d r e a m s


or, ephemera
This is a stream-of-consciousness kind of blog?

what is a puzzle?


To me, most puzzles have 4 parts: interpretation, information, mechanic, extraction. After extraction, there might be a few more steps but they usually fall under extraction. Puzzles usually have a final answer (a string) even though it may not contain a single character in the original puzzle. Puzzles might also come with a title and flavor text, both could be a hint for an obstacle in any of these 4 steps.


Often this is obvious or trivial. For a sudoku puzzle, it's clear that the puzzle looks like a sudoku grid. For a puzzle involving a set of pictures, maybe it would be helpful to know what the pictures are. If the only information is a string of numbers or set of plaintext words, it is much harder to interpret. The goal of this step is to identify what should be done next.


After interpreting the puzzle, the solver now needs to either gather information related to the puzzle and/or fill in blanks that are provided in the puzzle. This usually amounts to solving some clues or logic puzzles. It might also involve scouring the internet for obscure works of art, physically walking to locations, identifying melodies, or playing out a (physical, text-based adventure, flash) game. This is also an opportunity for the author to write a puzzle about whatever they want, and force the solvers to dwell into the subject. This is the part that clouds your search history with desparate searches like ''list of italian operas'' or ''locations of fake eiffel towers''. For the more generic or easier puzzles, this step tends to just be simple clue-solving, Google searching, or common knowledge.

Sometimes (rarely), this step is short because the next few steps are challenging.


A mechanic is the ''a ha!'' moment of a puzzle - an unexpected twist that you realize as you do the previous steps. Having a mechanic and extraction process separates a puzzlehunt-style puzzle from a generic logic problem, crossword, pub trivia questions or brainteaser. In puzzles that have a heavier information stage, this step might not exist. Otherwise, this is where the author can get creative, since anything is viable as a mechanic. Maybe the mechanic is noticing that all of the names found so far are also names of cities in Europe. Maybe you're supposed to extend each word in the crossword grid by a letter. Perhaps the puzzle is entirely in IPA.

This is where a lot of puzzles get stuck; when it is finally solved, it is often the most memorable part.


With all (or most) of the information, there is a (hopefully) insightful extraction process that will be performed on the final data.

Sometimes it's as simple as reading it out loud (or if the ''information'' is graphical, interpreting it as an English word). Other times, it involves taking the first letter of each piece of information.

However, it could get more complex. A standard (and boring) mechanic is indexing, where each string is accompanied by a number (both gathered through the previous step) and the nth letter is taken from the string. A trickier one is noticing that the final result looks like morse, braille, binary, flag semaphore, IATA, or your favorite code. A less obvious step could depend on the mechanic. If the mechanic ordered some words into a chain, maybe adjacent words differ by one letter. Sometimes, all that remains at this point is a solved crossword grid or a set of letters, which makes the solver stuck and sad. Other times, the final phrase is not the answer, and an additional step or two are clued.

This step is often the most frustrating to get stuck at because the final answer is so close. Because all the information in the puzzle has been used at this point, it is difficult to make more forward progress.



A few months ago, I was taught Sushi No, a misere version of the popular game, Sushi Go. There is also Seven Blunders, which is also the misere version of 7 Wonders. I wondered if there was a way to create a misere versions of other games that I liked and I came up with ideas for misere variants of Hanabi and Avalon. They were never playtested and there were some clear flaws. Recently, I thought maybe Coup can be played in a misere manner. It could be named Co-down, or just up.

The goal of the game is to lose as quickly as possible. On your turn, you can still take any of the actions allowed in the game. You can also call other people's bluffs if you wish, and you could choose to reveal that you were bluffing or ignore the call, depending on what benefits you.

UPDATE (15.9) I never finished writing out the rules that I worked out before and it no longer makes sense to me. Maybe one day I'll get to play a balanced game of up.

in the beginning ▽


This is the first post. I created the website a year ago (and learned some css/html/js) but stopped updating it. Then I lost my host and never migrated the website to this one.

I recently wanted to write some thoughts down somewhere, and a blog made the most sense. I considered and rejected wordpress because I had a domain name and a server to put stuff on. I hope that the freedom and flexibility of personal webpage outweighs the convenience and formatting options of wordpess.

Resetting this up took longer than expected. I had to put files in the right directory, and then I had to decide how much of the old website gets ported over. For now, the puzzles will not be moved over. I also delayed the migration because my files were on my old laptop and I kept putting off on transferring them over. Now, they're in version control and on the cloud so even if I lose this webserver, I'll be able to rehost everything.

This page will be used, as the subheading implies, to record stray thoughts on any topic. While some posts may be academic, I expect most of them to not be. I also hope to keep updating this page frequently.

- 🦕 (this is new, and my browser doesn't support it yet!)