Stephen Fry MADNESS: 'New domain names GENERATE NEW IP NUMBERS'

Autocue extraordinaire Stephen Fry has cheerily claimed new domain registrations "generate new IP numbers which so far show no sign of giving out". He blogged the explanation while celebrating being the first person in Blighty to sign up for a .uk domain. Dot-UK registry Nominet has not challenged Fry's latest boob, even …

[Anonymous Coward]

Re: Well he is sort of right

"IP adress show all the signs of being infinite actually"

How do you insert 'infinite' into a DNS record. Would you not have run out of atoms in the known universe a long, long way before that point? As an IP Address is always a collection of digits or characters then they will always have to fall into a finite space.

[TRT]

Re: Well he is sort of right

Right. I'm buying infinite.uk today.

[NumptyScrub]

Re: Well he is sort of right

To pull a Fry and talk authoritatively about something I don't actually understand properly, an infinite IP address range would have cardinality Aleph naught, because it would be directly translatable to the set of positive integers. Effectively what Pierre was arguing is that as we reach whatever finite limit we currently have defined, we'll just extend the limit closer towards Aleph-naught.

We've gone from 32-bit to 128-bit addressing this time, suggesting the next shift could be to 512-bit addressing, then 2048 and so on. Until we have extraterrestrial colonies though, I'd suggest one address available per atom in the planet is more than enough; apparently there are 1.33e50 of them, give or take, which is a bit less than 2¹⁷⁹ or a 180-bit address space.

IPv6 should be ok for a little while yet.

[Anonymous Coward]

Re: Well he is sort of right

"IP adress show all the signs of being infinite actually. IPv4 addies are finite, IPv6 are too, but if we keep implementing new protocols as the old ones are used up, then "IP adresses" in general are infinite..."

So if we redefine what the term "IP address" means, then they become infinite. Brilliant. Thanks for playing.

Similarly, if we change what "The Duke of Kent" means, then he's also infinite, or if we change "World War Two timeframe" to not mean 1939-1945, but instead to cover an infinite period of time, that's also infinite.

[ElReg!comments!Pierre]

Re: Well he is sort of right

There is no intrinsic limitation in the number of "Internet Protocol address" that we can create. Each subset (protocol) that we implement has finite limits, but creating new protocols (and therefore exponentially more adresses) is "just" a matter of adding numbers*. It is very difficult to run out of numbers. I suggest you start counting up and stop when you have run out of numbers to add. Should take you quite a while.

*or letters or whatev' we use to make it more practical.

[Fred Dibnah]

Well...

.. I think I'll let him off with a bit of artistic licence, because someone from an engineering company wrote this:

" Q. How many addresses will IPv6 accommodate? How does that compare to IPv4?

" A. IPv6 supports addresses that have four times the number of bits as those of IPv4 addresses (128 instead of 32). IPv6 is expected to accommodate, theoretically, an almost infinite number of IP addresses..."

Isn't a number that is almost infinite, er, infinite?

BTW the company is Cisco:

http://tinyurl.com/nll67w4

[sabroni]

Re: Isn't a number that is almost infinite, er, infinite?

Is a number that is almost 100 100? Is 98 100?

And Mr. Dibnah wouldn't approve of that attitude when it came to engineering. Is this strong enough to support 100 tons? Yes, it's strong enough to support 98 tons....

Finding another idiot on the internet doesn't make Stephen Fry correct, it's not the facts with the most votes that are right.

And relax......

[NogginTheNog]

Re: Isn't a number that is almost infinite, er, infinite?

Upvote just for mentioning the Bolton Legend!

[Bob Wheeler]

Re: Isn't a number that is almost infinite, er, infinite?

When ever the word 'almost' is used, it mean "not quiet" or "less than"

e.g

Q: Are we there yet?

A: Almost there: in other words, not quiet there yet.

Q: Is 0.999 recuring equal to 1.0

A: It almost qual to 1.0: in other words it is less than.

[Stephen W Harris]

Re: Isn't a number that is almost infinite, er, infinite?

0.999 recurring is exactly equal to 1.

Real numbers are different if there's another number that can be placed between them (a !=c if there exists b such that a<b<c. I typical example would be "(a+c)/2"). In the case of 0.999(rec) there is no value "b" which can be placed between it and 1; therefore 0.999(rec) == 1. It's not "almost 1", it's "exactly 1".

[Sir Runcible Spoon]

Re: Isn't a number that is almost infinite, er, infinite?

Infinitely (in language terms) can be used to describe a very large thing quite happily. Just don't try and use it in mathematics as such.

And don't even get me started on infinitesimal.

[Benchops]

> And don't even get me started on infinitesimal.

I want to get you started on infinitesimal.

But not very much.

[Slugworth]

Re: Isn't a number that is almost infinite, er, infinite?

