Slashdot Mirror

NEGATIVE STABILITY. All previous aircraft designs had been aerodynamically stable. That is, the center of gravity was well in front of the center of lift and the center of pressure (drag).

To illustrate the difference between stable and unstable designs, take a shirt cardboard and, holding it by the leading edge, pull it rapidly through the air. It will stretch out behind your hand in a stable manner. This is a stable design Now take it by the trailing edge push forward from there. It will immediately flip up or down uncontrollably. That is an unstable design.

The downside of aerodynamic stability is that the aircraft is nose-heavy and always trying to nose down. The elevator must therefore push the tail down to level the airplane. But in addition to rotating the airplane from nose-down to level, the elevator is exerting negative lift; that is, it is pushing the airplane down. In order to counteract this negative lift, the wing needs to be made larger to create more positive lift. This increases both weight and drag, decreasing aircraft performance. In pitch-up situations including hard turns which are the bread and butter of aerial combat, this negative effect is greatly magnified.

The YF-16 became the world's first aircraft to be aerodynamically unstable by design. With its rearward center of gravity, its natural tendency is to nose up rather than down. So level flight is created by the elevator pushing the tail up rather than down, and therefore pushing the entire aircraft up. With the elevator working with the wing rather than against it, wing area, weight, and drag are reduced. The airplane was constantly on the verge of flipping up or down totally out of control,. and this tendency was being constantly caught and corrected by the fly-by-wire control system so quickly that neither the pilot nor an outside observer could know anything was happening. If the control system were to fail, the aircraft would instantly disintegrate; however, this has never happened.

"This is not a troll comment but can it run on a cluster?"

No it's not troll, but it's totally uniformed. Currently SMP (multiprocessor/multithreaded) VMware is only supported on ESX server as an addon. As ESX runs on bare hardware (it's GSX who runs as a Linux application), there is currently no support for "virtual multiple CPUs in Linux". (Xen does this, but it's not the issue now).

Additionally OpenMOSIX (which comes with ClusterKnoppix - I guess you meant this by "those Live CDs"), does not to "SMP like" processing. Instead it combines the processes in a "global system view" state. (Too much technical details here, but a multiple threads are not migrated -- see HOWTO).

Moreover, it would be slow because of obvious issues -- as in network based access to disk and shared memory!

Finally multiple GSX servers managed from a single point is already possible with VMWare virtual center (google this yourselves is left for an exercise).

Sorry, but your suggestion will not work (at least under current circumstances).

Hyper Text Coffee Pot Control Protocol (HTCPCP/1.0)

http://www.faqs.org/rfcs/rfc2324.html

He could still get a patent.

And at least his program (presumably) works.

Look at Patent 5488364

http://www.faqs.org/rfcs/rfc959.html File Transfer Protocol

http://www.ccp14.ac.uk/ccp14admin/security/secure_ tunnelling_ftp.htm
Secure FTP transfers via Secure Shell Tunnelling

http://winscp.net/eng/docs/introduction
WinSCP is an open source freeware SFTP client for Windows using SSH. Legacy SCP protocol is also supported. Its main function is safe copying of files between a local and a remote computer.

etc. etc.

Interesting claim. See RFC 706 entitled, "On the junk mail problem", by Jon Postel, dated Nov 1975. There were less than 100 systems on the net when this RFC was written. This may also be the first mention of "denial of service" attacks.

But - are the NTP patents valid? I say they are perfectly obvious. Back before 1985 I was using fido-net systems and there were some running over packet radio. My neighbour across the street ran packet radio back then.

To send an email over a packetized transmission system is perfectly obvious to _anyone_ who thinks about this for a moment.

It doesn't even require a practitioner of the field. Even a retard would think of this.

I agree. One of the things that is so unfathomably stupid about some of the software patents I've seen is that they just completely ignore one of the fundamental principles of computer science. In order to prevent systems' complexity from getting totally out of control, you break things down into subsystems, and you give each a well-defined behavior, making it as general-purpose as practical.

For example, TCP/IP works the same whether it's running on top of Ethernet or PPP or FDDI or radio waves or avian carrier or whatever. This is done on purpose specifically to make it so that everything that depends on something doesn't have to be re-invented every time some change in an unrelated system is made. It's so vitally important to do this (because otherwise complete chaos would result) that it should be impossible to claim you know anything about computers without knowing this.

