>The point made with the help of Vinton and others was that the
>"Internet" or even the "internet" has to do with the interconnection
>of networks using a commonly agreed protocol, rather than the
>protocol in use by the individual constituent networks themselves.
That's true to a point... I would say a commonly agreed set of protocols.
And to really unravel the statement, it would be necessary to get into the
protocol layer stack, which has already been referred to in this thread.
Without going into the full ISO 7-layer model, which the Internet only
approximates in any case, it's useful to differentiate between:
-- transport protocols (such as PPP and Ethernet) which are only relevant
to the particular machines on that link,
-- network protocols (IP, TCP, UDP, etc.) which, generally speaking, must
be adhered by all machines ("hosts") on the internet, and
-- application protocols (SMTP (email), HTTP (World-Wide Web), FTP (file
transfer), etc.), which are only relevant to those machines providing and
using the services. (See Note 1, below.)
IP is the "heart" of the Internet (or of internets); indeed, it stands for
Internetwork Protocol. If a machine is "on the Internet", it must have a
physical connection to at least one other machine "on the Internet" and it
must use IP as its networking protocol. This is generally translated into
the statement that the machine must have a unique IP number; in fact, as
several people have pointed out, this is slightly imprecise: first, the IP
number needs only to be unique while it is being used; second, firewalls
can perform address translation in order to allow multiple machines to use
the same IP number. (Also, it's actually the network interface -- the
network card and/or dial-up connection -- which has the IP number. A "host"
(machine) will generally have one IP number per network interface, but it
can have more.)
A firewall could, theoretically, translate IP into some other protocol,
allowing machines which do not implement IP to use some other networking
protocol. However, that protocol would have to be a logical superset of IP
(and additional capabilities would not be functional). In practice, most
networking protocols (Appletalk, IPX/SPX, NetBEUI, etc.) are not that
easily mapped into IP. However, this sort of emulation is being proposed as
a step towards implementing IPv6, the "next generation of IP".
I am quite free to devise new transport and application protocols, though.
A new transport protocol might be used, for example, to enable the
transmission of IP over some new medium; perhaps a new radio system. A new
application protocol could be devised to provide new and different
services: in fact, these are proposed and even implemented quite often.
Such protocols are, by convention, initially defined by RFCs (Requests for
Comment) and then refined by use and discussion; if they are successful in
the sense that they are widely used, they become de facto standards and may
eventually become Internet Standards.
All of the above is a slight over-simplification, but I hope it captures
>We thus return full circle to a question of interest to me. What
>does it mean, really, to be "on" the Internet?
I hope I've answered that above. But I fear that it is not a sufficient
answer for your work, Jeff, particularly since it only answers the question
of what it means for a machine to be "on the Internet", not a person.
I think that these are really two completely different questions. Whether a
machine is "on the Internet" is a technical question, although there are
shades of grey. Using that definition, a person cannot be "on the
Internet" unless s/he has a very sophisticated electronic implant (with,
presumably, a built-in mobile phone). (But see Note 2, below)
When people say they want to be "on the Internet", they mean something
different. I share a bit of Dr. Lisse's cynicism about people meaning that
they've seen an article in the popular press, but I mostly think that it's
only fair to take people's expressed desires at face value. So I would
translate "I want to be on the Internet" to mean "I'd like to have access
to the services that are available to people who use computers which are on
the Internet." Which is what you say below...
>Ultimately we are or ought to be about the exchange of information.
>Hence it is my contention that one is "on" the Internet once one
>takes advantage of its services.
But there is more to the story than that.
While I now work for Oxfam, trying to install Internet access in Africa
(amongst other places) (and, yes, my work is confined to Oxfam offices and
I am keenly aware of the inherent problems and contradictions), I came to
the Internet by working to bring it to a different challenged region of the
world: the Canadian Arctic. Northern Canada certainly has resources not
available to many parts of Africa (for one thing, a somewhat supportive and
relatively wealthy federal government) but it shares other conditions:
difficult communications, a largely aboriginal population whose first
language is not English, transportation difficulties, a generally low
standard of living, and so on. For us (and I speak as a long-time
Northerner, which means something a bit different when you come from north
of the Arctic Circle), being "on the Internet" does not just mean having
access to services. It means being full and equal participants. It means
being both producers and consumers. And therefore it means owning and
operating Internet services (and not just English-language services!)
That was much more of a challenge. An Internet consumer can get by with a
low-speed, intermittent, and possibly firewalled dial-up connection. But an
Internet producer requires a constant high-speed Internet presence. It took
some work to make our federal government understand the difference
(including turning down the offer of a few free dial-up lines, which was
their original offer.) Indeed, it took some work to make the indigenous
business community understand the difference, and that work is still going
on. It's very tempting to price Internet access -- and particularly
high-speed permanent connections -- far out of the reach of
micro-enterprises (even micro-enterprises which have somehow managed to
acquire a computer and a consistent electrical power supply). But you can't
improve an economy by (in Shalom Aleichem's words) "eking out a living
taking in each other's laundry" -- in today's terms, becoming an indigenous
Internet Service Provider. The real development activity must be exporting
knowledge and skills -- and that requires low-cost high-quality
So it's not good enough to say, make do with these crumbs for now, and
learn to use them before you progress. ("But I think we can agree that for
some places and at some times, even in Ethiopia, a few steps in between
here and there may be required.") Because you may just be teaching people
to be better consumers.
I'll get off the soap box now. Sorry to chew up so much bandwidth.
Note 1: Some services, particularly e-mail, are not restricted to the
Internet. (Hence the "matrix" of BITNET, FIDO, and so on). At the
application level, it is often possible to transform a protocol into
another protocol (usually with some loss of function, though). As many
people have pointed out, you don't have to be "on the Internet" to exchange
e-mail with others who are. And you can (although I wouldn't really want to
if I had an alternative) access a variety of Internet-borne services via
e-mail (FTP-by-mail, Web-by-mail, etc.). So the "reach" of these services
exceeds that of the Internet. (In fact, there are e-mail to voice gateways,
so you could say that e-mail encompasses the entire international telephone
Note 2: I actually have taught (fairly young) children about Internet
protocols by literally putting them "on the Internet" using a specially
crafted transport protocol: YPOPP, the Yellow Pad of Paper Protocol.
Unfortunately, many common higher level protocols (particularly DNS) time
out during the long RTT (Round Trip Time) of YPOPP, but with a bit of
ingenuity and adaptation, it is actually possible to send and receive
e-mail over YPOPP. I'd be happy to provide more details if anyone is
Recommended Reading: If you really want to understand how the Internet
works, I strongly recommend Richard Stevens' excellent "TCP/IP Illustrated
Volume 1" published by Addison-Wesley.
International Infrastructure Advisor
--but: the opinions contained above are my own, and any resemblance to an
opinion of any organisation I currently or previously worked for is purely
a coincidence (though a happy one!)