"In the case of 0.999(rec) there is no value "b" which can be placed between it and 1"

(1 - 0.999...) / 2 will fit nicely.

[Vic]

Re: Isn't a number that is almost infinite, er, infinite?

> (1 - 0.999...) / 2 will fit nicely.

It won't...

Vic.

[Annihilator]

Re: Isn't a number that is almost infinite, er, infinite?

"(1 - 0.999...) / 2 will fit nicely."

Which would equal 0.

@Numptyscrub, thanks, didn't know there was a wiki on this subject, but it has my favourite demonstration. Namely that 0.999999r = 9 * 0.11111111r. 0.11111r is 1/9, so 0.999999r = 9 * 1/9 = 1.

Or, if you subtract it from 1, you get 0.0000000000r and never reach the '1' at the end.

[Benchops]

Re: Isn't a number that is almost infinite, er, infinite?

Decimal expansion is just one way of representing numbers. We're familiar with others, e.g. fractions, but there are deficiencies, e.g. can't represent sqrt(2). Also we would quite happily realise that 1/2 and 2/4 actually represent the same number. It's not just that they're equal -- they're just different representations of the /same/ number.

Well decimal expansion is just another way of representing numbers. It has its own deficiencies, e.g. you can't concisely represent 1/3 (without a recurring symbol), or sqrt(2). It also has multiple ways of representing the SAME number, e.g. 1.0 and 0.999... It's not that they're equal numbers, they're the same number written down different ways.

[My Coat]

Re: Isn't a number that is almost infinite, er, infinite?

0.999 recurring is equal to 1.0, not almost equal. Here's a test: 1-0.999… equals zero.

[Anonymous Coward]

Re: Isn't a number that is almost infinite, er, infinite?

That's just nonsense Bob. For example;

"Did you guess the right number?

Almost. I guessed eleven, but the answer was ten"

And the inappropriate phrase you're looking for is "not quite", not "not quiet", which actually means "noisy"

[Sir Runcible Spoon]

Re: Isn't a number that is almost infinite, er, infinite?

" "not quiet", which actually means "noisy""

I would have thought it was more 'loud'. Something can be loud and not noisy if it is noise that you like, but something can be quiet and noisy if you don't.

[Anonymous Coward]

Re: Isn't a number that is almost infinite, er, infinite?

> I would have thought it was more 'loud'. Something can be loud and not noisy if it is noise that you like, but something can be quiet and noisy if you don't.

Erm ... yes, but Steven Fry owns a huge number of ties that are loud without being noisy.

[TkH11]

Re: Well...

Uh no, a number which is almost infinite is finite. If a number is infinite then you can never run out and you have nothing to worry about, a number which is almost infinite is finite and you can run out.

In the first scenario you have nothing to worry about, in the second you may.

[Anonymous Coward]

Re: Well...

A number which is almost infinite, is not. It is billions of times less than infinite, or is that trillions less, in fact isn't it infinitely less than infinite?

[Terry 6]

Re: Well... FWIW

One for the mathematicians, but surely if it's infinitely less than infinite it's also infinitely close to infinite.

And since there are infinite numbers of infinites, almost infinite could be infinite

After all the numbers in the infinite set of odd numbers are all just one + or - the numbers in the equally infinite set of even numbers.

[Stephen W Harris]

Re: Well...

There's actually no such thing as "nearly infinite".

For a number to be "nearly infinite" it must be a finite distance ("delta" away, and so has a value of "inf-delta"... which is, itself, "inf". So any "nearly infinite" number is, itself, infinite... contradiction.

Alternatively, all finite numbers must be infinitely far away from infinity and so are clearly not "nearly infinite".

[DiViDeD]

Re: Well...

Or to put it another way....

'Since any finite number divided by infinity is as close to zero as makes no odds it can easily be proved that the total population of the universe is zero and that any of the people you think you see walking around can be put down to an overactive imagination or having had one too many Pan Galactic Gargle Blasters with Zaphod Beeblebrox'

[Purple-Stater]

Re: Well...

So, it's not quite infinite you say? How much less than infinite is it? Perhaps 1,000 less?

So: Infinite - 1,000 = <insert dramatic pause> Infinite

Hmmm, looks like infinite.

[Loyal Commenter]

Re: Well...

I think that Cisco can be forgiven, because that number, whilst not infinite, is practically infinite. The difference being that whilst the number of addresses might have a limit, it's probably bigger than the number of things you could practically do with those numbers.

A quick back-of-fag-packet calculation shows that it is sufficiently large to assign over 100 million unique addresses to every atom in the solar system, which, until we have mastered superluminal travel and for some inexplicable reason decode we need to individually address every particle in the galaxy should do us fine.

[Anonymous Coward]

Re: Well...

Fuck sake people, infinite does not mean "a really really big number", and isn't something a number can "almost" be, or "nearly" be. A number is either finite or isn't.

