Talk:Internet Protocol

From Citizendium
Jump to navigation Jump to search
This article is developed but not approved.
Main Article
Related Articles  [?]
Bibliography  [?]
External Links  [?]
Citable Version  [?]
To learn how to update the categories for this article, see here. To update categories, edit the metadata template.
 Definition Highly resilient protocol for messages sent across the internet, first by being broken into smaller packets (each with the endpoint address attached), then moving among many mid-points by unpredictable routes, and finally being reassembled into the original message at the endpoint. IP version 4 (IPv4) is from 1980 but lacked enough addresses for the entire world and was superseded by IP version 6 (IPv6) in 1998. [d] [e]
Checklist and Archives
 Workgroup category Computers [Categories OK]
 Talk Archive none  English language variant Canadian English

Note that Internet Protocol version 4 also exists. Chris Day 03:32, 9 June 2008 (CDT)

As does Internet Protocol version 6. This article deals with commonalities and requirements for the later version. DNS is a technology complementary to both.

There are more detailed articles under the main IPv6 article, and also on complementary techniques such as DHCP.Howard C. Berkowitz 16:24, 17 January 2009 (UTC)

Reorganization from earlier article

I have this pointing to both IPv4 and IPv6, and removed a good deal of text which tried to explain IP in terms of the Open Systems Interconnection Reference Model, rather than the applicable Internet Protocol Suite. Howard C. Berkowitz 20:43, 14 July 2008 (CDT)

Shouldn't the stuff about "L2" in Internet_Protocol#Local_versus_remote_principle go as well? Sandy Harris 05:37, 23 July 2009 (UTC)


I do cite several papers and presentations of which I was sole or co-author. All are peer-reviewed and are formally public domain, or, in the case of the presentation, was given to a public forum and no copyright was claimed. Please let me know if there are any concerns.

I have written professional books on addressing and routing; others may want to put them into the bibliography if they see fit. The 1998 book is more of historical interest; although it does consider both IPv4 and IPv6, it also discussed the plethora of proprietary protocols they replaced. Howard C. Berkowitz 16:21, 17 January 2009 (UTC)

Approval readiness

Given this is a "core article", with IPv4 and IPv6 arguably major sub-cores under it, and that it also is under the core Internet Protocol Suite and other architectural articles such as locality of addresses, I'd like it considered for approval. Editors, please consider its level of detail with respect to those other articles. Howard C. Berkowitz 16:28, 17 January 2009 (UTC)

What the devil does "protocol" mean in this sense, anyhow?

I've been using computers for 25 years now, and the Internet for 13 years, and I know my way around both of them fairly well in a rough and ready way. I'm also a college graduate with a degree in English, and I *still* don't know what "protocol" means in this sense. Since the word is being used about a dozen times in the first 100 words of this article, I think it would be not only a favor, but also extremely useful, to define it for the long-suffering reader at some early point in the article. Thanks. Hayford Peirce 22:03, 18 June 2009 (UTC)

Protocol (computer), linked in the lede sentence. If you prefer, the words of the magical spell. If one incorrectly follows a magical protocol, ribbit. ribbit. Howard C. Berkowitz 22:07, 18 June 2009 (UTC)
Yes, I saw that, and I reject the concept. If an article is called Spamity Leafton, for instance, I think that both words should be defined *within* the article, NOT by links. Maybe I'm just feeling particularly grouchy and stupid today, but that's how *I* feel about it, and I'm sure that there are lotsa equally grouchy and equally stupid people out there who would agree with me. Remember, this encyclopedia is still supposed to be written in normal English, with normal English conventions, not just in computerese because we have the ability to do so. Hayford Peirce 22:16, 18 June 2009 (UTC)
I reject the rejection. At the level of end users such as yourself, it should be reasonably obvious that a protocol is a set of rules by which the computer operates, just as it is a set of rules for diplomatic interaction. To get into how a computer protocol works, however, does take some preparation in principles of networking (e.g., computer networking reference models), formal logic and automata theory, binary arithmetic, and, quite likely, knowledge of operating system principles. I've taught IP for close to twenty years, and I have never been able to beyond "rules" without some preparation. The subset of the concept that is an IP address can, very loosely, be likened to a phone number, but that's about the limit of simplification.

