Kai's blog

Orah: The Insidious Temptation of False Robustness in Software-Based Solutions

Tue, 05 May 2026 08:00:00 -0400

I transferred to a new school in Ohio this year. The school has been truly great in many ways, but as soon as I settled in, I was hit with the managerial blunder of the leadership’s fervent implementation of Orah.

Orah

⊕
I’d like to talk to Mr. Trank Snith, please.

⊕Its registered NZBN is 9429041058619, if you want to look them up.

Orah is a self-proclaimed “school safety platform,” which is maintained by Boardingware International Limited, a small, private company. It was registered in Auckland, New Zealand, a country famous for their fully online registration policy which requires minimal documentation, with the company’s address still listed where a luxury restaurant and wedding venue now stands. You know you are in for a seamless integration experience when the company home page rotates between AI-generated images of happy students smiling with UI screenshots terribly photoshopped on, and especially when it exemplifies such impeccable traffic safety practices as a student pulling up a QR code when driving out of the school.

The problem with Orah is that it does not work. Even if it worked, the solution would still be impractical and possibly dangerous, but it just straight-up-frankly-for-real-seriously does not work at all.

Orah is meant to solve a few problems. Teachers need to take attendance fast, preferably automatically. The residential team needs to know where boarding students are. Students need to get emergency alerts reliably. The leadership needs a unified system to approve and track field trips, off-site extracurricular activities, boarder travel requests, and parent endorsements. Orah takes on the heroic task of amalgamating these requirements, working just often enough that it tricks you into thinking the problems are going to be solved eventually.

⊕
They are just regular passive NFC tags.

Attendance is based on NFC tags installed in every classroom. Orah mandates every student to use an NFC-enabled smart phone, and aggregates read events (“tile-in” actions) to a school faculty client. Attendance is updated automatically if the server detects a student’s recent interaction with the NFC tag. If the student is excused from a class period, they have to be labeled as excused with a specified purpose and time duration (a “pass”) by faculty members in charge of it. Residential students who wish to go off-campus apply for a “pass,” which becomes approved once it obtains parent endorsements and then staff endorsements. It becomes active with when it is activated manually by a faculty member or when the student “tiles in” at a physical NFC tag with the activation option selected; it is then ended with manual action or when the student “tiles in” again at a physical NFC tag. Parental approval is based on information on manually maintained student profiles, and parents confirm their endorsements through their own Orah clients. Emergency alerts are sent to student clients with options to confirm a student’s location and safety, and faculty members complete the roll call on their own clients. Residential staff members likewise complete roll calls at designated times in the dorms.

It seems to be a sophisticated system that solves all problems at once where management becomes effortless when user privileges are nicely controlled and when every single human being in the school becomes tightly-machined metal gears synchronizing in every move that culminates in a systematic, intricate dance. Why does it not work? To me, the better question is this: why should anyone expect this to work?

False Robustness

There are indeed many moving parts, and Orah does not always handle all edge cases correctly, even when nobody makes any mistakes. But it is easy to see that correct location labeling of any one student at any one given time depends on correct inputs from way more than just a few people. It requires an impossible level of procedural compliance. It is impossible for Orah to work.

⊕
They must have the fabled 10x reverse user retention engineer!

Let me give a few examples. We have an unusually high percentage of international students whose parents live in other countries and time zones. Orah is not available everywhere. Orah is not localized for many languages. Orah wants to send emails that certain non-US email providers refuse to process. Orah does not understand international phone numbers. Orah would like to track and compare your IP address to make sure a student is not approving their own pass on the same device. Orah does not understand information sent from censored parts of the internet, which exists in more countries than the average U.S. citizen can name.

⊕The IP tracking practices are confirmed in their Cloud Security Whitepaper. Yes, they are on AWS.

I have personally been given a few automated infractions throughout the year. One time, my “pass” has been accidentally ended by a school nurse which resulted in my account stacking up with an entire day of unexcused absences that equaled 10 hours of detention. Orah’s confident accusation took a few days to undo. Not too long ago, I missed a class period to do some technical theater work, which all relevant people were informed of, but Orah shenanigans still resulted in my parent getting a bureaucratically worded email about my infraction.