IPv6 addresses are not "practically" infinite, because that's meaningless, as in, the term makes no sense.

You can't just put an adverb in front of "infinite" in a sentence and think that means something.

All it means is you don't actually understand what "infinite" means.

[NumptyScrub]

Re: Well...

quote: "Fuck sake people, infinite does not mean "a really really big number", and isn't something a number can "almost" be, or "nearly" be. A number is either finite or isn't."

Actually, set theory gets really complicated and introduces various different sizes of "infinite", including the concept of a set being countably infinite vs "uncountably" infinite.

Since IP addresses are integers, the set of all IP addresses would have cardinality Aleph-naught, and an upper bound of Aleph-naught (aka omega, the least infinite ordinal). Feel free to read the wiki pages, because I'm doing a Fry and presenting information as fact whilst simultaneously not understanding it properly myself. Maths is hard.

quote: "Q: Is 0.999 recurring equal to 1.0"

A: Yes. Yes it is

[Loyal Commenter]

Re: Well...

I think that once you get to numbers beyond the scale of anything you can actually measure with them (such as numbers smaller than the Planck length, or larger than the scale, or number of things in, the universe), it doesn't matter whether that number is infinite (or infinitely small), or not. Unless you are a pure mathematician.

For the needs of humanity, IPv6 addresses fall firmly into "who cares whether it's infinite or not, we'll never run out" territory. They are numbers which are intended to be assigned to a physical, addressable, entity, the countable number of which could never conceivably reach the limits of that address space. From a theoretical point of view, that number is not infinite, but for the other 99.9999% of humanity who do not care to draw a distinction between uncountably large and infinite in scale, it doesn't matter one jot.

[NumptyScrub]

Re: Well...

quote: "For the needs of humanity, IPv6 addresses fall firmly into "who cares whether it's infinite or not, we'll never run out" territory. They are numbers which are intended to be assigned to a physical, addressable, entity, the countable number of which could never conceivably reach the limits of that address space."

IPv6 convention splits the 128-bit address space into a 64-bit network address (class X subnet metaphor) and 64-bit device address (which can completely encapsulate any existing IPv4 address in the first 32-bits, should translation be required). What this actually means is that we'll actually be assigning a 64-bit range publically (minus the conventional loopback / broadcast / multicast ranges of course) and let routing kit deal with the internal 64-bits.

So the whole thing has 2¹²⁷ unique values, but potentially only 2⁶³ "assignable" (as in by IANA) values, and a whole slew of dead space, since my home router is not going to need to address 2⁶³ devices on my internal network but will get assigned a network ID, and all the unused ones are just as "wasted" as the unused IPv4 addresses in one of the existing assigned class As.

I completely agree that 128-bits (even split 64/64) is more than enough for a planet, but I'd also suggest tacking more bits on once you have to deal with multiple celestial bodies containing addressable objects, maybe an extra 64-bits defined as the "planetary" identifier, and then 64-bits for the "galactic" granularity level, giving 2⁶³ objects per subnet grouping, or 9.2e18 devices per 9.2e18 networks per 9.2e18 planets per 9.2e18 galaxies. 256-bit network addressing should in theory let us deploy to a significant portion of the known universe using a homogeneous routing backbone, and still have unique identifiers per device even with the whole thing running DHCP.

quote: "There's actually no such thing as "nearly infinite".

For a number to be "nearly infinite" it must be a finite distance ("delta" away, and so has a value of "inf-delta"... which is, itself, "inf". So any "nearly infinite" number is, itself, infinite... contradiction."

Have a read of the thoroughly confusing wiki page on ordinal numbers to see why set theory defines ω (equal to the cardinal value ₀א) is called the "least infinite ordinal" and thus why ω+1 (infinity plus 1) is considered a perfectly valid term. There is other stuff regarding the infinite different sizes an infinite set can be and all sorts of other horrible maths in there.

The good news is I found out where omega and aleph were hiding in Character Map ^^;

Edit: the bad news is you can't put a zero as a subscript on the correct side of an aleph. Oh well :(

[harmjschoonhoven]

Re: Q: Is 0.999 recurring equal to 1.0

That is where L. E. J. Brouwer's mathematical philosophy of intuitionism kicks in. It simple forbids "Gedankenexperimente' as "0.999 recurring".

[Mark 85]

Re: Well...

Let me run a quick math... Infinity - 1 = Almost Infinity. ??? or maybe it equals "damn near infinity".

[Annihilator]

Re: Well...

"Isn't a number that is almost infinite, er, infinite?"

Depends. Some numbers are infinite, yet finite. Such as 0.99999 recurring, which is actually equal to 1.

Similarly, infinite sets can be of different size (compare integers with real numbers - both infinite)

[TRT]

Re: Well...