Anyway, the point is a wireless network works the same as a wired network. It's an abstraction. Whether it's wireless or wired doesn't matter. And yet, through the collective lunacy of the government and others, somehow people are being granted patents for every possible combination of interchangeable things they can put together.

The thing that I just do not understand is how people cannot see this. Nobody would contemplate allowing separate patents for a microwave oven that draws its electrical energy from a natural-gas fired power generation facility and for a microwave oven that draws its electrical energy from a dam that generates electricity hydroelectrically. They wouldn't because they see that wall outlets are general-purpose, and that plugging a microwave into a different wall outlet is not an invention. Why, then, is the equivalent thing allowed for software? How can it be possible that people are so ignorant that they can't see that networks (and other well-defined, standardized interfaces in computers) work exactly the same way?

How would you know that an email is a reply?

Well, grabbing a reply e-mail at random from my inbox, I find these nifty headers: "References" and "In-Reply-To" (see section 3.6.4 of the linked RFC).

Whenever you send an e-mail, your mail client (whatever it may be) should generate a Message-ID, and any replies to that message should include this ID in "In-Reply-To" and "References" headers.

So, identifying a reply is very simple: If the "In-Reply-To" or "References" headers contain the ID of a message that was sent from this account, then the message is a reply. There are two obvious ways to know if a given message ID in a received e-mail was actually sent out by this account: A database of message IDs or, better yet, using a keyed encoding to generate message IDs. If the message ID were generated, for example, by concatenating the sender's username and a timestamp, then encrypting the result with, say AES (because AES is *very* fast), then base64-encoding the result of that, then hotmail servers could easily verify the validity and origin of the message ID when it came back.

I'm sure with a little thought, some even better approaches could be developed. This isn't a hard problem.

The only real way to check it would be for hotmail to keep a track record of everyone you send mail to, add them to your adress book and then let those emails bypass your spam filters.

In other words, an automatic whitelist. Sure, that's also a very good idea, and also very easy to implement.

I ran into a site that claimed most Mac viruses were actually MS Office macros, but they didn't show any data.

This site shows that as of 2000, there were about 40 and were mostly Mac system-specific:
http://www.faqs.org/faqs/computer-virus/macintosh- faq/

This is something a good router will take care of, but very few of them do, not even the customized Linksys routers, nor the linux routers like Smoothwall do BOTH of the required things.

All you need to do to get outstanding performance on an asymetric line is the following:

[ON THE ROUTER]
1. Prioritize TCP ACK Packets as #1 to always go upstream first to your connection
2. Restrict upload rate by 2% - 5% of the actual upload rate of the connection

Do these three things, and enjoy a fast connection in both directions.

The reason your downloads choke when you upload, is because the TCP ACK packets that need to go UP to the site you are downloading from (to get the next chunk of your download) can't get out because your saturating your upload connection. By prioritizing these ACK packets, they go out first, and your download runs full speed, while your uploads fly as well.

The reason for #2, is that most DSL and Cable modems have a large buffer in them... If you saturate this buffer, then your router can't guarentee the TCP ACK packets go out first since other packets are sitting in your DSL/Cable modems buffers... This is very important to do #2.

There are more things you to can do to help, but they get more sophistated. If you are interested in managing asynchronous TCP connections, check out this document, which covers everything including the kitchen sink:

http://www.faqs.org/docs/Linux-HOWTO/ADSL-Bandwidt h-Management-HOWTO.html

- Voxel

This should work:

RFC 2549 - IP over Avian Carriers with Quality of Service

--Woof!

Of special interest might be RFC 706, "On the Junk Mail Problem." They saw it coming...

Of special interest might be RFC 706, "On the Junk Mail Problem." They saw it coming...

This article was dead on.

My first ever encounter with this mentality was in high school when my music theory instructor told us that she loved her Mac and when I tried to argue with her about a number of things, she'd repeatedly reply with "No Mac has ever been hacked or had a virus on it."

Now, at the time, I was a young nooblet and probably should have let it slide but instead I snuck into her office and opened up her Macintosh's word editing software with the intent of some lil' bastardry. I found the option to replace a mistyped word with another that the user entered. After that, whenever she typed the word "the", it was replaced with "WARNING! VIRUS DETECTED! PULL PLUG FROM OUTLET AS SOON AS POSSIBLE!" Unfortunately, her son knew enough about computer to fix it so my fun didn't last very long (only one or two lunges at the wall).

