With the next RIPE Meeting coming up, some of us might be dreading the networking we take ourselves to know less about: social networking. But its functions show inherent parallels to the base functions of computer networking. So accompany me to a deep dive on how the OSI layers relate to human communication theory and find out why small talk is not in the least at superficial as it may seem...
I myself have hated small talk for most of my life, because I viewed it as tedious exchange of banalities without content or function. As I deepened my understanding of psychology, communication theory, and epistemology, I began to realise its functions and the inherent parallels to the base functions of computer networking.
Emma - A layered use case
Let us accompany Emma - a network engineer from Germany - on her visit to the RIPE Meeting in Rotterdam:
After a long and tedious first plenary session full of new information Emma tiredly grabs coffee and snacks. All tables are already occupied, so she steers towards one where only one person is already standing, hoping to keep the communication to a minimum.
„Ist hier noch frei?”, she asks in German.
Jim, a network engineer from the Netherlands speaks basic German, but isn’t quite confident with it, so he just nods and gestures for her to settle in.
"Jim”, he says and extends his hand.
"Emma”, she replies and shakes it. "Nice to meet you.”
Let’s look at this exchange from a network perspective. A lot of things have taken place that mostly go unnoticed. I don’t know if a computer would notice or be able to tell you if we could ask it how a TCP handshake works, but for us humans, it’s mostly a subconscious experience.
Starting with the OSI physical layer, we’ve established a connection on two wireless mediums and one wired connection:
- The visual medium: transporting gestures over the air (you could as well call it optical transport) like pointing, nodding, and facial expressions. Although seemingly banal, keep in mind that this would be impossible for a blind person, or in a phone call.
- The acoustic medium: transporting voice. This could be easily jammed by the loud atmosphere of a conference centre or similar.
- The haptic "wired” medium: transporting the handshake and its related information - strength, hand temperature, roughness of skin, and similar.
So we have established a physical connection between the two network nodes, although calling it "physical” may be strange from a human point of view. Now let’s go up one OSI layer.
This layer is probably the most unintuitive one to compare, because things like flow control in human communication is mostly regulated via cultural standards and an ominous thing psychologists call "Theory of Mind”. Logical Link Control determines whose turn it is to speak, and when - a complication that is quite easy in the exchange above, but something I believe we all have experienced various problems with in large company meetings.
The question of Medium Access Control in our example is answered on a deeply unconscious level, because it determines things like whether it is socially acceptable for Emma to join Jim’s table - a question that is subconsciously answered on her side during the decision making process for choosing the table. Knowing that the question might be answered differently on Jim’s side, she is inclined to ask if it’s okay to join him. But we all know people whose medium access control works quite differently from ours.
Another layer that is more on the hazy side of Theory of Mind is the network layer. Most functions are only perceivable in larger groups and social constructs and have little relevance in point to point communications like the exchange above. One could probably write books about the finesse of multicast group management and routing in the development of cliques and the spreading of gossip. While IPv4 had quite the rigid structure of an elementary school classroom, IPv6 and especially the neighbour discovery protocol conveys all the social and communicative beauty of the average high school cafeteria.
With the transport layer, we are moving close to the core of the actual exchange taking place above, but also towards a direction where the human communication layers are more interlaced than the OSI layers.
While we are used to people speaking UDP in company meetings where everyone just talks and nobody seems to listen, Emma and Jim are busy performing a TCP handshake. Coming from the example above, the usage of the term handshake for TCP seems to stem quite obviously from human behaviour:
Emma sends a SYN package. We’ll tackle the issue of language later, but as there is enough information available, Jim can identify it as such. He then responds with a SYN ACK, where we can imagine the SYN part as the invitation to join the table and the ACK part as him actively reaching out and introducing himself. Emma counters the introduction with one of herself, her own ACK. The three way handshake is complete. They are now ready to establish a session.
We could now go on assuming that the interchange only went this far regarding the communication layers and continue the exchange, where Jim asks Emma whether this is her first RIPE Meeting and they talk about the weather. But let’s yet hold this off for a moment, because then we’d disregard too much that has also happened above.
Moving on to the session layer, we will not clearly be able to see it in most point to point exchanges, since the finesse of session handling only really shows in larger groups.
Think of a table full of people at the conference dinner: Some conversations will be held between all people at the table while some time will be spent on sub-discussions in smaller groups. Maybe someone in the middle of the table will even be able to listen and join in on both discussions at the same time - a feat I have always found admirable since my session layer implementation will simply not support doing that (I’ll just drop both connections by accident). Between two people the initiation or termination of a session is usually equivalent to the initiation or termination of the complete connection, so while present in principle does not play any relevant role.
