Being creative with cr

Last updated: Nov 2018

I don't believe I was showing an extraordinary creativity when I thought, early October 2018, that it would be good to have a tool that would generate random associations between my own concepts. I was surprised, however, to fail at finding such a tool. So I coded a very simple program, called cr, which does exactly that: it takes a number of .txt files as input, each file containing a series of concepts, and generates random associations.

Here are, for instance, some possible outputs:
Example 1 Example 2 Example 3 Example 4
attention
bayes
area
volitional philanthropy
evolution by replacement
false flag
by no means
r/k selection theory

(Those are the most publishable out of 6 executions, with my current concepts).

This page is an account of my experience with this method, particularly of the highest quality associations that have been generated.

First, what is a concept? The answer is simple once the purpose of this method is understood: anything that may stimulate your creativity as you wish is eligible.

To make things clearer, here is the beginning of my list in c.txt:

a posteriori
a priori
ability
abolitionism
about
absence
absurd
academia
academic research
acceleration
active recall
ad
add
admit
adrenaline
adult education
advocacy
aesthetics
affordable
agi
agility
agreableness
ai
ai risk
airbnb
airbus
airport
alcohol
alien
alien god
alone
alright
alternative
altitude
amazon
america
american dream
amf
amnesty international
amusement park
android
anger
angle
ani
animal welfare
anki
anniversary
anonymous
anonymous negative feedback

Current progress

I currently have two files: p.txt (44 people, some famous, some I know personally) and c.txt (everything else, 1194 concepts so far). I'm filling these files at a steady rate.

How I had the idea

I have been interested in methods of learning from a relatively young age, which led me to Anki. At some point I read (from Robb Seaton) about using Anki for idea generation: by putting all cards in the same deck, the reviews should lead to collisions between unrelated topics and hopefully, new ideas.

However, the efficiency of the learning process with spaced repetition makes collisions quite rare. I've become frustrated with such a low density, which led me to this idea.

Now that I think about it, cr has other advantages over Anki:

  • The user can be focused on creativity, instead of multi-tasking with trying to recall the answers;
  • The overwhelming majority of the concepts in my files could never have landed in my Anki collection in the first place (as you can see from my letter C).
  • Remarkable associations

    Nothing really stands out so far.

    Thoughts on this method

  • It makes sense to believe that the method will be increasingly powerful as more concepts are added.
  • I still can't believe I haven't found any program doing this on the Internet. I think some implementations must exist but are too obscure to be easily reached (after all I haven't spent such a long time searching); also, perhaps some people have written a code similar to mine for themselves, but simply haven't published it.