When thinking about how to represent and model relationships, it is important to determine what properties define or distinguish between possible kinds of relationships. It turns out that some of these same distinguishing properties are the units of design that can be manipulated to create Designer Relationships.
There are a very large number of ways to characterize relationships: it is similar in magnitude to the very large number of kinds of relationships you or I know about or can imagine, and the number of ways they might differ from each other. Here I will look at one possible distinguishing property: "The number of times I have participated in the relationship." Next time, I'll look at how this and similar properties can be used in designing relationships.
When I interact with someone at least once, I say that I have a relationship with that person. The relationship can be very infrequent: my relationship with the police officer who pulls me over for speeding (hopefully just once!). Or somewhat more frequent: my relationship with the cousin I see now and again at family events. Or even more frequent: "I see my sister, I don't know, like 10 times a day."
One property which seems to help distinguish the relationships above is the number of times I have interacted with the other person in the relationship. Let's refer to this property as "Interaction Count."
Interaction Count is a property not only of my relationships to people (more on the special case of People below), but a property of my relationship to anything I interact with, such as a movie. Modeling my relationship to a movie is a good thought experiment, because it is simpler than my relationships to people.
How often did I see movie X? Just once: my relationship with movie X is in part characterized by an Interaction Count of 1. (I say "in part" because there are many dynamically interacting properties that characterize a relationship as a whole)
Now, let's say that I have just seen movie X again: my relationship with movie X is in part characterized by an Interaction Count of 2.
Simple enough, and very useful. Yet as soon as I start thinking about assigning Interaction Count values to my relationship to a movie in real life, some practical concerns come to the fore.
Quantitative-Qualitative:
Quantitatively assigning the count is simple. Yet for values over 1 or 2, the meaning quickly becomes something more easily summarized in natural language:
“No, I’ve never seen it [Interaction Count = 0],” “Oh yeah, I saw it last week for the first time [Interaction Count = 1]”, “I’ve seen it twice [Interaction Count = 2].” But more than twice? “I work at the movie theater and I’ve seen it so many times that I dream about it [Interaction Count = ?].” Lets replace the question mark with something a little more useful: Interaction Count = N, for all N greater than 1.
And so, it becomes quickly apparent that the Interaction Count has both quantitative and qualitative meanings. This raises a practical concern as to how to assign and interpret the value of this property. The actual count is most important when the value is low: the difference between 0, 1 and 2 (never, once, twice) is very meaningful and easy to interpret. But for counts greater than 2, it is usually only important to know that I have seen it more than once or twice, and not so important to know exactly how many times. "I love that movie! I saw it plenty of times, but whether it was 6 or 7 times, I don't know, and I don't care." Interaction Count = N.
For most of the more interesting relationships in my life, [Interaction Count = N] is the most frequent case.
Part-whole:
Another practical concern is part-whole distinctions, or "how much" or "what part". "The second time I watched the movie, I only watched the last 10 minutes." Yet, the count would still be incremented from 1 to 2. In that case the Interaction Count can be said to be blind to "how much" or "what part" of the movie I watched. "My kid watches that movie over and over again, so I'm exposed to random snippets of it, over and over again!" The part-whole concern turns out to be crucial when modeling relationships of any complexity.
Yet another practical concern might have to do with qualities of a specific interaction, such as attention or state of mind. "I watched that movie twice, the first time was with my old girl friend, but the second time we had just broken up and I was mostly just thinking about her the whole time." Here again, the count would still be incremented by from 1 to 2. The Interaction Count can be said to be blind to qualities of attention and state of mind.
And many, many more practical concerns become apparent as I work on the model. The concerns can be different depending on the possible kinds of relationships I might have to: my relationship to a book ("I've read that twice"); my relationship to a place ("I've only been there once"); my relationship to an emotion ("I've never felt THAT way about them"); or my relationship to anything in the world. And of course, the kinds of relationships I might have with persons.
Persons and Interaction Counts
"I have never met her [Interaction Count = 0]."
"We've met once [Interaction Count = 1]".
"I talked to her just now and once before [Interaction Count = 2]."
"Oh gosh, I talk to her all the time, she's my wife! [Interaction Count = N]!
And with persons the practical concerns of the Interaction Count become even more obvious than with movies, books and other objects.
"Yeah, I met her once but for only 2 minutes So…."
"I met her at the Holiday party, but I was drunk. So..."
"Her? Yeah, I met her once, but she was with that other girl who I was trying to impress. It's embarrassing, but I hardly remember her at all."
You get the idea. People are complex, and the possible kinds, types, and details in relationships between them is, in a sense, that complexity times the number of persons relating.
[next time:
What number of properties would it take to describe a person-to-person relationship of any complexity? What is the connection between that number and the number of properties needed to represent a Person? In other words: How many ways can persons differ from each other? Let's say that I come up with a specific number of ways, let's say 100 ways, what does this tell us of the number of distinct kinds of persons I can generate, as a game, just from the model? And, importantly, should a model of persons include models of their relationships?
I have noted above that when a property fails to distinguish between some distinguishing characteristic of the thing being modeled (qualities of attention and state of mind as in the example of movie watching), that the property, and therefore, the model it is part of, are blind to that characteristic. In what ways might this blindness be adaptive, a productive design tool (this touches on the usefulness of incompleteness and indeterminacy)?
]
concept fields: [designer relationships, relationship model, All Models Are Wrong, person (data structure), no representation of a person is the person, part-whole relationships, achievement of representation systems, combinatorics, large numbers of small things, interaction count]
If you would like to reference this page, here is an APA Style citation:
Smith, T. (2026). Modeling Relationships: Properties for Design. Small-Infinities. https://small-infinities.com/surfaces/modeling-relationships-properties-for-design