This article, incidentally, is far more accessible to nonspecialists than many in mathematics and medicine. I'm not a beginner, and I'm still trying to find an absolutely clear statement of the relationship between an adrenergic beta-agonist and a sympathomimetic, although I can use the concepts. Howard C. Berkowitz 22:30, 18 June 2009 (UTC)
Well, nevertheless it will not hurt to start with: "The Internet Protocol (IP) is a precisely defined set of rules (a protocol) which describe how computers on the internet have to communicate and interact and how they pass information from one to another" (or so) Peter Schmitt 22:42, 18 June 2009 (UTC)
I disagree. That doesn't identify the critical aspect that IP is one of many protocols required to make the Internet work; it oversimplifies. I'd be willing to extend It is the protocol on which the Internet is built to add by providing a framework over which packets of information can be transferred, independently of the types and numbers of underlying physical media and administrators. Howard C. Berkowitz 22:47, 18 June 2009 (UTC)

But this again uses the "incriminated" word without explanation ... another attempt:
The IP is a set of precisely defined rules (one of many protocols required to make the Internet work). It is the protocol on which the Internet is built by providing a framework over which packets of information can be transferred, independently of the types and numbers of underlying physical media and administrators.

It's an improvement, but I remain unconvinced it's necessary. Take, for example, continuum hypothesis. In the first paragraph, the idea of a set, natural numbers, real numbers and perhaps cardinality are implied prerequisites, not even linked as I did protocol (computer). Mathematics articles often are simply not accessible without background.
There's a serious tradeoff between having linkable definitions, and having a lede that flows for someone with at least some background. In my articles on intelligence, for example, I link to terms of art. Howard C. Berkowitz 23:40, 18 June 2009 (UTC)
When I read the introduction I didn't notice any problem ("protocol" seemed to be clear). I just try to mediate.
Concerning the continuum hypothesis (though this does not belong here): There may well be a need for more links -- but I certainly hesitate to link natural and real numbers -- they should be known in the common "naive" sense, at least (if not, a link probably will not help, while countable set should help.) I tried to avoid as much technical terms as possible (and links that lead in the wrong direction), but you can't avoid natural and real numbers (or equivalent).

(This was written before the text below -- edit conflict) I shall have to read it tomorrow)Peter Schmitt 00:20, 19 June 2009 (UTC)

Peter is attempting to make the article more user-friendly, and accessible to idiots like me, while you, Howard, seem to be arguing that it should remain obfuscated. Why? Aren't you the person who, within the last week or so, argued eloquently that a lot of gearheads, programmers, and longtime computer-types simply didn't recognize the fact that most people in the world didn't have their background and that the simple, elementary stuff (to them) was incomprehensible nonsense to the rest of the world? Why not try a simple test: walk into a shopping mall, grab 100 adults at random, and ask them to tell you, in twenty words or less what "protocol" means? If five of those people can give you an even approximate answer I would be astonished. Hayford Peirce 23:52, 18 June 2009 (UTC)
Apparently, I did not explain well the point last week. My comment on "simple elementary stuff" was directed to the computer scientists, to get across the idea that the simple stuff is not and will not be. Templates are unlikely ever to be at a level where the average person can pick them up and use them, without endangering the wiki.
It's a Rodney Dangerfield sort of problem. Some disciplines, computer science being among them, don't get the respect that their terminology is nuanced and needs hard work. As I just pointed out to Peter, no one seems to expect all mathematical articles to be understandable without background. We have a large number of medical articles that will not be accessible without an understanding of biochemistry and physiology. They don't remotely have the linked definition such as protocol (computer).
There are areas in computing, and indeed other fields, where the proper understanding of someone who does not have the background is "much magic. Gods say leave alone." Think, now, of the beginning computer science student, who does have the background -- at what point does "user friendliness" make it hard to read a focused introduction? Apropos military, there is a British tradition that while commanding a warship takes a lifetime of expertise, "any fool can command a regiment." Library and information science, I've noticed, is another such area, which many assume anyone can do and doesn't have nuanced rules. Howard C. Berkowitz 00:00, 19 June 2009 (UTC)

Vanishing it

I fail to see the improvement by "set of rules", given that if necessary, that's a single click. I'm not going to remove it out of hand, but I think it's a hand-waving oversimplification that adds nothing. Had it been that straightforward, I would not have written protocol (computer).