My father passed away in February this year. I believe faculty members were quickly informed of this, but Orah kept imploring me to ask for a dead person’s endorsement. I, although it was the last thing I wanted to bring up with someone else, asked for my profile to be changed as soon as possible. I kept asking for a month, and it was finally fixed. Ever since that happened, my mom always showed up twice on endorsement lists in Orah.

These are not insane, inconceivable edge cases; these are things that occur regularly in a small school. From my time teaching ESL to primary school students and from my time being a student, I believe education has fascinating formative influences on people. Most students live in incredibly small worlds—the choices available to them are decided by a very small group of people. The limited number of people, places, and institutions they interact with inevitably start to take unbelievable importance because the world you do not have access to might as well not even exist for you. Small infrastructural obstacles can seem like insurmountable boulders. This is the consequence of false robustness.

⊕Well, it is still enough of an inconvenience that I wrote this whole thing.

To me, Orah problems are mostly just inconveniences that I feel annoyed having to give nonzero time and energy to address. It is a solution to the intertwined, complex system of school management with another more frustratingly intertwined, complex system. I believe it is common sense that systems like Orah which depends on correct input fail at the weakest link. It does not matter if the links are hidden with a sleek user interface when the links are real people with people lives and people problems logging into a software ten times a day.

Insidious Temptations

Orah is dumb, but its absurdity is mirrored in many more supposed solutions to issues.

Managing schools is understandably difficult. Orah provides software that models this complex problem, in which procedures that require actions from faculty members become abstractions like “passes,” “tiles,” and “endorsements.” From a system design standpoint, abstractions are not productive if they do not come with a set of implied constraints on the sorts of actions you can do with the abstract objects. Naturally, Orah cannot offer infinite customizations and flexibility; it is optimized for the company’s idea of a conventional use case.

However, a system that solves issues in another system is still just a solution; the solution’s existence does not justify itself. There is a balance between the complexity, or the upkeep of the solution and the complexity of the problem itself. For more personal tasks, this balance is usually incredibly variable. Some people get by remarkably well without any sort of task management, while others are motivated by the process of writing carefully planned-out bullet journals. In an organization, however, this balance is usually narrow and completely dependent on what individual members tolerate. It is painstakingly obvious when it goes wrong.

⊕A system’s complexity might not be obvious at first. Orah is sleek and somewhat user-friendly but extremely complex to manage. An opposite example is Johnny Decimal; it might look intimidating but usually only needs about a day to set up and minimal time to maintain.

People nevertheless stick to these systems. There are the obvious factors like sunk cost and the simple fact that people who provide solutions for money want you to stay dependent on it, but there is also a subtle, insidious temptation of contrived systems—it provides comfort. It is a convenient wall of impenetrable bureaucracy to hide behind. After all, if individual links are connected by the system, of course we can just blame the weakest link when it fails! One does not have to take on very much accountability at all when people inevitably make provable mistakes.

⊕
Come on! It’s not even a real status page.

And of course the system gets better as time goes on. Maybe people will even stop making mistakes! Orah probably will improve in the practicality standpoint, but us participants bear the burden of user testing. Every edge case we encounter and struggle with is something that Boardingware International Limited should have fixed long ago, yet the school pays subscriptions to the company, who in turn gets real user data instead of having to do good old fashioned grown-up software engineering with unit testing and whatever kind of design awareness that has been paraded around the magnificent seven tech companies these days.

⊕To school faculty if you are reading this: do not get any ideas about AI developed all-in-one student management software.

At the time of writing, web applications assembled by AI agents are definitely on the brink of being good enough. As people begin to realize the unbelievable ease of building haphazard, contrived systems with a sleek user interface, I hope it will be an opportunity for more people to experience the tradeoffs of system design first-hand. Layers upon layers of bureaucratic defense will fall into the hands of even the most ordinary of people, and hopefully we can quickly realize the absurdity of it all.

