Independent Submission                                            R. Hay
Request for Comments: 5841                                     W. Turkal
Category: Informational                                      Google Inc.
ISSN: 2070-1721                                             1 April 2010


                    TCP Option to Denote Packet Mood

Abstract

   This document proposes a new TCP option to denote packet mood.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This is a contribution to the RFC Series, independently of any other
   RFC stream.  The RFC Editor has chosen to publish this document at
   its discretion and makes no statement about its value for
   implementation or deployment.  Documents approved for publication by
   the RFC Editor are not a candidate for any level of Internet
   Standard; see Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc5841.

Copyright Notice

   Copyright (c) 2010 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.

1.  Introduction

   In an attempt to anthropomorphize the bit streams on countless
   physical layer networks throughout the world, we propose a TCP option
   to express packet mood [DSM-IV].

   Packets cannot feel.  They are created for the purpose of moving data
   from one system to another.  However, it is clear that in specific
   situations some measure of emotion can be inferred or added.  For
   instance, a packet that is retransmitted to resend data for a packet
   for which no ACK was received could be described as an 'angry'
   packet, or a 'frustrated' packet (if it is not the first
   retransmission for instance).  So how can these kinds of feelings be
   conveyed in the packets themselves.  This can be addressed by adding
   TCP Options [RFC793] to the TCP header, using ASCII characters that
   encode commonly used "emoticons" to convey packet mood.

1.1.  Terminology

   The keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD,
   SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL, when they appear in this
   document, are to be interpreted as described in [RFC2119].

2.  Syntax

   A TCP Option has a 1-byte kind field, followed by a 1-byte length
   field [RFC793].  It is proposed that option 25 (released 2000-12-18)
   be used to define packet mood.  This option would have a length value
   of 4 or 5 bytes.  All the simple emotions described as expressible
   via this mechanism can be displayed with two or three 7-bit, ASCII-
   encoded characters.  Multiple mood options may appear in a TCP
   header, so as to express more complex moods than those defined here
   (for instance if a packet were happy and surprised).

              TCP Header Format

         Kind     Length     Meaning
         ----     --------   -------
          25      Variable   Packet Mood

   In more detail:

           +--------+--------+--------+--------+
           |00011001|00000100|00111010|00101001|
           +--------+--------+--------+--------+
            Kind=25  Length=4 ASCII :  ASCII )

           +--------+--------+--------+--------+--------+
           |00011001|00000101|00111110|00111010|01000000|
           +--------+--------+--------+--------+--------+
            Kind=25  Length=5 ASCII >  ACSII :  ASCII @

3.  Simple Emotional Representation

   It is proposed that common emoticons be used to denote packet mood.
   Packets do not "feel" per se.  The emotions they could be tagged with
   are a reflection of the user mood expressed through packets.

   So the humanity expressed in a packet would be entirely sourced from
   humans.

   To this end, it is proposed that simple emotions be used convey mood
   as follows.

      ASCII                Mood
      =====                ====
      :)                   Happy
      :(                   Sad
      :D                   Amused
      %(                   Confused
      :o                   Bored
      :O                   Surprised
      :P                   Silly
      :@                   Frustrated
      >:@                  Angry
      :|                   Apathetic
      ;)                   Sneaky
      >:)                  Evil

   Proposed ASCII character encoding

      Binary          Dec  Hex     Character
      ========        ===  ===     =========
      010 0101        37   25      %
      010 1000        40   28      (
      010 1001        41   29      )
      011 1010        58   3A      :
      011 1011        59   3B      ;
      011 1110        62   3E      >
      100 0000        64   40      @
      100 0100        68   44      D
      100 1111        79   4F      O
      101 0000        80   50      P
      110 1111        111  6F      o
      111 1100        124  7C      |

   For the purposes of this RFC, 7-bit ASCII encoding is sufficient for
   representing emoticons.  The ASCII characters will be sent in 8-bit
   bytes with the leading bit always set to 0.

4.  Use Cases

   There are two ways to denote packet mood.  One is to infer the mood
   based on an event in the TCP session.  The other is to derive mood
   from a higher-order action at a higher layer (subject matter of
   payload for instance).

   For packets where the 'mood' is inferred from activity within the TCP
   session, the 'mood' MUST be set by the host that is watching for the
   trigger event.  If a client sends a frame and receives no ACK, then
   the retransmitted frame MAY contain the TCP OPTION header with a mood
   set.

   Any packet that exhibits behavior that allows for mood to be inferred
   SHOULD add the TCP OPTION to the packets with the implied mood.

   Applications can take advantage of the defined moods by expressing
   them in the packets.  This can be done in the SYN packet sent from
   the client.  All packets in the session can be then tagged with the
   mood set in the SYN packet, but this would have a per-packet
   performance cost (see Section 5, "Performance Considerations").

   Each application MUST define the preconditions for marking packets as
   happy, sad, bored, confused, angry, apathetic, and so on.  This is a
   framework for defining how such moods can be expressed, but it is up
   to the developers to determine when to apply these encoded labels.

4.1.  Happy Packets

   Healthy packets are happy packets you could say.  If the ACK packets
   return within <10 ms end-to-end from a sender's stack to a receiver's
   stack and back again, this would reflect high-speed bidirectional
   capability, and if no retransmits are required and all ACKs are
   received, all subsequent packets in that session SHOULD be marked as
   'happy'.

   No loss, low-latency packets also makes for happy users.  So the
   packet would be reflecting the end-user experience.