What are you going to do with the information that it's a set of rules? What does that tell you more than the Common Law is a set of rules? Howard C. Berkowitz 00:04, 19 June 2009 (UTC)

At the very least, it tells people like me that Protocol in this sense is NOT a single book somewhere, such as "The Protocols of Zion". It tells me that it MAY be similar in some way to the notion of "Court Protocol", or "Diplomatic Protocol," but that it MAY also be more structured than that. It tells me that it is NOT simply a mechanism to download cool new songs and videos. It tells me that it is NOT a blueprint for designing a new Website. It tells me that it is NOT the same as Internet Explorer or some other browser. It doesn't tell me a LOT, but it at least gives me a LITTLE more of a notion of what the article is going to be about. "Internet Protocol" is NOT, I think, Rocket Science in the sense that Boris's new article about the Non-Borel set is, where he uses "set" three or four times in the first sentence or two. I think there are billions of people in the world who can lead useful lives without knowing about sets, whether they are Borelian or not. Internet Protocols, however, are something that *many* of us have heard about, and may have even wondered about, since many, many millions of us are using the Internet every day. It therefore behooves the authors of the article not to have it start out like (and I do exaggerate here for effort, I'm not trying to be rude or procovative), "A porterhouse steak is a steak that is larger in some ways, and tastier, than other steaks, at least to those people who eat steaks." Please -- if we have a sentence like that, let's at least make sure that *everyone* is clear on the "what is a steak, anyway?" concept.... Hayford Peirce 16:31, 19 June 2009 (UTC)
You see, I'm also thinking of the person who already knows that and wants to focus on what he doesn't know. To me, the nature of a hyperdocument, as opposed to a hard copy, serves both by having the link to protocol (computer).
As far as rocket science, I wrote that, and I remember some engineers in my class at Cape Kennedy thinking this was more complicated. I'm quote serious that there is, as we saw with templates, a hope or belief that a complex matter in a given field will be understandable to the beginner with just a bit more experience. Long ago, I gave up on ever trying to play tennis. Howard C. Berkowitz 16:48, 19 June 2009 (UTC)

Summary - and a suggestion

I try to summarize the arguments expressed above (I think I understand both sides) and what I think.

Whoever wants to read the whole article can be expected to know the meaning of protocol in this context, or to read the linked article if he wants to learn about it.
But the introduction should be as readable as possible for someone who only wants to learn what the Internet Protocol is, and in this case following the link could be too much.
To give a good explanation would be too complicated if one wants to be precise, and I understand that Howard does not like to be vague.
But I also understand that Hayford wants some help to understand protocol, in particular, since protocol is not a "strange" scientific term, but a common word (obviously?) used in a special sense (and therefore could be easily misunderstood).
Moreover, "protocol" is repeated several times. This draws the attention (and is also a stylistic question).
One could try to avoid the use of "protocol" (except in the names of the protocols, of course). The they need not be explained (and the link would come later or could go in "Related articles"). Instead of explaining "protocol" in general, only "Internet Protocol" has to be explained -- and that has to be done anyway.

I give a example to show what I mean (but the wording could be improved):

The Internet Protocol (IP) describes/defines the standards used for communicating across a heterogeneous network. It is the basis/fundament on which the Internet is built.

Peter Schmitt 23:48, 19 June 2009 (UTC)

Peter, I think you have summed up the positions (and my own position) perfectly! Please effectuate something along those lines -- I will be away from computers until Sunday evening, so this is the last thing I can contribute for a couple of days. Best, Hayford Peirce 23:53, 19 June 2009 (UTC)

prefix aggregation

I am not sure of the intended meaning of the following sentences:

"Only when the call reaches the area code level of switch does the telephone routing process look for the additional information pointing to an exchange. Only at the final exchange does the switch look for the specific line. <brY The example used at the very beginning of the locator section, with a fixed-length /8 prefix field, had the actual im used the continent-country split, could have aggregated addresses: the continental link could be selected just on the /4 high-order bits of the address."

Peter Schmitt 23:36, 21 June 2009 (UTC)

3 unexplained terms

What does "hub-and-spoke", "avian media" and "carrier pigeon" mean? Are there links missing? Peter Schmitt 23:41, 21 June 2009 (UTC)

I'd have thought "hub-and-spoke" was a fairly obvious metaphor for describing how things are connected, the network "topology". However, I'd say if we're going to use "topology" in that sense here, then it needs an explanation or a link. In any case, I'd say that entire section is unnecessary & confusing.
"avian media" and "carrier pigeon" refer to data transfer by bird, a bit of an in joke referring to RFC 1149. I'd say the comment is necessary; it makes an important point about just how "medium agnostic" IP is. However, it might be stated more clearly. Sandy Harris 05:55, 23 July 2009 (UTC)


I do not feel this is straightforward enough, at the level of an encyclopedia article intended to be readable by non-specialists. Of course in this context "protocol" needs some explanation; I'm amazed anyone could think otherwise.