Back to the issue--I think it is a grave mistake for anyone to ever feel 100% invulnerable when it comes to computers that are connected to the internet in anyway. I would diagnose this as a standard case of a false sense of security. This is something that has plagued many people throughout history and often led to their downfalls.

What message am I trying to get across to Mac users? First, realize you're not invulnerable. Second, just browse around and look at what's out there for you to use as anti-virus and virus blocking tools. And if you don't want to, read some horror stories, perhaps that will motivate you to become aware of possible worms in your Apple.

What we have here is that, in addition to doing this, Windows is also offering to set up an ad-hoc (i.e. computer-to-computer) network on the link-local subnet with the same SSID as that of the last network the laptop connected to. I wonder what the rationale for doing this could have been. It seems to me that a machine should not offer to set up an ad-hoc network unless specifically directed to do so by the user. When such a network is set up then it is appropriate to use link-local addressing to auto-configure the interface.

It's simple destructive interference. Maybe the active noise control FAQ will help. See section 2.5 in particular.

See e.g. the fusion faq.

This He3 thing keeps coming up as an economic argument for a Moon base. But in addition to the point made in the review, that much development remains to make a He3 fusion process feasible, it also requires that, during the same time, nobody finds a better solution that does not require He3. This seems like a real slender thread to hang a big Moon infrastructure project on.

Let me add one other factoid. The last time we bought some He3, it was somewhere around $5000 per mole. So even if you went with straight He3 + D, which yields about 18 MeV, the cost for the He3 alone would come to around 1 cent per kilowatt hour. My electric company charges me a bit over 4 cents per KWh, so this seems in the ballpark already, without involving the moon. Of course, global He3 production will have to grow by a zillion times if people start using it for electricity, but the resulting economy of scale should make this cost even lower.

UDP based HTTP GETs

Take a look at OpenBSD. Once you understand the UNIX philosophy, using it is a breeze. Combine a bells-and-whistles shell like bash with a terminal multiplexer like screen and you have a powerful and elegant user environment. And your investment in time will pay you back double.

We've replaced half of our servers at this point-- very little new hardware required. The new machines stay up longer, they require less maintenance, and they work the way you expect them to. They aren't susceptable to viruses. When they finish running nightly jobs, they email me reports, even if the programs aren't designed to do so because I have the control. Need new functionality with your app? Usually piping the output to some other built-in utility gets the job done. None of our users know the difference; they still use Outlook to check email, use file and print servers. What they don't know is that most of these services aren't running on a MS platform anymore. But MY job is a lot easier. And that is only the tip of the iceberg.

Obviously, I don't know what you spend your time doing, and if you say you don't have the time, then maybe you really don't. But I've heard this same argument before, from admins at my company who work in different offices, and you just sound just like them. They just can't be bothered. Too bad. I'm not working nearly as hard. Remember the Perl motto: "The three principal virtues of a programmer are Laziness, Impatience, and Hubris."

Ok, I've done a little research into Evolutionary Programming and Evolutionary Algorithms. Exactly what good has it done? What I'm reading shows a method of simulating the evolutionary process in a computer. But so what, you can make a computer do anything you want.

But here is something even more interesting. The information I found on EP and EA says, "In nature, we see that the encoding for genetic information (GENOME) is done in a way that admits asexual REPRODUCTION. Asexual reproduction typically results in OFFSPRING that are genetically identical to the PARENT."

But then on the same page it says "Evolution, by definition, absolutely requires diversity in order to work. In nature, an important source of diversity is MUTATION." Now, according to the dictionary, the definition of MUTATION is "A change of the DNA sequence within a gene or chromosome of an organism resulting in the creation of a new character or trait not found in the parental type." It seems to me that single celled organisms from the billions of years ago that evolutionist promote would have never produced anything but the same exact cell. Mutation would have never happened because the genetic material would have always been produced asexually. This is all based on what I read at http://www.faqs.org/faqs/ai-faq/genetic/part2/sect ion-1.html. Now I'm sure from here there are obviously "outside forces" that would cause mutation. And I'm sure there are others. But there may also be "outside forces" that would protect a species from mutation. Point being, you can't quantify everything.

