn straight.
The US was created by people with essentially no "legal standing" - who just got approval from a bunch of like-minded people and went out and DID it (not necessarily in that order B-) ).
The legal system was INTENDED to be simple enough that anyone could understand the rules he must live by, to allow anyone to bring his own case when disputes arise, and to guarantee that he would be able to have advice by someone more knowledgable if he didn't feel confident to do it all himself. (Note that even supreme court justices do NOT have to be lawyers - and some were not. It "just happens" that lately they all are.)
But the government grew, and the legal code grew, and the precedents accumulated, and the spoken language drifted - causing the words whose precice meanings had been tested in court and thus frozen to become a technical jargon. And as the amount of law grew to mind-boggling proportions the schools gave up and/or deteriorated, so that people of legal adult age may have nothing but lore about the laws they are expected to live by.
Meanwhile the government decided to limit the people who would be allowed to charge for legal advice and "waste judges' time" to those it certified as having some expertese. As always, it cloaked its market-limiting move as consumer protection. But what about the guarantee of the right to council of your own chosing? Why, you still have that - just chose them from the licensed lawyers and pay the fee. Can't pay? That's OK - if you're a criminal defendant we'll draft a random lawyer and make him represent you. If you're really lucky he might actually know something about that branch of the law. Civil case? Tough luck!
Hmmm....
Bind
Telnet
Most of the other infrastructure of the Internet - especially in the early days.
vi
Sendmail
Netnews
Mosaic
BSD
UNIX itself - up through at least Version 6 and maybe 7.
I could go on.
Yes, they weren't explicitly "Open Source". But they WERE developed in environments where the source - either deliberately or defacto - was open or opened (for some value of "open") or otherwise not kept proprietary. Many of them were developed before there WAS an explicit open-source software model.
And yes, some of them didn't run straight from conception to execution according to the orignal design. (But what project does?)
Take Unix, for instance. It was a spare-time hack done in the back room by a handfull of techies trying to turn an abandoned computer into something useful. When it first become part of an official project it was officially the underpinnings of a word-processor application for the legal department - under a strict injunction NOT to spend their time writing an OS. (The minimalization of the kernel proper - with important core functionality such as command interpreters as applications - may have arisen partly to provide plausible deniability: "That's not an OS. An OS does THIS and THAT and THIS OTHER THING, right? Well this doesn't!"). Its source was freely distributed to hundreds of educational institutions, for the cost of media and shipping, until upper management finally took notice when the kernel appeared in its entirety in a set of OS course notes that quickly became an underground classic.
There were major questions about the proprietary status of Unix. It was written before copyright had been revised to apply to software or the patent system hacked to apply, so the only protection it had was trade secret - which evaporates when the cat leaves the bag. It was written by a regulated monopoly, which was mandated to publish and make available anything it did that had applications outside of telephony. The entire EXISTENCE of SVRn may have occurred because a proprietary rewrite was needed to "close" the Unix source.
Perhaps there haven't been a lot of high-profile breakthrough applications out of open source just lately. But how many breakthroughs have you seen from CLOSED source lately? How long has it been since something as fundamental as the Internet, the Web, or spreadsheets came out of either open OR closed source projects, hmmm?
I talked with one of the programmers of the modem some time back. (We both used to go to a periodic social event.)
According to him there's a single configuration bit in the modems (at least the ones available several years ago) that switches it between being a portable leaf node and a base station. By flipping a user could donate bandwidth from his high-speed connection to his neighborhood. (Of course he'd have to run it through ipmasq on his firewall...)
Don't know the details. Perhaps somebody from the company could verify and expand, or deny, this story.
very interesting, but... if you were making use of the thing, how can you say it was obsolete?
I wasn't actually "making use" of it. They assigned me to writing a program to make sure all the parts were working.
But my personal definition of "obsolete" is "If you were going to buy something to do the same job would you consider buying this or is there something so much better/cheaper now that this wouldn't be considered."
Obsolete does NOT mean it stops doing the job.