4.2.  Sad Packets

   If retransmission rates achieve greater than 20% of all packets sent
   in a session, it is fair to say the session can be in mourning for
   all of the good packets lost in the senseless wasteland of the wild
   Internet.

   This should not be confused with retransmitted packets marked as
   'angry' since this tag would apply to all frames in the session
   numbed by the staggering loss of packet life.

4.3.  Amused Packets

   Any packet that is carrying a text joke SHOULD be marked as 'amused'.

   Example:

      1: Knock Knock
      2: Who's there?
      1: Impatient chicken
      2: Impatient chi...
      1: BAWK!!!!

   If such a joke is in the packet payload then, honestly, how can you
   not be amused by one of the only knock-knock jokes that survives the
   3rd grade?

4.4.  Confused Packets

   When is a packet confused?  There are network elements that perform
   per-packet load balancing, and if there are asymmetries in the
   latencies between end-to-end paths, out-of-order packet delivery can
   occur.

   When a receiver host gets out-of-order packets, it SHOULD mark TCP
   ACK packets sent back to the sender as confused.

   The same can be said for packets that are sent to incorrect VLAN
   segments or are misdirected.  The receivers might be aware that the
   packet is confused, but there is no way to know at ingress if that
   will be the fate of the frame.

   That being said, application developers SHOULD mark packets as
   confused if the payload contains complex philosophical questions that
   make one ponder the meaning of life and one's place in the universe.

4.5.  Bored Packets

   Packets carrying accounting data with debits, credits, and so on MUST
   be marked as 'bored'.

   It could be said that many people consider RFCs boring.  Packets
   containing RFC text MAY be marked as 'bored'.

   Packets with phone book listings MUST be marked 'bored'.

   Packets containing legal disclaimers and anything in Latin SHOULD be
   marked 'bored'.

4.6.  Surprised Packets

   Who doesn't love when the out-of-order packets in your session
   surprise you while waiting in a congested queue for 20 ms?

   Packets do not have birthdays, so packets can be marked as surprised
   when they encounter unexpected error conditions.

   So when ICMP destination unreachable messages are received (perhaps
   due to a routing loop or congestion discards), all subsequent packets
   in that session SHOULD be marked as surprised.

4.7.  Silly Packets

   Not all packets are sent as part of a session.  Random keepalives
   during a TCP session MAY be set up as a repartee between systems
   connected as client and server.  Such random and even playful
   interchanges SHOULD be marked as silly.

4.8.  Frustrated Packets

   Packets that are retransmitted more than once SHOULD be marked as
   frustrated.

4.9.  Angry Packets

   Packets that are retransmitted SHOULD be marked as angry.

4.10.  Apathetic Packets

   When sending a RST packet to a connected system, the packet should be
   marked as apathetic so that the receiver knows that your system does
   not care what happens after that.

4.11.  Sneaky Packets

   When a packet is used in a particularly clever way, it SHOULD be
   marked as sneaky.  What is "clever" is rather subjective, so it would
   be prudent to get a few opinions about a particular use to make sure
   that it is clever.

4.12.  Evil Packets

   It is hard for a TCP packet to discern higher moral quandaries like
   the meaning of life or what exactly defines 'evil' and from whose
   perspective such a characterization is being made.  However,
   developers of TCP-based applications MAY choose to see some
   activities as evil when viewed through their particular lens of the
   world.  At that point, they SHOULD mark packets as evil.

   Some organizations are prohibited from using this mood by mission
   statement.  This would also prohibit using the security flag in the
   IP header described in [RFC3514] for the same reasons.

5.  Performance Considerations

   Adding extensions to the TCP header has a cost.  Using TCP extensions
   with the ASCII-encoded mood of the packet would detract from the
   available MSS usable for data payload.  If the TCP header is more
   than 20 bytes, then the extra bytes would be unavailable for use in
   the payload of the frame.

   This added per-packet overhead should be considered when using packet
   mood extensions.

6.  Security Considerations

   The TCP checksum, as a 16-bit value, could be mistaken if ASCII
   characters with the same number of zeros and ones were substituted
   out.  A happy ":)" could be replaced with a frown by a malicious
   attacker, by using a winking eye ";(".  This could misrepresent the
   intended mood of the sender to the receiver.

7.  Related Work

   This document does not seek to build a sentient network stack.
   However, this framework could be used to express the emotions of a
   sentient stack.  If that were to happen, a new technical job class of
   network psychologists could be created.  Who doesn't like new jobs?
   :)

8.  IANA Considerations

   If this work is standardized, IANA is requested to officially assign
   value 25 as described in Section 3.  Additional moods and emoticon
   representations would require IESG approval or standards action
   [RFC5226].

9.  Informative References

   [DSM-IV]  "Diagnostic and Statistical Manual of Mental Disorders
             (DSM)", http://www.psychiatryonline.com/
             resourceTOC.aspx?resourceID=1.

   [RFC793]  Postel, J., "Transmission Control Protocol", STD 7, RFC
             793, September 1981.

   [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
             Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an
             IANA Considerations Section in RFCs", BCP 26, RFC 5226, May
             2008.

   [RFC3514] Bellovin, S., "The Security Flag in the IPv4 Header", RFC
             3514, April 1 2003.

Authors' Addresses

   Richard Hay
   Google
   1600 Amphitheatre Pkwy
   Mountain View, CA 94043
   EMail: rhay@google.com


   Warren Turkal
   Google
   1600 Amphitheatre Pkwy
   Mountain View, CA 94043
   EMail: turkal@google.com