On the other hand, the presentation layer’s main functionality is to provide a common format for the data exchange taking place. Words like translation and syntax are by no accident part of nearly every description of its functionality, be it in a college class for computer science or an online encyclopedia.
This is where our two protagonists agree on language. Emma starts out with her native language, but quickly realises that a solicitation is necessary. From Jim’s reaction of seemingly understanding her, but refusing to verbally answer, and maybe his facial expression - paired with the contextual knowledge of being at an international conference - she can refer that he is not (sufficiently) able to, or is uncomfortable to, speak German. Knowing that English is the common ground because the sessions are held in it, she defaults back to what she knows must be a common ground.
Note how we have not yet talked about data or content. I’m sure you agree with me that there hasn’t really been any, besides the exchange of names (are you already thinking of them as more or less unique identifiers like MAC or IP addresses?). Yet, there has been a lot of meaningful exchange which we often disregard, because the superficial topic is only the weather. Keep that in mind, next time you’re watching people doing small talk.
Last but not least, the application layer is all about how we communicate. Although the word 'presentation' is used for the previous layer, it is the application layer that is all about how we present information. When we’re just providing information, it comes close to rhetorics. Think of our verbal and body language as a website: This is HTML - how we present what we say, how we structure our talk, the semantics we choose to convey meaning. Let’s add some CSS in the mix on how we dress, style our hair or design our slides and we have the complete base for data exchange with other human networking nodes. Just think of queen Elizabeth II and how she signalled her opinion with clothing, colour and jewelery choices when she wasn’t officially permitted to say it out aloud.
Now, Jim and Emma have established a connection. They can now start to discuss the various topics of the RIPE meeting, their opinions and ideas and contribute to the future of networking - in every sense of the word.
If you followed me this far, I’d like to thank you for your patience and open-mindedness and add a little bonus to the mix. Many of us have often jokingly talked about "Layer 8” - mostly when referring to error sources originating between keyboard and chair. But there is an actual point here, a full-fledged layer of communications, complete with headers and glitter, which we refer to when invoking layer 8:
Psychologist Paul Watzlawick once introduced the idea that messages contain content as well as relational information. Anyone smelling header and data here? Me too. Together with similar revelations by psychologist and linguist Karl Ludwig Bühler, psychologist Friedemann Schulz von Thun introduced the currently very common four-sides or four-ears model. It extends the encapsulation of data to an additional layer.
Now many "nerds” assume, that the point of communication is the exchange of factual data. I’m not diving into the more than complicated debate on whether this is true or even what factual means, but we are at a RIPE meeting so let’s just assume that we want to exchange actual information on networking and tech.
While in the four-sides model, all sides are presented equal for our networking perspective I would like to shift that view and declare the factual information as data or payload of the interchange. The other three sides are self-revelation, relationship and appeal. While it may be arguable that they are always present and thus create a fixed header for the payload, I personally prefer to view them as extension headers, since I see more parallels. My personal understanding is that they MAY be omitted, they SHOULD be evaluated by the receiver (but we all know how RFC2119 SHOULDs are treated in real life, don’t we?) , and when interpreted wrongly they can poison and break the whole interchange.
Most "Layer 8 errors” I encounter in my daily business are actually header problems and seldom found within the factual data payload.
So, in case you hate small talk and human networking like I used to, I hope my remarks helped you realise its job and value. While the superficial flow of information about the weather, the food or whether this is Emma’s first RIPE Meeting and how she likes Rotterdam may be nice - but it is completely irrelevant.
The role of small talk is to establish a connection, about solicitation, flow regulation and protocol negotiation. The more people there are in an exchange, the more small talk is necessary to test the waters with all members of the local network and establish a complex understanding. Two people will always be faster to move on to content - that is, if they have any real interest in it - while you might be going to bed after dinner with twelve people at a table and thinking that there was no actual talk happening. In that case: congratulations! You spent the entire evening exploring the endless depths of the neighbour discovery protocol.
 Andrew S. Tanenbaum, Computer Networks: International Edition, 4th Edition, (Prentice Hall International, 2003) ISBN 978-013038488
 X.225 : Information technology – Open Systems Interconnection – Connection-oriented Session protocol: Protocol specification
 ISO/IEC 7498-1:1994 Information technology — Open Systems Interconnection — Basic Reference Model: The Basic Model
 Anne Böckler-Raettig, Theory of Mind, 1. Edition, (UTB, 2019) ISBN 978-3825251338
 Friedemann Schulz von Thun, Kathrin Zach und Karen Zoller, Miteinander reden von A bis Z: Lexikon der Kommunikationspsychologie, 4. Edition (Rowohlt, 2012) ISBN 978-3499628306