I actually programmed a 606 back in my cutting-hacker-teeth days. A high-school summer job at a university lab - and it WAS obsolete at the time. The IBM 14xx and 709x and the Control Data 1604 (discrete transistors and diodes germanium mainly - on cards) were the state-of-the-art.
The 606 was a later, bigger box of the same pluggable modules. They had vacuum tubes at the end for the output amplifier/switch, and little baseless peanut-tube vacuum DIODES along the side of the module for the logic gates. (Think DTL, but with the "T" standing for "Triode" rather than "Transistor". At least I think the amplifer tube was a triode - perhaps a dual triode. (Two logic gates per plugin! Miniaturization!) Didn't get into the circuitry.)
606 had one K of memory - decimal K, addressed from 000 to 999. If I recall correctly, a word was ten four-bit digits plus sign. (I don't recall if it was BCD or bi-quinary.)
Ten program-readable registers - each consisting of a rotary switch for each digit plus a toggle for the sign - made up the middle part of the processor's end panel. Bottom part was a giant plugboard, top part a neon light display, with a neon lamp for each state of the program sequencer and a bunch more for an output register.
There was a knob (bakelite pointer on a pot) for adjusting the clock speed.
One I used had a printer/reader, separate punch, and a drum memory. UofMich guys had mad a plug board that turned it into a stored-program machine - booting a program from the console card reader. Printer was strictly numeric and took about a second per line to print. A row of heavy metal typebars would rise up, each one stopping at the correct height, and when enough time had elapsed for all of them to have reached maximum height the whole thing would "whack" forward through the ribbon onto the paper.
There was a prime-number sieve program that they used for basic checks. After about the first 13 primes or so it was taking longer than a printer cycle to compute them. B-)
the most common type of password attack comes in the form of "social engineering"
*cough*
Like giving your password to someone doing a study on passwords?
I figured someone would catch that. B-)
Which brings up the question of how many "cryptics", confonted with such an obvious piece of social engineering as asking you to disclose your password for a survey, would lie, masquerading as one of the other categories.
It's just like someone asking for information about whether you own a gun, what kind, where you keep it, etc. in a situation where the person giving the answer can be identified. Even if you are otherwise scrupulously honest, the canonical thing to do is to lie. No one else has a right to that information, revealing it reduces your security, anyone asking is suspect, and refusing to answer leaks part of the information you want hidden (because people who DON'T own one generally won't refuse).
Pollsters asking how you voted/will vote is a similar situation. B-)
Each telecom laid plenty of extra dark fiber, to be sure, but they also assumed that a much larger portion of what they laid would be utilized ...
Not true.
If you dig a trench from California to New York, are you going to lay in a couple of fibers and when you need more dig another trench, or are you going to lay enough fibers that you don't need to dig another trench for a hundred years? Especially when essentially ALL the cost is digging the trench.
The tellcos did NOT expect all the fiber to be lit up soon. And they do NOT buy equipment to light it up until they NEED to light it up - because Moore's Law applies to the equipment. (Are you going to buy equipment today to light fibers that won't be needed for two years? Or are you going to put the money to earning more money for a couple years then buy a box that can light twice as much bandwidth and have bux left over?)
So with the economic slowdown the fibers aren't getting lit up as fast as some people expected. That's no skin off the carriers' noses - they had to dig the trench and lay the fiber to have a business at all. The box makers are having a hard time because the carriers aren't buying many boxes, so the box makers' stocks have tanked.
But the box makers' stocks are bottoming out. (The carriers ARE still installing boxes, and before all the ones they already bought are in service the carriers will have to order some more.) And some market players would like them to tank even more, so they can make more money by anticipating the move.
So suddently the press "discovers" the dark fiber. And you see scare articles that make it sound like it's a disaster, a major failure of planning on the carriers' part. "Oh, dear! Only 5% of the fiber is being used. (wring hands) The SKY is falling!" As if the carriers were in trouble because only 5% of their investment was generating revenue, instead of all but a small delta of it - just like they intended.
And the stocks tank for a couple more days, letting the players pick up more at fire-sale prices before they recover.
Disaster my aching tail. The only thing that would be a disaster for a carrier would be if the dark fiber ran out and the carrier had to dig again.