Solutions

Bad decisions like this of course happen to almost every institution. This seems to me, however, an instance where no improvements would ever be seen as really necessary. People annoyed by it do complain, but none of us are really that seriously harmed and therefore equipped with the time, energy, and motivation to say anything too crazy. Nevertheless, I am sure it will never stop being absurd and frustrating as it permeates so much of our daily routines.

I do still remember how people would apologize to me every time I mentioned my father was still on Orah’s list and it needs to be fixed. I know what I am supposed to say—I know, it is Orah. I say in a defeated tone, as if implying it is a well established fact how much we all hate Orah. You cannot possibly do anything or be of any fault, I say. It is so unfortunate that Orah is here and everything is like this. We just simply cannot have nice things. No! No, this is not how it works. Why would anyone think that is how it works? I know that these people I interact with are somehow motivated to keep using Orah. It is not us versus the system, it is me versus your good intentions! How can you pretend you are the system and it is just an unfortunate fact? We see each other every day. I know you are a nice human being, and my assumption of your human-best-effort decision making does not stop upon any mention of Orah.

Anyway, I hope this illustrates the dilemma we all may face one day in system design. Let us keep living in this happy world together and solve all the problems because I believe we can do all things. This website is indexed by Google and even has decent performance for certain articles sometimes. I hope it becomes the second search result after Orah’s official website.

The Scenography (and the rest) of She Kills Monsters

Thu, 26 Feb 2026 07:00:00 -0500

My school recently completed a production of She Kills Monsters written by Qui Nguyen, in which I played Vera/the Beholder in addition to doing miscellaneous technical work. This is the first time I played any role in a serious production, and I also spent quite some time on building the scenography that accompanies the set. This post is a retrospective on the technical aspects of the show.

I really do not think I will ask for everyone’s consent to be in my blog post, so this is my attempt at painting over everyone else’s face.

Setting the tone

⊕
Yes, this is really how it looks.

Candidly speaking, I am not a big fan of the writing of the play. I do think it tells a somewhat cohesive story with a message that I really, really appreciate, but the somewhat fragmentary story arcs can distract the audience from it. Some of the plot points it opens up are unresolved or unexplained, and some of its humor are extremely unsophisticated. However, most of these are excusable considering that it debuted in 2011.

When I got the script, it came with a really, really awful design that looked like it was drawn with a mouse in powerpoint. Maybe this made sense for the publisher because most of the story happens within a highschooler’s custom DnD module, but this really is unacceptable for me.

The two lowercase L’s were tweaked to improve legibility.

⊕

I started with the title typography, for which I imagined a variety of different styles as concepts, including a romantasy-style serif, modern stretched sans-serif, and maybe just Helvetica. In the end, I settled for a stereotypical fantasy visual style with Roman drop capitals, insular minuscules, and Latin marginalia.

The minuscules went through some iterations for legibility, but there is only so much you can do without completely forgoing historical, or even stylistic accuracy. The Celtic knot behind the drop capital is a traditional design fused with a Borromean knot inspired by Lacan’s diagram of the sinthome. It is then furnished with the Tiamat drawing, which I hypothesize may be completely inaccurate due to my lack of DnD knowledge.

I collected other photos online and from the school’s photography class. Most of the fighting and drama occurs in the DnD fantasy world that reflects the dead protagonist’s struggle in high school, so I decided to use real photos of the school throughout the scenography. To delineate real-world from fantasy, I drew over the photos used in the fantasy setting and added a colored glow to some of these elements. I did not have the time to plan what to add or make separate backgrounds for different stages in the adventure through the fantasy world, but the intention is, at the very least, legible.

Tech Chaos

Some scenes required some hand-drawn “manuscripts”, for which I reused the background from the title typography. This included another sketch of Tiamat, which I tried to give a faithful rendition in reference to the 5th edition’s official art. There are two separate maps: a regional one used at the beginning of the show, and a large, detailed map of the entire “New Landia”. Since the show is set in Athens, Ohio, the prop team gave me the bright idea to make the shape resemble Ohio, like what was done for a separate map prop.