You may have a point that ID doesn't really contribute anything SCIENTIFICALLY, I don't know, but I don't think that is the goal of ID. I think ID is an approach to the origins of life that allows for a creator and not a haphazard glop of primordial ooze. But I don't see evidence that the study of evolution has produced any scientifically usablility either. You can tell me that "we have found this cure", or "we have discovered this process" due to evolution. But that's just you saying that. I want to hear from the group that discovered it to tell me and prove to me that some evolutionary process helped them in discovering what they found.

Now don't get the impression that I know alot about evolution or Intellegent Design. I'm actually a Creationist. I believe the God of the Bible created the universe and all that is in it. The same God that Washington, Adams, and Lincoln prayed to and asked for wisdom with the running of this country. Which is why this is the greatest country on the planet. I don't waste valuable time on either approach. ID'ers are just tiptoeing around what they really want to say. But I like the idea of ID because it is a way to approach the atheist evolutionist that wouldn't believe a god created anything. Alot of atheist evolutionist think that ID'ers or creationist are morons that blindly believe something that has no proof. So for a creationist to approach an evolutionist would be of no merit. I've noticed alot of evolutionist pre-judge creationist and ID'ers. I know this because I'm an ex-evolutionist and an ex-athiest. But anyway, I digress because I'm getting off the subject. I'll not continue this bantor with you. I appreciate you're input though, I've discovered something I didn't know before. Cheers

I don't know if Parry would be happy working there.

Speaking of unknown compression programs, does anyone remember OWS?

I had a good laugh at that one when I figured out how it worked, way back in the BBS days.

I'll get excited when I see the collective realization that just the technology we currently have enables the defeat of many entrenched, obsolete social constructs.

It's the withering away of the state; Lenin forsaw it but mistook it. It turns marxism and capitalism both ass-over-teakettle.

Why not use these techniques to defeat fearful democratic and republican governments, as well? They are equally egregious, just the authoritarian regimes are less duplicitous and a damn sight rougher.

Plagiarized from here.

You're a few years too late.

Actually, the effects of the depletion of the ozone layer and global warming on each other is pretty circular- UVB destroys small phytoplankton in the Antarctic. This contributes to global warming [see HERE], as well as a collapse in the polar and sub-polar oceanic food supply.

I could point out dozens of flaws in this theory (e.g. what's the actual, real rate of such destruction? how do you know if it is anywhere near significant?) and counter that competent scientists themselves believe little in such shallow factoids as you claim, e.g. here:

http://www.faqs.org/faqs/ozone-depletion/uv/

"Subject: 7.) Are sheep going blind in Chile?

If they are, it's not because of ozone depletion.

For a short period each year, the edge of the ozone hole passes over Tierra del Fuego, at the southern end of the South American continent. This has led to a flurry of reports of medical damage to humans and livestock. Dermatologists claim that they are seeing more patients with sun-related conditions, nursery owners report damage to plants, a sailor says that his yacht's dacron sails have become brittle, and a rancher declares that 50 of his sheep, grazing at high altitudes, suffer "temporary cataracts" in the spring. (_Newsweek_, 9 December 1991, p. 43; NY Times, 27 July 1991, p. C4; 27 March 1992, p. A7).

These claims are hard to believe. At such a high latitude, springtime UV-B is naturally very low and the temporary increase due to ozone depletion still results in a UV fluence that is well below that found at lower latitudes. Moreover, the climate of Patagonia is notoriously cold and wet. (There is actually more of a problem in the summer, after the hole breaks up and ozone-poor air drifts north. The ozone depletion is smaller, but the background UV intensity is much higher.) There may well be effects on _local_ species, adapted to low UV levels, but even these are not expected to appear so soon. It was only in 1987 that the hole grew large enough to give rise to significant UV increases in southern Chile, and cataracts and malignant melanomas take many years to develop."

The deterioration of the Ozone layer was very well proven, and Crutzen, Molina, and Rowland won the Nobel for conclusively proving that CFCs caused the depletion of the layer.

http://news.nationalgeographic.com/news/2003/08/08 05_030805_ozone.html

Now the issue is why this is happening - the argument about whether this and CFC ban is causation or merely correlation could probably go on forever.
However, if the phenomenon is cyclical (after all ozone is produced by UV radiation from regular O2 oxygen, so increasing/decreasing ozone levels might be just yet another complex cyclical phenomenon), we're going to find out. This might take time, though.

Look here:

http://www.faqs.org/faqs/ozone-depletion/intro/

And Jon Postel, who is the principle author of RFC 959. Probably Tim Berners-Lee, too.