Yes the slowdown is hard on the box makers, who'd like to sell more boxes than they are. But it's a slowing of growth, not a retreat. So the boxes still sell - and when things pick up (or when the carriers exhaust their pre-bought inventory even before a recovery) the boxes will be selling a lot faster.
So let me get this straight.
They pay $30m for a company worth $2m, don't include the $28m loss and everyone is ok with that?
No.
They pay $30m for a company that is worth $30m. But $2m of that is desks and chairs and computers and office supplies, and $28m is that it's Cygnus Support, a money-making (or potentially money-making) company, which is worth a lot more than the desks and chairs in its office.
Now they didn't really "pay" anything but printing-press stock (which dilutes the stock in their shareholders hands, so it's not really nothing). But the combined pie got bigger, so it's appropriate that each piece of stock is a smaller part of the bigger pie.
But they paid more stock for it than if they'd gone to the market and somehow sold stock at the Red Had market price (without driving that into the ground) and bought the stock of Cygnus Support (without blasting that into the stratosphere). So the difference is treated as an expense over the next several years. That means the part of the company profits that support the valuation of the stock only gets taxed once (when the former Cygnus Support shareholders sell their Red Had stock), not twice (also when Red Hat makes the money in the first place).
If what you're interested in is whether the basic business of Red Hat + Cygnus is actually making a profit, ignoring the financial noise from the stock-certificate printing press, you need to look at the "adjusted" numbers, not the "reported" ones.
Just because a cost is a one time expense doesn't mean it is irrelevant.
Yes. But I think we're talking about the difference between "pro-forma" and "reported" here.
Pro-forma excludes several things, in order to get a measure of how the underlying business is actually doing.
Some of them are actual one-time costs that you're concerned about. But the main components are "amortized good-will" from merger accounting and "deferred compensation" from stock options. Now IANAnAccountant, but if I understand them correctly, those last two are taking advantage of provisions in the tax law to avoid the shareholders being double-taxed on a couple things.
Here's my understanding of this - which may be flawed, so you HAVE been warned.
Amortized good-will avoids double-taxing the shareholders of the acquired company (typically people who got their stock near the founding when it was almost free, then busted their butts and risked their houses during the startup period) for the appreciation of their stock from the acquisition.
When a merger is accounted as an acquisition, extra stock in the "surviving" company is printed to replace the retired stock of the "acquired" company. What really happens is the two pies are merged and everybody gets a proportional piece of the bigger pie. But it's accounted as if the company that printed the stock actually sold it, then spent the money to buy the other company's stock. The amount they paid will be at a premium over what the other company was worth on the open market. So the difference is treated as if the acquired company had an asset called "good will" which makes up the difference. Uncle lets the acquiring company act as if they "spent the money" to "buy the good-will", and then treat like any other capital expense and "amortize" it - deducting the cost in little chunks over a number of years. (Uncle eventually gets his cut - but only ONCE - when the former holders of the stock in the "absorbed" company finally sell it. Their stock price went up by the premium paid - the value of the good-will asset - as a result of the merger.)
"Deferred compensation" is similar, but relates to stock options. What actually happens is that the company offered the employees some stock when it was cheap. The employees don't actually get the shares into their hands and have the right to sell it until it's "vested" - once they've been around for a while. (And of course they might not actually bother to exercise the option - especially if the stock goes down.) When they sell it they'll be taxed on the difference between what they paid for it and what it sold for.
But it's accounted roughly as if the company sold the stock on the open market at the exercise price, then had to buy it back at the higher price to give to the employee once it's vested. So the company gets to write off the difference against profits on ITS taxes. If you think of it the way it's accounted that's what happened - the company had an expense and deducted it. But if you think of it as the company selling the stock to the employee at the lower price, the company got credited for the taxes the option-holder paid (in return for not making the money from the stock's price change for itself).