In the prologue, one of the protagonists dies in a car crash. I decided to pick up 3D animation again and rendered a somewhat awkwardly animated car crashing into the virtual camera and fading to white. This led me to wrestle with Blender’s volumetric fog rendering for quite a bit, but was overall easy thanks to an awesome, almost ready-to-use model of the Land Rover Defender 110.

Surprisingly, it did not take too long to render.

The finale, a fight with a symbolic Tiamat, is the most complex asset in this set of projections. I decided to go with a procedurally generated 3D glowing spiral for the fight because we have people wearing dragon masks instead of a giant prop dragon to wiggle on stage. The ending, to me, lacked some of the necessary explanation and emotional build-up that would make this work as an true resolution. Because of this, I took the idea from the director to do a very literal flashback with bold typography that made sure the audience understand this is a mental battle for the protagonist to reconcile with her past.

The Show Begins

All the projections were programmed into QLab, which we just recently got a license for. QLab uses a very pure procedural logic, which was confusing at first. I anticipated that an expensive, professional software would prioritize ease-of-use, but it turned out that every parameter is controlled with a fader and not even a crossfade is built-in.

QLab had automatic infinite looping for videos and audios, but the complexity of the goto cues—I cannot believe the programming is more similar to assembly than video editing—made me decide that it is better to cut the video from the middle and stitch the two ends together with a crossfade in KdenLive to make the video assets seamless loops.

Running the cues demanded some practice from the technical crew, but the projection cues turned out to be the smoothest compared with lighting and audio. The two performances went mostly smoothly, which were a lot less stressful than I anticipated. It was a welcome success for my first time performing on stage.

I also want to mention that my involvement in the entire project started from several unlikely coincidences that made me consider participating in theater, which is something I have never even thought about before. I am surprised to see that a big part of my skillset became useful in this context, and especially in a small production with few professional resources outside of artistic direction. It gave me a reason to produce a larger-in-scale project of graphic design, which I appreciate as an addition to my archive.

mi tawa.

A Stupid Person Believes That Memorialization Blocks Anticipation

Sat, 23 Aug 2025 12:00:00 -0400

My friend recently drafted a lengthy yet dense article about the mourning and becoming of queer identities. I want to make a point here that part of the thesis, unduly summarized in the title, is a little stupid.

Jido makes a terrifyingly tangled analysis about objects that we think can be mourned and objects that we think can be desired. The outstanding citation to support this comparison is Testo Junkie, which contains statements that are too correct.

It is too correct that traits are stated to be divided into this dichotomy by ideology, because things have always been this way, things have increasingly been this way, and things will always be this way. Queer people sometimes reject one side of the dichotomy and sometimes the other, but never both.

Take this example: I am me when I am respected. I am also me when I am desired. The identity, which I am not going to go into today, is an intertangled mess of how one desires to be respected and how one’s desires are respected. We should not confuse this, Jido, as I have said, and as many people only seek being desired through traditional means of being mourned. You are partly right; marginalized subjects mourn the lost opportunities to be accepted, to be respected, and to be desired. But what do you feel when you say this?

Does it not feel like the object of mourning is something to be, and the verb is something to do?

Memorialization seems like blocking a future, but it is shaping the future. Many speak of shaping an identity, but who browses in the identity store and selects base packages and add-on packages and premiums and markups? I do not choose the clothes I wear, the body I have, or the things I am accustomed to. I resort to them. A resort is a memorial, a memorial of things I have done in my clothes, the people I have disappointed with my body, and the actions that has passed through my instruments. The only things I also resort to is anticipation; be it redemption, revenge, or repetition.

A living memory is a memory that hopes—that hopes to be forgotten.

In case of any serious readers, this article is just as unorganized and unedited as the draft that inspired me. As always,

This is a short-short article in a series of short-short opinion pieces. Ideas are of my own except when they are not.

