Request for Comments: 0
A fictional memo
Anyone
Still open

THE MEANING OF THE INTERNET

Status of this Memo

This memo asks a question. It does not specify a standard of any kind, because the question is prior to all standards. Distribution of this memo is unlimited. That was always the best part.

1. Abstract

The Internet grew in part through numbered memos called Requests for Comments. In April 1969, Steve Crocker’s RFC 1, “Host Software,” set out tentative agreements and open questions. Very little was firm. Reactions were expected.

It is a good way to begin something large.

The name survived after the network stopped being small enough for everyone involved to know one another. Somewhere between RFC 1 and the present day, the network scaled its answers a billionfold and quietly stopped asking the question underneath them all.

This document requests comment on that question:

What is the Internet for?

2. Terminology

RFC 2119 uses MUST, SHOULD and MAY to distinguish levels of technical requirement. This fictional memo borrows the capitals to ask what we owe one another. That part is ours, not the standard’s.

packet
A finite sequence of bits sent from one place to another.
router
A machine that decides where a packet goes next.
the middle
Routers, cables, switches, protocols. Carries everything; understands nothing.
an end
Where a connection terminates and comprehension begins. Usually a person. Occasionally two, meeting.
person
Formerly "user." Not a resource. Not inventory. See section 4.
meaning
Not specified.

3. The Record So Far

A few things people found to do with a connection:

From research networks to everyday company
1969The first ARPANET message from UCLA to SRI was meant to say LOGIN. The system crashed after LO. Two letters were enough to begin a very long conversation.
1971Ray Tomlinson sent email between computers sitting beside one another. Distance could come later.
1980sUsenet carried conversations about science fiction, cooking, jokes and the people having them.
1993A camera watching a Cambridge coffee pot became available on the Web. It had already been saving colleagues a wasted trip since 1991.
TodayA new baby. A repair explained. Someone saying they got home safely.

4. The Ends Decide

In their end-to-end argument, first presented in 1981, Jerome Saltzer, David Reed and David Clark argued that some functions need the knowledge available at a system’s endpoints to work completely and correctly. Checks inside the network may still help, but they do not always replace the checks at the ends.

The network was built that way. TCP has moved love letters, lab results, revolutions and spam with perfect equanimity, because the middle was engineered not to know what it carries.

What each layer of that middle can know:

Physical
Did a signal cross?
Ethernet / Wi-Fi
Which local interface?
IP
Which destination address?
TCP
Did the bytes arrive in order?
TLS
Is this connection protected and its peer authenticated as configured?
HTTP
What was requested?
Unicode
Which characters are represented?
The reader
Why did this matter?

The 1981 argument does not prove anything about meaning. It concerns where functions belong in a network. It does leave an architectural coincidence that is difficult to ignore: the middle transports; the ends interpret.

You are an end.

5. One Successful Transmission

payload:

49 27 6d 20 68 65 72 65 2e
decoded:

"I'm here."

Every checksum passed. Every byte arrived, in order, to the correct recipient, authenticated, decrypted. Every device involved did exactly what it was designed to do.

What does it mean?

sent from the train platform
sent after the operation
sent from outside the house
sent after six months of silence
sent by the person you were waiting for

Same bytes.

5.1. Send Something

A local demonstration. Nothing typed here leaves this page.

5.2. Did It Travel?

Suppose the message is: I miss you.

a finger touches glass
voltage changes
bits
radio
fibre
routers
more voltage
pixels
photons
a retina
neural activity

Every physical event in that chain can be accounted for. At no point can an engineer point and say: there, that is the missing. The person at the other end may cry.

Did the meaning travel? Or did something happen at the other end? If it happened there, was it the same meaning?

6. Security Considerations

A question left unasked does not remain unanswered. It gets answered by default, by whoever benefits from the default. The primary vulnerability introduced by this memo's long absence is that "what is the Internet for?" was quietly resolved, at scale, by parties whose honest answer was "for holding your attention until the next quarter."

A system that cannot determine meaning can nevertheless influence the conditions under which meaning is made. A network need not understand a conversation to decide which conversations are easy to have. This is a considerable amount of power.

Implementers, and every person online is one, SHOULD treat any purpose they did not choose as unverified input.

7. IANA Considerations

No registry for meaning is known to exist.

Creation of one is not recommended.

8. Comments Requested

In the tradition of the founding documents, this memo is unfinished on purpose. The question is reopened; the comment period never closes. Responses need not be filed anywhere.

A response looks like this: the next thing you build, send, teach, or repair through the wire, on purpose, for someone.

Author's Address

You
an endpoint
(present location)

comments to: everyone you love

9. Unresolved

A book does not understand its words.

Air does not understand a voice.

A photograph does not remember the person in it.

And yet something arrives.

Every layer below you can explain how the message arrived. None can explain why it mattered.

Why anything should mean anything at all is outside the scope of this memo.

It does anyway.

Request for Comments.

The real documents behind this fictional one