Pro-forma accounting treats the stock as being sold at the exercise price (what you and I would think of as having happened), but the reported income treats it as if the company paid the employee the difference. The latter is another valid way of looking at it: The company was trying to pay the guy extra by letting him assume some of the risk in return for sharing the rewards, and if it had actually paid cash instead it could have deducted THAT and everybody would have been in the same tax situation. But if you're evaluating the health of the company's business you want to look at things the pro-forma way, not the reported income way.
... GPL explicitly denies the software itself as a source of revenue ...
No, it doesn't. You can sell GPLed software. You just can't do so EXCLUSIVELY - somebody else could start selling it, or giving it away, once they have a copy.
That doesn't mean they WILL. And it doesn't mean that, if they do, your market will dry up entirely. Think: Did you buy the distribution of Linux you last loaded, or did you download it from the net?
Yes the financial model pays you for the service you provide rather than the software itself. But even with proprietary software a very large component of the price is for the service rather than the underlying code. People buy software for what it will do, not what it is, and when they pay they expect support.
Microsoft wants to (or already does) use GPL'ed code but does not want to release what they did with it.
If so, I'd bet it was incorporated by low-level workers, in violation of the company's policies and the wishes of the upper management. Too much of Microsoft's business model is built on keeping the source to themselves for an exec to risk having to give it all away for a few extra features - or even a lot of very powerful features.
Especially since the GPL doesn't stop you from reverse-engineering the code and writing your own equivalent to create the feature! It even allows a single person to do this - though a large company would want to "clean-room" it, with one team doing the analysis and another the coding, to avoid risk of conatmination with enough code snippets to cross the boundary between a genre member with fair-use quotes and a derived work.
GPL and the other open-source licenses are built on copyright - which protects an expression - not on patent - which protects an idea. (Despite the way some companies are trying to stretch copyright into a super-patent.)
And the open-source social contract (not to be confused with the Social Contract license B-) ) is this:
- Here's what I did. Some nice ideas, and a lot of drudgework to make it run.
- You like it? Use it.
- You tweak it and keep it to yourself? That's fine.
- You tweak it and sell it, or give it away?
- Don't keep the tweaks to yourself,
- don't keep ME from using your tweaks, and
- make sure everybody else who gets it does the same.
- (You want to take the ideas and do your OWN drudgework to make another version run? And sell that? And NOT share the guts? I can't stop you. Just don't use the fruit of MY drudgework in YOUR version.)
Cygnus doesn't even exist anymore. It was acquired by RedHat a long time ago.
/. since before the acquisition and haven't seen anybody complain. So I'm assuming that they're still in business, doing what Gilmore and crew created them to do.
While Cygnus has been absorbed by Red Had, my understanding is that they still provide open-source support contracts for a fee. It's just that they're now part of Red Hat (and doubling as Red Hat's support department), rather than an independent entity.
If that has changed I (and a lot of other people) would like to know about it. B-)
But as far as actually HIRING them I have indeed been out of the loop for a while - since I don't have to buy open-source support in my current situation. So if Red Hat DID break them I would only know about it if somebody complained - like here on Slashdot.
I've been reading
And that includes spiking the perenial "Who will support me?" objection to open source applications and system components.
Whenever I try to get my employer to try some open source app, the main thing they scream about is upport. If it breaks, they want someone to blame/fix it, and they are willing to pay big bucks for that.
Then contract with Cygnus Support!
That's what they DO!
(Besides upgrade many of the Gnu tools, cut distributions, and maintain archives for them - with money from the support contracts.)
Think of Cygnus as a software company that happens to put its products under the GPL. Or think of it as a service organization specializing in GPLed software tools.
And they're no pikers, either.
Because the company in question has already threatened to sue them if they publish HOW the watermark is defective and how to circumvent it - and the potential penalties are draconian.
The company probably threatened suit because it would like to do a somewhat tightened-up version of the watermarking scheme and sell THAT. Publication of the research would give the readers (researchers in the field):
a detailed list of what the researchers found
powerful tools to investigate related schemes
a strong insight into the ways the company's designers were hiding the watermark information
This would compromise all related schemes, effectively destroying not just the original set of schemes, but everything the company might develop in the future.