Ken kama suno lili la

Sat, 23 Aug 2025 08:00:00 -0400

This is a quick cover of はるまきごはん’s 彗星になれたなら, which is my first toki pona cover featuring my vocals. I did not have the luxury to put in a ton of effort on the music and graphics, but I think the homebrew feeling is somewhat charming in retrospect.

The music featured a guitar with a pretty aggressive distortion, which hides the sloppy execution and makes the mix more solid when there is no drum or bass. In some places, the vocal is just out of tune enough to make me uncomfortable; I tried to add a bit of pitch correction, but I thought the robotic snapping sounds worse. It is indeed very time-consuming to post-process vocals that are not great by themselves. When the tone wobbles a lot, the pitch correction needs to be adjusted dynamically to reduce the “snapping” sound. At some point, it would be more efficient to just record a few more takes.

However, I am fairly satisfied with the translation. It seems to be more in line with the toki pona philosophy than previous translations. I also did not have to make too many weird grammatical decisions, but it still generally rhymes. Since this video does not include them, I include the comments on the translation with this blog post:

laso sewi pi suno pini li kule pona pi nanpa wan tawa mi
The blue-green of the end of the sun is the best color, in my opinion
tenpo ni la, linja lawa jelo sina li suno pi utala ala tawa lukin mi
At this time, your yellow hair is a peaceful light to my eyes
suno tawa lili pi tenpo ni li
The small moving star of now
sona ala e toki pi jan tu ni
Does not know about the conversations between these two people

The song has come out for quite a long time, but frankly I find the original lyrics here a bit weird. It reads like lore for lore’s sake in the Harumakigohan universe. “suno tawa lili” might be a bit unclear, especially when the entire translation uses the word “suno” frequently, which also appears as a verb. I am not sure if “star” or “comet” will occur to the reader at first. The word order at the end was changed, but the implied third person is preserved.

mi wile kama suno ni
I want to become this star
mi wile tawa laso sewi pini
I want to go to the ending blue-green up above
sina kama weka tawa mi li tan ala tan ni
Is it because of this that you are becoming distant to me:
jume suli li lon insa lawa mi
There are big dreams inside my head?

“laso sewi pini” is somewhat forced by the i-rhyme. “tan ala tan” is ungrammatical, but occurred to me very naturally with the X ala X construction. This might also be the only place where this construction would make sense.

tenpo seme la mi kama sona ala e sitelen sina
When will I become ignorant of your writing?
mi li kama jan moli ala taso la
When I become a person who merely does not die
ken la pilin ike suli mi li kama ala
Perhaps big bad feelings of mine will not come

“sitelen” is interesting in the way that it can be interpreted as writing or signs in the figurative sense. “taso” used in this way is a bit roundabout, but interpreted as “someone who merely lives” here is a surprising efficient way to express this idea.

tenpo kama la, ijo ni li lon la
In the future, when this becomes the case
o weka e ni, suno lili o
Destroy it all, little star

“weka” is kind of a callback to “o weka”. I did not like the original lyrics of this very much, either; I did not think it made too much sense. Translated into toki pona, however, everything has the potential to make every sense and no sense at the same time, so this is not a concern at all.

ken la, pali mi li pini
Perhaps my deeds are done
ken la, ni li ike tawa jan suli
Perhaps this is bad for the big men
mi sona pona e pilin pona lon uta sina
I know well the pleasant feelings on your mouth
mi sona pona ala e ni: mi lon
I do not know well of this: I exist

Harumakigohan does cartesian philosophy. “jan suli” is convenient because it can mean “grown-ups” and the idea of “authorities”, which is fitting for a different interpretation. I quite like the contrasting structures here, which is a very toki pona way to make comparisons. The simple ideas somehow sound poetic in toki pona. Another interesting thing of note is how “pilin pona lon uta sina” can also be interpreted as the pleasant feelings of kissing, but I doubt that Lili and Nana canonically kissed.

sama jume la olin o
As one dreams, please love
sama olin la pali e jume o
As one loves, please make dreams
jume li suli taso pilin li suli suli la
Although dreams are big, feelings are bigger
tenpo kama la wile ale li lon
In the future, all wishes will come true
sijelo mi li ken kama suno ni ala taso
Although my body cannot become this star
o pilin o olin, jan lili o
Please feel and please love, little person

These first two lines were how this translation started. You might notice that I skipped a verse, mostly because I did not want to put in too much effort. The use of “suli suli” is probably another very toki pona one, which also just sounds good when sung aloud. “jan lili” is a stretch but contrasts with the previous “jan suli”.

