Randomized Zettelkasten Test

Motivation and Method

I read the book How to Take Smart Notes a couple years ago and have tried to keep my notes together in a Zettelkasten ever since. I currently use Obsidian, which contains 2500 individual notes collected over two or three years—some abstract to encourage being hooked up to anything and some as concrete as possible to attain specific insight.

However, despite the enjoyment of keeping my notes in one place and looking at all of them in a big network, very few of my additions have come from connections between notes. I want to see if I’ve hit diminishing returns or if original insights are still available with this method.

To that end I’m generating 25 random numbers between 1 and 2500, collecting the results to make a toy Zettelkasten a hundredth the size of the full one, and seeing how many meaningful ideas can be developed from it. Playing fast and loose with priors here, a graph with 25 vertices can have a maximum of 300 edges, and by Sturgeon’s Revelation I’d expect only 10% of these to be good. So: if I find 30 worthwhile connections or products, I’ll be satisfied.

Now to roll.

The 1%-Zettelkasten

  1. Satellites (onto-cartography) - ontological machines caught up in the “gravity” of “bright objects”. (See this post for more details.)

  2. Ancestors don’t die young—for some definition of “young”; all of our ancestors lived long enough to reproduce.

  3. Biennials—plants that take two years to fully grow. An important distinction if you ever need to plant crops for food.

  4. “Death extinguishes envy”—a quote that stuck with me, though it’s not fully true. Death removes us from the social world and solidifies a much kinder impression of us than the average person might have during our lives.

  5. Diffuse responsibility—the well-discussed state where 100 people seeing a problem will do nothing because they each feel 1/​100th the responsibility.

  6. New senses make new media—speculation. A new medium can be developed by finding some new material or way of arranging things, but if it were possible for ex. to feel microscopic etchings with one’s fingers I would expect a new artform to arise from it.

  7. Phenomenological barriers are symmetric—just as cisgender people can’t fully feel what it would be like to be transgender (because to have the experience with any fidelity would make them no longer cisgender), transgender people can’t completely fathom being cisgender. (Or replace cis and trans with gay and straight or another binary of your choice.)

  8. “Type A” and “Type B” drivers—the folk distinction between drivers who move as fast as possible towards their destination and those who prefer to enjoy the ride.

  9. Campaign progression—tabletop advice: a campaign should proceed in steps, with the players creating situations with greater control or freedom only to deal with increasingly nasty threats.

  10. Three-person fair coin tosses—the process of flipping a coin twice, where one person wins on HH, another HT, and a third TH. On TT, you start over.

  11. Voting antipatterns—conditions where voting systems produce outcomes people don’t want: someone gets elected no one wants (Dark Horse), one candidate soaks up votes for the majority choice (vote-splitting), the winner by some metrics get squeezed out on either side (center squeeze), two people team up to eliminate a third but whoever helps more will lose when they go head-to-head (Burr dilemma), or pairwise winners proceeding in a cycle with no majority victor (Condorcet cycle).

  12. Wonder rooms became museums—the progression from eccentric intellectuals collecting interesting things to those things becoming a public educational resource.

  13. Do the obvious—see this post.

  14. Jungian extension—idle thought after briefly reading about Jung’s concept of the anima as the unconscious feminine part of male psychology (or the animus as the masculine part of female psychology): shouldn’t the anima possess an animus of its own, the inner feminine having an inner-inner masculine with an inner-inner-inner feminine, etc.?

  15. Mereological fungibility—the state of a thing being interchangeable with others of its class, type, kind etc regardless of ordering; the condition of paying ten one-dollar bills where any one could be the last without effecting the outcome.

  16. Maze cryptography—idea for a creating simple Morse-like code based on cardinal directions, then turning a message into the correct solution to a maze.

  17. Medical anatomy improves, artistic anatomy doesn’t—while there is always more art to be made, our understanding of all the sketchable parts of a person is complete while our understanding of the internal workings of a person and how they work is still expanding.

  18. Catachronism—recently coined, the redefinition of the past and present in terms of a potential future.

  19. Unmotivated attackers—the security fairy-tale of attackers who simply give up after their first few tries, or at all.

  20. Aleatoric novels—early choose-your-own-adventure style novels where readers proceeded in the text by flipping coins at certain points. Or so I’d thought—aleatoricism does exist as a method of using randomness in creative work, but I can’t find any mention of coin-flipping in novels and I’m not sure where I got this impression.

  21. Band names should be used for other collaborations—idle thought.

  22. Current-event entertainments—kinds of entertainment that can be checked daily or semi-daily and that involve moment-to-moment relevance, like streamers or political commentators. Arguably includes soap operas.

  23. Derangement (permutations) - when a set of items undergoes a permutation and every item winds up in a different place than it started.

  24. Extend a stone in atari—beginner Go advice, essentially not to let a single stone be surrounded.

  25. The Multiple-Stage Fallacy—outlined in this post, the phenomenon of driving the probability of anything to near-zero by breaking it into “stages”, assigning probabilities to each “stage” and multiplying them, without allowing updates on likelihood once earlier “stages” are passed.


  • 1-4. Death is a bright object—many things in life get inescapably defined in relation to death, in the same way that they’re defined by being on Earth or by the nature of time.