I think the poster really meant practicably infinite, rather than practically infinite. It's the difference between an all-you-can-eat buffet and an infinite buffet.

[Stephen W Harris]

Re: Well...

You're confusing an expansion of a finite number with the value of the number. "PI" and "sqrt(2)" are both finite numbers (we can bound them; eg 3<PI<4) but any expansion of that number would take infinite space.

The "decimal expansion of 1/3" is infinite (0.333 recurring) but no one would claim 1/3 was infinite :-)

[Rogue Jedi]

Re: Well...

Assuming I have not miscalculated there are 28,147,976,710,656 possible IPv6 addresses, properly written out that is:

Twenty eight trillion, one hundred and forty six billion, nine hundred and twenty six million, seven hundred and ten thousand and fifty six possible IPv6 addresses,

this is far from infinite at about forty thousand addresses per person on the planet

[Badvok]

Re: Well...

"Assuming I have not miscalculated ..."

You have.

[Loyal Commenter]

Re: Well...

Assuming I have not miscalculated there are 28,147,976,710,656 possible IPv6 addresses

The actual number is (approximately) 3.4 x 10³⁸, which written out is:

340,000,000,000,000,000,000,000,000,000,000,000,000

I don't believe you could write it out in words (at least not ones people would recognise).

[JDM]

Re: Well...

I think its a little closer to 340,282,366,920,938,000,000,000,000,000,000,000,000 (340 undecillion)

[JDM]

Re: Well...

I think its a little closer to 340,282,366,920,938,000,000,000,000,000,000,000,000

[Purple-Stater]

Re: Well...

"Isn't a number that is almost infinite, er, infinite?"

This is exactly correct, for all useful purposes.

For the pedantic, yes, almost-infinite is not infinite but, again, for useful purposes, when the numbers you have available are more than you will conceivably use... that's as good as.

Now, one must continue being pedantic, else we'd not be able to ridicule Mr. Fry, who's simply speaking to the tech-commoners. It's rather like listening to people who love heavy metal music talk about why Justin Bieber is so bad.

[Anonymous Coward]

I am sick to death of the bashing Fry gets on the register. Well done, hes not an expert in IT, but guess what, he has never claimed to be (quite the opposite in fact). Fry has a fascination with technology and likes to use it but admits he himself has no clue how most of it works. It is the press that has painted him as some sort of ambassador for technology and all this article is is an attack against a misconception perpetrated by the gutter press and poorly researched content lacking articles on "technology news sites".

[Anonymous Coward]

Are you new around here?

[Anonymous Coward]

The point is people treat him as if he is.

[Anonymous Coward]

Well that's hardly his fault,

...so why slam him?

Is it actually all about trying to inform a nation who loves to think as Mr Fry as a full-on techy, and isn't aimed at him directly? 'cos it seemed pretty well targeted at him to me.

No of course Mr Fry is a pseudo-techy, probably knows enough about tech to get by, and even to apparently fix the PCs of his luvvie celeb mates, has no real tech depth, but rather ironically I'd actually say the same about The Register journos slamming him, so, possibly some pot...kettle...black syndrome kicking in a bit, I dunno.

[tin 2]

Nah. I agree with the first part: the bashing is unpleasant, and he is not an expert in IT, but the problem here is by the very existence of the article (and the previous faux pas on QI) that he is holding himself out to be one. No need at all to mention IPs and running out or otherwise of them, unless one is trying to be a clever-clogs.

[g e]

Except that

QI portrays him as a font of all wisdom, a veritable tree of knowledge and people believe it.

Which means he needs to be corrected - especially as his track record seems to be not to give a toss and fact-check after whatever his previous drubbing was. And the one before that. And before that. And so forth.

A sign of intelligence is supposed to be learning from your mistakes. Nothing against the guy at all, he's a very entertaining fellow and quick-witted but desperately needs to know when he's frothing in a field about which he knows cock-all.

[TkH11]

I partially agree with you. Fry is a very clever guy but not in the area of technology (heck, we can't all be engineers and understand how the tech works). I have watched him on tv be supportive of technology, he clearly finds it fascinating, and I think this is good, the fact is, most celebs don't give a hoot about tech and have no understanding of it.

Last night, I watched Michael Mosely in an episode of "The Story of Science" cover a little on Quantum Theory, valves, transistors and the integrated circuit culminating with his saying that he considers the transister to be in the top 10 of all inventions ever created by mankind.

He left the audience with no doubt as to the importance of Quantum Theory and how transistors could not have been developed without an understanding of it.

This is the kind of publicity that the tech industries need. We need far more than that.

The UK general public is in my opinion, technologically illiterate, sure they know how to use tech, but they know nothing of how it works, and anyone that works in the industry is a geek.

in countries such as India, the USA, anyone in the tech industry has a far higher level of respect shown to them by the public. That's where we in the UK need to be.