[ih] 'Internet' vs 'internet'

Joe Touch touch at strayalpha.com
Fri Oct 19 08:11:23 PDT 2018 


> On Oct 19, 2018, at 5:58 AM, Toerless Eckert <tte at cs.fau.de> wrote:
> 
> On Thu, Oct 18, 2018 at 09:41:38PM -0700, Joe Touch wrote:
>> Interesting and absolutely complex - but perhaps because it takes pains to explain all the varieties of ???how??? rather than focusing on a much simpler ???what???.
>> 
>> IMO, Internet access is defined by two properties, each of which can be ???full???, ???partial???, or ???none???:
>> 
>> FIRST CLASS - Internet protocol participant
>> 	these nodes can participate and Internet protocol network in any role (client, server, peer) for any service AND can reach the root DNS servers
> 
> Agreed. This is i think what your initial "On the Internet" means,
> and thats what i thought was easy and good to "The Internet" / "The IPv6
> Internet" 
> 
>> SECOND CLASS - Internet information access
>> 	these nodes can access information provided by any node in set (A), but are not themselves in set (A)
> 
> Right. And "Internet Access" might be the simple term to use here.
> Not sure about exact definition. "can exchange data" might be a
> better term because i fear "information" would be read by non-technical
> people more like "Facebook" or the like.
> 
>> Most home Internet service is ???partial SECOND CLASS???, i.e., can???t run as a server at all (so not FIRST CLASS) and some ports blocked for SECOND CLASS.
>> 
>> However, most hosted web servers are ???partial FIRST CLASS???, i.e., run as a server but not on all ports (most hosting services block certain ports). 
>> 
>> AFAICT, the only meaningful variants are:
>> 	full first class (which implies full second class too, trivially)
>> 	partial first class (which implies full second class because you can always contact a first class node on at least one port and get access to anything)
>> 	full second class (implies no first class)
>> 	partial second class (implies no first class)
>> 	no access at all
> 
> Why introduce partial first class ? Any form of data access to
> the Internet that does not fully meet the definitions of
> "On the Internet" is simply "Internet Access”.

Largely to allow for the case where some ports are blocked and to avoid a debate on “which ports” and whether they’re important or not. See below regarding home gateways.

>> 
>> Further, note that full first-class nodes can help other nodes become any class except itself. 
> 
> Well, the interesting explanations for laymen are something like:

To me, FWIW, a layperson only needs to know:

- can you control your content, how it is served, and how it is logged 100%? (first class)
- can you get at Internet data managed by others, either reading or writing that data, but not under your control? (second class)

We can’t get down into the definition of a ‘host’ for laypeople.

> A users computer is called a "Host" in the Internet technology.
> A Computer/Host is called "on the Internet" if its connection to the
> Internet meets the following requirements ...
> 
> A computer "On the Internet" can only extend the Internet
> to allow more computers to be "on the Internet" if it can
> become a "Router on the Internet”.

Strictly speaking, routers don’t need IP addresses themselves (unless they start also acting as hosts, e.g., to participate in protocols for in-band configuration, etc.)

> A home gateway for example
> can not do this when it just has IP because then it
> only gets one IP address and because to be "on the internet"
> every computer needs its own Internet IP address, the home
> gateway needs to give private IP addresses to computer behind
> it, granting them only more limited "access to the Internet”.

A home gateway doesn’t 'give away' addresses; it translates addresses and ports.

Arguably, if the public side of a home gateway has a real, public IP address, then NAT’d devices behind it CAN be ‘partial first class’, e.g., for some subset of ports assigned to each private-side host.

> With IPv6 on the other hand, the home gateway can become
> a router "On the (IPv6) Internet" and make computers behind
> it be Hosts "On the (IPv6) Internet”.

IPv6 alone doesn’t magically allow a home gateway to become a router; this still requires that your ISP not block the addresses of your “private side” hosts (which they can do, and I suspect will do unless you pay more money for “”commercial service”). Yes, the point of IPv6 is to allow that to happen, but ISPs can still interfere...