Off the top of my head I can think of severals ways around that... modem dial long distance into another country, or buy a T-1 into a neutral country into another neutral country etc etc, you know...

You've forgotten about the evil bit

BitComet was also found to ignore the "evil bit."

Some security folks had a similar idea a couple years ago. Setting an "evil" bit in all network packets that were malicious. For some reason it has had difficulty gaining acceptance and buy-in from administrators:

  - http://www.faqs.org/rfcs/rfc3514.html

Just implement RFC 2795.

All public ipv4 transit networks in existence use a routing protocol called BGP4 (Border Gateway Protocol v4 - rfc1771)

In my old geazer defense, I studied networking shortly after that RFC was written. Most of the characteristics you describe appear to have been introduced as new features in BGP-4. My networking book devotes a whopping two pages to BGP. At any rate, it appears BGP-4 solves the problem being discussed.

Thanks for the info!

Considering that my networking professor told the whole class about it, there are more than a handful of people that know.

For those that don't know, the issue arises out of the way the internet does routing. IPv4 uses a flat routing system. Every key router on the internet knows how many hops away it is from all of the other key routers and which direction the router is in. Consider (the dots are placeholders so slashdot will display my beautiful ASCII art properly):

[cute but erroneous diagram clipped to avoid lameness filter]

Router D knows that it is one hop away from router E. B knows that it is two hops from E. How? Because D tells B that it is one hop away from E, so if B sends a packet to D, D can deliver it in one hop. C knows that it is three hops away.

Now suppose router B goes down. C knows that it can't reach E through its usual three hops, but when it talks to its neighbor to the right, it sees that F can reach E in three hops, so C is now four hops away from E. Now when C sees traffic headed for E, it sends the traffic to F.

How do you poison the system? If one of the key trusted routers, such as C, tells everyone that they are two hops from everywhere, large portions of the internet will try to route through C. If you can take control of a trusted router in each of several key locations, you can confuse the overwhelming majority of the internet into thinking you are offering the best route to their destination.

The short route won't make a big difference for nearby traffic, but traffic headed ten or twenty hops away will wind up going towards C when it should go someplace else.

The above-described mechanism for updating the routing tables is the key to the internet's ability to automatically route around cities that have been destroyed by a nuclear weapon.

Oh good god, what complete and utter BS. Lest anyone believe this is actually how transit routing works:

All public ipv4 transit networks in existence use a routing protocol called BGP4 (Border Gateway Protocol v4 - rfc1771). BGP is an "inter-autonomous system" routing protocol. That means, as a whole, it has no network awareness of individual routers, links, specific static addresses or locations. Essentially, all it knows is that a set of ip networks comprise an Automous System (labeled via an ARIN/RIPE/APNIC assigned Autonomous System Number). When a bgp router in one AS has an established bgp session with a router in a different AS, it tells the other router all the foreign ASNs that the network is willing to take traffic for and prepends its own ASN to the front of the list. The same is done for networks that originate within the local AS (i.e. the ASN is appended to "nothing" and is thus respresents the final destination AS) [there is also an origin ASN field, but ignoring that for the sake of simplicity]. This list is known as a bgp path. Thus, to find a route(s) to any accessible ipv4 address, a bgp router need only look at all the paths that contain the destination ASN, and the shortest path is generally the best route (although certainly not always). The actual job of routing packets is handled on a per-AS basis; i.e. each network is responsible for knowing, internally, how best to move packets to all the AS' that are connected to it.

You will note, however, that the core problem you describe continues to exist in this model, simply not on a per-router basis. If AS999 sends a path such as "9999 701" to all neighboring ASes, they'll believe that a viable route for traffic destined to AS701 is via AS9999, which, given a large major network, could be extremely distruptive.

However, in reality, this has not been a grave concern for a number of yea

Even with 100ms latency, using a 30bit TCP window size [RFC 1323] you can theoretically transfer data at 5 gigabits per second.

1,073,741,824 bits per 200ms (100ms RTT)... and thats with the receiver just ACK'ing after the transfer of each 1 gig chunk, not providing intermittant ACKs throughout the transfer.

You need something beyond statefull inspection, you need to be able to magically tell what is a "good" or "safe" http connection from a "bad" http connection.

And it's called ssh.

lol. Sounds like the Evil Bit :)

The names of domains identify organizations, not categories of information. The domain name hierarchy was never envisioned as a system for classifying information by category. It was created to solve a technical problem.