I'd say the article should start with a very basic explanation of the role of IP — given a bunch of networks, create one big network. That leads directly to IP addresses and the idea of being agnostic about the underlying network technologies.

Then talk about how to send a packet — either send directly on your local network or give it to a gateway if the target is not local. Explain the netmask or address prefix in terms of that. Illustrate with something the average end use can see on his PC.

Then talk about gateways and routers. IP is hop-by-hop and not based on "connections". Routers have neither storage nor state. TCP gives end-to-end connections. Routers may use prefix aggregation beyond the basic is-it-on-my-net level.

The whole section "Address structure" strikes me as confusing. Parts of it are quite good, but I have the feeling it could be clearer overall. It starts with "There are two fundamental parts of an address: the locator and the identifier." I've never seen either term in a networking context before; am I just out of it? Sandy Harris 06:36, 23 July 2009 (UTC)

What you are saying is not irrelevant, but I believe the core article should be "technical", but with a very clear tutorial subpage or subarticle. The role of IP really belongs in a more general context such as Internet Protocol Suite. You're describing a networking tutorial, hardly limited to IP but getting into routers, etc.
Would this be a service to the networking student who wants some nuances of IP, noting there are subordinate IPv4 and IPv6 articles, or a disservice for the PC user? Where does the new networking student go to understand addressing? I think you are confusing things that should be in "networking" with things specific to IP, such as the critical IPv4 problem of overloading locator and identifier onto the same address. As I remember, it was Noel Chiappa that first made that observation in the NIMROD project. Howard C. Berkowitz 15:58, 23 July 2009 (UTC)


Howard first suggested this might be ready for approval above, just over two years ago. Now I am looking at it because I've recently become an editor, and this is he only article at CZ:Ready_for_approval#Computers where I'm not an author, so I could conceivably nominate it. However, I'm not ready to nominate it, so I'm starting a discussion instead.

Several reference links were broken. I fixed some, but not link 8 which points to a NANOG (North American Network Operators' Group [1]) paper. I found only an abstract for that.

Howard is cited three times, twice as an RFC co-author & once as sole author of the NANOG paper. I think that's OK; they are relevant publications in highly relevant places. However, we do have a policy against self-citation, so it seems worth asking for opinions from more experienced editors. Do we need to have the constabulary rule here? (I think not.)

I regard the questions in the three immediately preceding sections — sections #prefix_aggregation. #3_unexplained_terms and #re-organise.3F — as unanswered and in need of being dealt with.

Howard says above "You're describing a network tutorial." I thought I was describing a reasonable structure for an encyclopedia article on IP, at the correct level for such an article. Anyone else care to comment, or Howard to rebut?

The paragraph in section "Locator" starting "What is important ..." and describing classful routing seems confused to me. Much of it is fine, but it uses the /8 /16 /24 notation which was developed later as part of CIDR (Classless inter-domain routing), is not in the RFC cited, and does not belong here. (Or at least, not without a disclaimer: Today we would call these /8, /16 and / 24 networks.) On the other hand, it fails to mention classes A, B and C, or to mention how many of each type were possible. Only 256 Class A, for example, allocated to government and large companies.

The section on "Local versus remote principle" is not clear. It talks about "Layer two" and "L2" without defining them. Don't bother defining them, though; those are OSI terms and do not belong here. Anyway, it would be better to talk of "on your local network" or "directly connected" or some such.

Mentioning IP using carrier pigeons is fine; a great demonstration that it is medium-agnostic. However, providing three links here on that topic is excessive. Ideally, move the links to an article on IP on avian characters (or create that title as a redirect to a section of April 1 RFCs) and wikilink to it from here. Alternately, just reference RFC 1149. Sandy Harris 13:55, 21 February 2011 (UTC)

All good points. I agree with the discussion, although I don't have time at this moment to have it. I will be glad to do so if I'm still around.
I have to check the NANOG reference, but a number of their older presentations have become corrupted on their archives and I don't still have a copy -- they remain matters of record. We can certainly check to see if they have other copies, or, in some cases, ARIN might have a version. Howard C. Berkowitz 17:36, 21 February 2011 (UTC)
Good idea on April 1 RFCs, and maybe on network agnosticism--the latter can actually extent into issues such as the Interplanetary Internet. Interplanetary Internet ideas are, in fact, being used with one real network with extremely long delays -- Lapp reindeer herders who can connect every two weeks or so. Avri Doria is the co-principal investigator, which, as I remember, is under the University of Lulea (sp?) in Sweden. Howard C. Berkowitz 17:39, 21 February 2011 (UTC)