Joe 

> 
> 
>> I would thus define "the Internet" as "those nodes that are first-class AND connected to the DNS roots".
>> 
>> I would never say that second class nodes are ???on the Internet???, but rather ???can access the Internet???.
> 
> Right
> 
> Cheers
>    Toerless
> 
>> Joe
>> 
>> 
>>> On Oct 18, 2018, at 5:47 PM, Brian E Carpenter <brian.e.carpenter at gmail.com> wrote:
>>> 
>>> Toerless,
>>> 
>>> Have a look at RFC4084. To me, that explains why this is a more tricky
>>> question than you might think, and it might have wider implications for
>>> consumer protection, monopolistic behaviours, and whatever interpretation
>>> you put on the phrase "network neutrality".
>>> 
>>> Another interesting thing to think about is the question: How many
>>> hosts are there on the Internet? Historically (30 years ago) that
>>> was a meaningful question to which you could answer "about 56000".
>>> Today??
>>> 
>>> In the room I'm sitting in there are currently 4 devices switched on
>>> running TCP/IP. One of them has unique IPv4 and IPv6 addresses; three
>>> others have NATted IPv4 and native IPv6 addresses. So if you count the
>>> routed IPv4 Internet, there's 1 apparent host. If you count the routed
>>> IPv6 Internet, there are 3 hosts and a router.
>>> 
>>> Regards
>>>  Brian
>>> 
>>> On 2018-10-19 11:03, Toerless Eckert wrote:
>>>> IMHO it does not make sense at all to talk about the "Internet" as
>>>> a scientific/technical term (as opposed to pure marketing) unless
>>>> someone provides an agreed upon definition. The absence of a clear
>>>> definition ha always annoyed me.
>>>> 
>>>> I like the idea of defining "The Internet" as the set of IP hosts
>>>> that are "on the Internet" and the transit infraatructure
>>>> connecting them. Its also fine to add to such a document
>>>> definitions for "access to the Internet" such as via NAT,
>>>> application layer gateways or the like. Those add-on terms wouldn't
>>>> be so important and probably harder to categorize given all the
>>>> variety of constraints vs. being "on the Internet".
>>>> 
>>>> Why has nobody tried to revisit that subject in an RFC after rfc1775 ?
>>>> None of the discussion points on this thread seem to be blockers
>>>> but IMHO easily aligned. So i wonder whats the big blocker.
>>>> Just nobody who cares enough about precise terminology ?
>>>> 
>>>> Cheers
>>>>   Toerless
>>>> 
>>>> On Thu, Oct 04, 2018 at 05:40:40PM -0700, Joe Touch wrote:
>>>>> 
>>>>> 
>>>>>> On Oct 4, 2018, at 11:22 AM, Jack Haverty <jack at 3kitty.org> wrote:
>>>>>> 
>>>>>> Hi Joe,
>>>>>> 
>>>>>> Your criteria are a good description of what, IIRC, it meant to "be on
>>>>>> the Internet" back in the 80s.  But today, I suspect the vast majority
>>>>>> of people who think their computers/phones/devices are "on the Internet"
>>>>>> wouldn't meet one or more of the criteria.  So they must be all on
>>>>>> something else, if not the Internet?
>>>>> 
>>>>> Lots of people use airplanes to get access to goods without traveling on an airplane themselves. 
>>>>> 
>>>>> The same is true here. There???s a distinct difference between ???access to Internet information??? and ???Internet access???. The latter allows users to run their own servers; the former is dependent on a ???distilled??? product only.
>>>>> 
>>>>>> 
>>>>>> Meanings of words are determined by how people use them.   I think "the
>>>>>> Internet" changed meanings long ago, and continues to change.
>>>>>> 
>>>>> 