The explosive growth of the Internet strained the scheme for distributing host name to address mappings. Mappings were maintained in a single file and distributed to all hosts. Two key drawbacks of this scheme were:

1. The network bandwidth required to distribute the file was proportional to the square of the number of hosts in the network; and,
2. local organizations had to wait for a central administrator to make their name changes visible to the network.

The domain name system was designed to eliminate bandwidth and administration bottlenecks without sacrificing consistency. The hierarchical structure of domain names made it easy to subdivide the mappings into zones. Each local organization could then maintain and distribute its own zone, instead of depending on a central administrator. The ability to query a remote zone file was added, along with a caching strategy. These greatly reduced the network bandwidth required to distribute and query the mappings.

The reason single-letter domains were reserved until now was also apparently technical. Engineers worried that the system might develop bottlenecks around top level domain names. One day it might need to be reorganized by initial letter (archive.org becomes archive.a.org, slashdot.org becomes slashdot.s.org, and so on). To keep this option open, single-letter domains were reserved. The conclusion seems to be that this is unnecessary. The problem then becomes how to manage the crowd of would-be registrants of single-letter domains.

Much of this information comes from the authoritative documentation on the domain name system. See: RFC 1034, DOMAIN NAMES - CONCEPTS AND FACILITIES.

AFAIK, the standard for representing international characters in a hostname is via IDNA encoding (see RFC3490) which maps international characters to ascii characters. It looks like name components in this spec have to begin with ascii "xn--", so a single international character domain name would be at least a 5 character ascii domain name which may already be allowed.

That said, I'm really not too familiar with this subject at all (I just looked up the RFC) so please someone correct me if I'm wrong...

Why not just use the "evil bit"? http://www.faqs.org/rfcs/rfc3514.html

Technically speaking, it'd be much easier just to try and get all the pron people to put some kind of broadcast flag on their own stuff, and then just filter by that.

Sounds like you just reinvented the evil bit!

SCNR this one, so don't mod me down for not knowing that RFC3514 is an april fools day joke.

If you are having mime problems perhaps this will help?

I'll admit that I haven't looked at Usenet in quite a while, despite founding talk.politics.libertarian nearly 12 years ago, yet when I look at the newsgroup today, it still looks like it has useful information.

For a while I worked at a company that delivered a full Usenet feed over satellite, but they went out of business. I suspect you can't even do a full feed with a single transponder at QPSK 3/4 today. Does anyone still deliver Usenet feeds over satellite?

[it's full of thugs, imbeciles, etc...] "All this is true mostly of the Alt newsgroups, which were designed to have few inhibitions. Other groups, such as Comp, Sci, Soc and Humanities, fare much better, largely because they can be moderated. They contain lots of valuable stuff.

Newsgroups in the alt.* hierarchy can be moderated too, it just happens that most aren't. There's a list of active newsgroups, including which ones are moderated and which aren't, at faqs.org.

Apparently, 1 and 2-character identifier names are all the rage in China.

Actually, it's more that they're all the rage in specifications for hashes. RTFRFC.

What does the call to macro 'F' do?

By convention, the permutations in a hash or cipher are given single-letter names beginning with F (for 'Function'). If you ever decide to implement a crypto suite you get used to names like F, G, H (permutations), or X0[19] (cell 19 of section 0 of the primary execution context).

How about the call to 'RR'?

It performes a Rotate Right on the word it's called on. (((x) >> (y)) | ((x) << (32 - (y)))) is a right circular shift of the word X by Y places. RL is a Rotate Left, or left circular shift.

As a final caveat, that implementation of RR and RL is faulty for general values of X and Y but works within the constraints of the program.

With Linux, you don't even have to install the OS on the hard drive. It can run off of a CD.

As Snagglepuss would say:

It can run off a network, even
Time for Windows to exit, stage left!

If you RTFRFC you might notice that those are the variable and section names used in the algorithm specificaiton.

It seems that people forget that for the most part of the early internet days it was all SLIP/PPP. People didn't use NAT. They got a public IP when they dialed up. SIP has been around a long time. They banked on address availability (or later IPv6) when they designed it. The widespread adoption of NAT was not a foregone conclusion at that point. NAT just gained momentum a lot faster. It is one solution to a shortage of IPs, I would just argue that it isn't the best.