Reverse Association Problem
TL;DR: Figuring out how associations work. (Spoiler: they are a painful type of graph.)
Note: this is purely my own analysis, I didn't read anything on this topic, so these might be incorrect and obsolete ideas.
After playing Dixit, Chameleon, and other association games, I noticed a quirk that I called the "reverse association problem." But lets start simple.
(I think) associations are chains of thoughts, you go from one to the next as you think. So I wrote an example association chain starting from an object in my room. (You should stop and do it for yourself.) and I wasted lots of time scripting a CLI tool for it
eg. Table -> Knights -> Jedi -> Lasers -> Presentations -> Takahashi Method -> Minimalism -> KISS principle -> Confusion
Association Chain:
+--------------------+ +----------------+
|Starting Association|-->|Next Association|-->...
+--------------------+ +----------------+
As I went from Table
to Confusion
, I took some notes. First off, associating is fast, you can associate faster than you can type, if you write down every detail that comes to mind. Which I didn't do skipping some inbetweens, so here is the full expanded chain:
eg. Table -> Knight of the Round Table -> Knights -> Jedi Knights (from Star Wars) -> Lightsabers -> Lasers -> Laser pointer -> Presentations -> Takahashi Method -> Minimalism -> Simplicity -> KISS principle -> (If you don't follow KISS) -> Confusion
Leading to a second observation: You can insert a (more detailed) association between almost any two links, and you can cut links out of the chain making the connections less obvious.
These two properties describe what happens in board games. Given a starting point (prompt) you create a short chain and then carefully adjust it to be just right, usually the word after two links is the sweetspot. In Table -> Round Table -> Knights -> Jedi
picking Jedi
feels to far but Knights
is just right or a bit easy. (Lets call it "endpoint")
Typical pick is two "links" from the starting point:
+--------------------+ +-----------------+ +---------------------------+
|Starting Association|-->|First Association|-->|(Second) Ending Association|
+-------Table--------+ +---Round Table---+ +----------Knights----------+
The second phase of a board game is reversing the association - finding the starting point from the endpoint - which is hard. In Dixit and Chameleon you are given the endpoint and a list of potential start points with lots of context allowing for some tricks.
Reversing association chains:
+--------------------+ +---+ +------------------+
|Starting Association|<--|.?.|<--|Ending Association|
+-------Table--------+ +---+ +------Knights-----+
To eliminate variables of social games lets simplify further to an operation the typical guesser repeatedly performs: If/How are two specific words/associations connected?
Word Chain Game. My "reverse association problem," in simple terms, is connecting two associations (which one is the start and the end doesn't matter). A solution is quickly generating association chains that are "on topic" and trying to get a connection, but I can think of better methods eg. chaining from both sides and looking for a connection between chains. (similar to how computers solve Rubik's cubes)
Connecting two associations:
+-----------+ +---+ +-----------+
|Association|---|.?.|---|Association|
+---Table---+ +---+ +--Knights--+
At this point you should have realized that associations are a graph. A graph is a bunch of interconnected dots. More on Graph Theory: here, here, or Wikipedia
Casually exploring a graph is easy, you are just looking around trying to find something interesting. Trying to get from one point to another on a graph is called pathfinding and one of the big topics in computer science.
Associations are a graph:
+-----------+ +-----------+
|Association|---|Association|
+-----------+ +-----------+
| /
| /
| /
+-----------+ +-----------+
|Association|---|Association|
+-----------+ +-----------+
Good news, we can slap existing mental models (graph theory) on associations and have an easy time understanding what's going on. Bad news, the models are a bunch of hard math and pain is inevitable.
Consider Wikipedia as an approximation of associations, the links between wikipedia pages should be somewhat representative of associations. We already stole graph theory from computer science, now lets steal knowledge from Wikipedia, specifically Getting to Philosophy, Wikiracing, and other Wikipedia Games about exploring the graph of links between wiki pages. Before long you will realize the absence of backlinks, Wikipedia pages often don't link back to where you came from. The Wikipedia graph is in fact directed (eg. Apple links to Central Asia, but Central Asia doesn't link back to Apple.)
Associations are a directed graph:
+-----------+ +-----------+
|Association|-->|Association|
+-----------+ +-----------+
| /
| /
V L
+-----------+ +-----------+
|Association|<->|Association|
+-----------+ +-----------+
Here I notice some cryptographic potential, whenever something is easier forwards than backwards you can use it as a cryptographic tool. For verification you could take an association chain, write down the start and end points then to prove your identity state the inbetweens. Cryptography and humans famously don't work well together, so I doubt practical use would be as smooth as it sounds, especially for one last reason:
The association graph is also weighted. Means that some connections are stronger than others (eg. Apple is more a fruit than it is an archery target). Worse, human psychology makes the weights vary segnificantly between people and change with time. The weights are different for each person based on his environment, occupation, lifestyle, and most importantly recent thoughts have enormous effects on associations.
Associations are a weighted directed graph:
+-----------+ +-----------+
|Association|-1->|Association|
+-----------+ +-----------+
| /
6 4
| /
V L
+-----------+<-7-+-----------+
|Association| |Association|
+-----------+-3->+-----------+
You will know more about a person after reading their search history for 10 minutes, than talking to them for 2 hours.
Association chains (probably) tell a lot about you because you will lean towards fields and topics you are fimilar and active in. Remeber, it is basically eavesdropping on thoughts. Social and association games (in my impression) push people to learn about each other this way (and discussing why certain associations came up). Though I know some people who demand social games to be played "normally" (like trivia games).
Now with a defined phenomenon it might be easier to find an applications for my "reverse association problem," at least designing and playing association games is easier now.
Why does everything always end up being graphs and vectors?
- very qualified association expert
Friendly reminder that the above post is based on absolutely nothing and should be taken with a handful of salt.
Footnotes
Other on topic stuff:
I use asciiflow for the text diagrams, I think they are a nice substitute for images.
Ruby is great for scripting (but Unix isn't.)