>>>>> Agreed, but lots of people learned the difference between AOL and Internet access too, One was distilled services presented through a specific interface; the other is extensible based on agreement of the endpoints. If we value that latter principle, we need to encourage the most complete Internet access we can - that???s partly what net neutrality is all about.
>>>>> 
>>>>> Joe
>>>>> 
>>>>>> /Jack
>>>>>> 
>>>>>> 
>>>>>> On 10/04/2018 07:33 AM, Joe Touch wrote:
>>>>>>> 
>>>>>>> 
>>>>>>>> On Oct 3, 2018, at 12:12 PM, Jack Haverty <jack at 3kitty.org
>>>>>>>> <mailto:jack at 3kitty.org>> wrote:
>>>>>>>> 
>>>>>>>> ...
>>>>>>>> 
>>>>>>>> At Oracle, we had our own world-wide internet, and assigned our own IP
>>>>>>>> addresses, regardless of whether or not the particular number was in use
>>>>>>>> in the public Internet.  But we were connected to the Internet through
>>>>>>>> computers which were dual-homed, and thus could receive email, use FTP,
>>>>>>>> etc. as needed. We could interact with the obvious players, e.g.,
>>>>>>>> Yahoo!, but also with computers inside our customers' private internets.
>>>>>>> 
>>>>>>> I would call that ???accessing Internet content???, but definitely NOT being
>>>>>>> ???on the Internet??? (note: I appreciate this also applies to nearly all
>>>>>>> consumer access because of NATs).
>>>>>>> 
>>>>>>> Being ???on the Internet??? IMO has minimum requirements; I presented these
>>>>>>> as candidate requirements at a meeting in 2004:
>>>>>>> 
>>>>>>> Internet User ???Bill of Rights"
>>>>>>> 
>>>>>>> The Internet is an association of communicating parties. Consenting
>>>>>>> parties should be able to communicate in an unrestricted fashion,
>>>>>>> insofar as they do not impinge on the corresponding rights of other
>>>>>>> parties. The following is a list of specific rights to that end:
>>>>>>> 
>>>>>>> 1. REAL IP: Users have the right to a real IP address, routable from
>>>>>>> anywhere on the Internet.
>>>>>>> 
>>>>>>> 2. REAL DNS (& REVERSE-DNS): Users have the right to a valid reverse DNS
>>>>>>> name for that IP address, and the forward lookup of that name that
>>>>>>> matches that address.
>>>>>>> 
>>>>>>> 3. RECEIVE ANY: Users have the right to receive any valid IP packet,
>>>>>>> using any valid transport protocol on any valid port (if applicable), up
>>>>>>> to the limits of their local resources and network connection.
>>>>>>> 
>>>>>>> 4. SEND ANY: Users have the right to send any valid IP packet to any
>>>>>>> valid real IP address, using any transport protocol, on any valid port
>>>>>>> (if applicable), provided it uses an inconsequential amount of resources
>>>>>>> of the network and potential receiver until mutual consent is established.
>>>>>>> 
>>>>>>> 5. ENFORCEMENT: Users have the right to know the ISP responsible for
>>>>>>> traffic from any valid IP address, sufficient to register a complaint
>>>>>>> regarding violations of any of these rules. 
>>>>>>> 
>>>>>>> ??????
>>>>>>> Everything else is, at best, access to Internet *information* but
>>>>>>> undermines the ability to participate directly in Internet protocols
>>>>>>> themselves. That???s sort of like saying you can watch TV, but only from
>>>>>>> still photos taken across the street through a smudged window.
>>>>>>> 
>>>>>>> Calling that ???the Internet??? isn???t evolution of terms to common usage.
>>>>>>> It???s misleading advertising.
>>>>>>> 
>>>>>>> Joe
>>>>>>> 
>>>>> 
>>>> 
>>>>> _______
>>>>> internet-history mailing list
>>>>> internet-history at postel.org
>>>>> http://mailman.postel.org/mailman/listinfo/internet-history
>>>>> Contact list-owner at postel.org for assistance.
>>>> 
>>>> 
> 
> -- 
> ---
> tte at cs.fau.de

More information about the Internet-history mailing list