If the company can sue and win, it means major financial damage for the researchers, their college, their peer-reviewed journal, and the specialty's academic organization. This amounts to the bulk of the academics and professionals that are currently working in the field. Such a result would have the proverbial "chilling effect" on further research into weaknesses of copy-protection and watermarking systems. (Chilling as in liquid nitrogen.)
The problem with DSL is that telcos are trying to do it over ANCIENT POTS lines. Much of it untwisted, oxidizing, all connected at rats nest type junction boxes, alligator clips, etc. Good enough for voice. But no more than that.
And this whole "how many feet are you from the switch" contingency only proves that DSL is a HUGE KLUGE.
It's not just the quality of the old lines. The distance limit is inherent in the use of the copper pairs.
A copper pair has stray capacatance, inductance, and resistance. The stray capacatance and inductance combine to form a transmission line. If it's balanced well enough that the audio isn't crufty, it's balanced well enough that the first hundred or two megahertz is pretty OK, too.
Deviations from a smooth twised pair (like the stuff you see in junction boxes) distort the smoothness of the transmission line, producing some reflections and other pathologies. But the amount of havoc they produce is dependent on their size relative to the wavelength of the signal - and the wavelength of one megahertz is pushing a kilometer. (That's why T1 (and Primary Rate ISDN), with a carrier at 1.544/2 MHz and harmonics several times higher, can travel over ordinary pair, junction boxes and all.) Just don't splice in a branch with a few hundred feet of wire to give options on what house to feed. (Or cut the branch off when somebody wants to use the line for DSL.)
But the wire also has resistance. And that makes the transmission line a LOSSY transmission line.
The distance limit comes from the fact that the stray resistance is in SERIES and the stray capacatance is in PARALLEL. That forms a low-pass filter. The farther you go down the wire, the more AC is attenuated, and the higher frequencies are attenuated more than lower frequencies. They are also shifted in phase as you go farther down the wire, and the combination of the two effects distorts waveforms as well as losing power.
After a few miles, even audio becomes muffled. (Then phone companies may add "loading coils" - lump inductors which level out the high frequencies a bit up to the resonance - a couple KHz - but eat everything above that. Which is why you can forget DSL, base-rate ISDN, and 56k modems in rural areas.)
A DSL modem acts like a bundle of narrow-band modems operating at a comb of frequencies. Each modem handles only a narrow range of frequencies, so the phase distortion within its band does not produce significant waveform distortion of its portion of the signal. The modems are independent, so phase differences between the separate carriers can be ignored. And the modems can independently adjust their gain, so the selective fading of the higher frequencies can be compensated for - up to a point.
Eventually the signal in the higher bands is too weak to reliably extract from line and modem internal noise. So the modem drops off the higher carriers and runs at a lower data rate. And the farther out, the more you lose, until there's really no point to it...
Want to go farther? Use thicker wires, spaced farther apart. (Lower resistance and lower capacatance.) Want to go still farther? Use an amplifier/filter combination (like the cable systems do) or a repeater (like some extended-DSL systems). Want higher data rate, ditto - and also use a different insulation. (Insulation can attenuate REALLY high frequencies by "inductively coupling" a "stray resistance" into the "stray inductance" of the line.)
But if you replaced the copper with a type-II superconductor and the plastic with vacuum insulation, and didn't foul up the smooth twist too much, you could run your DSL line to the other side of the continent.
The main thing that "category >3" lines are doing is twisting the copper pair more often and more evenly, so frequencies in the hundreds of megahertz don't couple to nearby wires and get interfered with, distorted, or lost. For the first couple megahertz, phone lines are OK except for the selective attenuation issue. Cat 5 still has the same selective attanuation as Cat 3, which is why you can only run it around in a building rather than all over the city.
I think you missed a point...
The license (as "clarified") bans modified versions.
The version used with OpenBSD is modified.
So they had to take it down.
Now you have two choices:
Download and install a copy of the modified version. This violates the license.
Download and install a copy of the UNmodified version. This means you don't have the OpenBSD modifications.
Now since the whole POINT of OpenBSD is that it has been heavily vetted for security bugs, do you REALLY want to install the UNmodified version? Of the FIREWALL code?