  • 1-5. Responsibility (onto-cartography) - too many agents in one system, or at a certain moment in the system’s history, can lead to very little action.

  • 1-18. Catachronism as temporal gravity—whatever future we expect defines our actions in the present, and likely future states of a system can affect its behavior in the here and now.

  • 2-3. Biennials are successful—it seems to work as a growth strategy.

  • 2-4. Our ancestors were envious—they had more time to play these social games and have all the available negative reactions to each other.

  • 2-11. Evolution as avoidance of antipatterns—granting of course that evolution also involves running into many many antipatterns and then dying.

  • 4-13. Death as escape from social world—taking death here more metaphorically, the motivation behind someone deleting an account and making a new one under a different name, or a famous author publishing under a pseudonym.

  • 4-18. Death defines life up to death—treading much the same ground as before, whatever the destination of life is will inevitably define the journey of it.

  • 5-13. They’re waiting on you—the obvious trained reaction to noticing you’re a bystander in some situation, taking the burden of action onto yourself because someone has to.

  • 5-15. Psychological fungibility of strangers—perhaps the problem of diffuse responsibility is not that we ourselves feel less responsible, but that we see everyone else in the crowd as interchangeable and assume that like us they would do something in this situation if it were presented to them alone.

  • 5-19. Extramotivation in foes—just as some people can take it upon themselves to help when everyone else is paralyzed, expect that some people will use this special level of agency to work toward goals against your own.

  • 6-7. Fewer senses make new media—speculative again, but it’s possible to communicate things through a lack of understandable media, whether by painting messages only the colorblind can read or by including sections that are unreadable as an artistic point.

  • 8-9. Type A and type B players—really a gross generalization of the player typology seen in most game-mastering guides, but could eventually converge on this genre of advice given enough nuance.

  • 9-11. Group campaign failures—collaborative storytelling can go in directions no one wants it to

  • 9-18. Campaign attractors—likely end states of a campaign, usually either of it being left unfinished or of a certain ending being achieved.

  • 9-20. RPGs as aleatoric—the injection of randomness into these games being one of the key things that separate them from other kinds of collaborative art.

  • 9-22. Play-by-post as current-event entertainment—possibly lacking relevance to the outside world, but changing daily and encouraging regular checkups (unless it’s scheduled differently).

  • 10-16. Coin flip cryptography—it’s possible to encode coin flips into Morse, so if someone could communicate a large series of them discreetly then a message could be sent.

  • 11-18. Steering toward antipatterns—in a system you want to crash, you should look for common failure modes (there will likely be several flavors) and move towards them.

  • 12-13. Turn wonder rooms into museums—when you have a collection of eccentric odds and ends that it’s fun to show people, make it into a resource that can enrich a wider audience.

  • 12-17. Medicine as extension of art - (Edit: I forgot to include the details on this one) the connection between the human form and curing illnesses was not always historically obvious, and had to be developed as those who studied the former informed those who studied the latter.

  • 12-21. Galleries as art—galleries are mostly named after places, people or subjects, but there seems to be plenty of room for them to take on some more stylistic elements.

  • 13-17. Consider nonimproving corpuses—distinguish between bodies of knowledge that are still being completed and bodies of knowledge that have moved on to being more about themselves—the difference between group theory and James Joyce studies.

  • 13-18. Aim for a contextualizing future—look for a destination that redefines the journey as good.

  • 13-24. Look for beginner advice—self-explanatory.

  • 15-23. Fungibility is derangibility—kind of a cheat; when you can do things in any order you can do them in any order.

  • 15-25. Negative or Zenoic mereology—arguments that involve splitting things into infinitesimal parts to prey on our intuitions about parts that are only very small.

  • 19-24. Unmotivated opponents—the opponents imagined in “wouldn’t it be nice” play, where you make moves in the hope that your strategy is not discovered instead of making moves that are best whether they’re understood or not.

  • 23-25. Multiple Stage-ing when derangible—breaking things into stages and multiplying their probabilities seems less of a sin if those stages could happen in any order /​ the probabilities are independent of each other. (Typically if you’re trying to break something into smaller pieces you’ll find that they’re dependent on each other, though.)


I have 29 results, which is suspiciously in line with such an off-the-cuff prediction, so there’s probably some motivated stopping here—take it as my having reached the finish line, not testing to see how much can be produced in total.

The prediction I made doesn’t quite hold anyway, because a Zettelkasten in the wild would also include connections between these new results and the old ones. I didn’t consider that at the start, but in hindsight I’d expect there to be further, diminishing growth from here.

The usefulness of these combinations might be questionable, as are the original 25, but I don’t curate my notes very heavily and take them on many subjects. It might also be leveled that my connections are not fully connective, or that they introduce too much outside information, but I remain skeptical that trying to undergird extremely disparate fields will lead to anything other than statements of the obvious or category theory.

All in all, I’m surprised to see this working. At first I found almost no combinations at all, and it was only through specific effort that I produced most of these. It took probably two hours distributed across a couple of days to generate all 29, which implies that the kind of work I’m doing here takes more energy generally than reading off a list and being struck by inspiration. Mulling over connections this way isn’t the best use of my time, typically, but I’m glad to see that it works roughly as advertised.