Overall speaking, this is another lighthearted cover without much effort. I do think that I am getting better at translation and poetic language in toki pona, so I am looking forward to what might be to come when I get inspired next time, when I will hopefully put more effort in the music.

mi tawa a!

It Is Called Footnotes Because We Are All Getting Stomped to Death

Mon, 11 Aug 2025 20:00:00 -0400

“If our union is destined to turn into separation; if it is destined to be pathologically misread and abbreviated, then so be it if we can only embrace each other under the footnotes of DSM-5 and ICD-11.”

This is my rough translation of a Tumblr-style post I have come across lately. I absolutely hate it, but I hate it only because it sounds a bit cute and even a bit charming, which I am sure other people will agree. It inspires a few questions, although posited countless times and answered in countless ways, that need to be answered in this exact context. How are diagnoses and relationships connected? Why do we think about mental struggles in terms of diagnoses? How is it romantic that we are in the footnotes?

There is a terrifying yet promising amount of discussion about mental struggles. However, the bulk of the discourse in mainstream media hinges on diagnoses. Is this disorder real? Are people faking that disorder? Is this disorder overdiagnosed? Do doctors know what they are doing when they treat that disorder?

This discussion is useful; it is certainly important that we classify people into increasingly neat little boxes and decide if they deserve human rights based on the neat little boxes. However, it is concerning how much the validity of a certain feeling or getting a certain treatment (both in the social and medical sense) hinges on the validity of a diagnosis. People are increasingly paranoid about not being able to tell if an emotion is genuine that they pass the ball to a medical profession that sees the person for thirty minutes every two weeks, then the ball gets passed to a piece of paper with questions on it, so on and so forth. Much like how I do not have to explain the utility of psychiatry, I do not have to talk beyond making the point that psychiatry currently features some pretty funny stuff, either. So let’s get to the point. How does it all relate to relationships?

Diagnoses are when the conversation stops. Like all explanations, we would like to not explain why we are sad, angry, or anxious up to a certain point. A diagnosis is a proxy to a non-answer through science; you can stop asking about why when I say it is a chemical imbalance, or it is a developmental disorder, or science just does not know yet. Diagnoses help people identify other people who are also annoyed by having to explain themselves to no end, and more broadly, oppressed in roughly the same way. But at the same time, the label of a diagnosis provides a means to a fantasy, where the illusion hinges on the mythical, unifying, hidden cause. In relationships, a shared fantasy is a good fantasy.

A fantasy backed by science is a convenient fantasy. We would like to think that everyone has different experiences, and therefore different struggles. But all this becomes very inconvenient when we want to find a way out of this, especially some quick, proven, and commercialized way. By relegating the burden of proof on science, we are free to treat mental struggles more like natural disasters or accidents than we are forced to think, when free will is assumed without access to free choice. Does this sound dangerous? Yes, a popular fantasy is also a dangerous fantasy. The fantasy itself might not be harmful by itself, but popular fantasies often enter policy-making before good explanations of the fantasy does.

It is called footnotes because we are all getting stomped to death. Everyone subject to a label must endorse, implicitly or explicitly, a position on the accuracy of the label. Some argue that a label is the first step towards awareness; others argue that any generalization will backfire and jeopardize people’s trust on others being able to make decisions for their own. But what we are often forced to do is to accept a label on a condition. Do people think that it is romantic to be in the footnotes because they can finally have labels, or because it is brave to enjoy (or suffer from) relationships despite they are getting stomped to death?

In most cases, it is a little bit of both. Psychiatry is a special science for confidently generating so much utility and so much suffering despite the unignorable uncertainties in its findings. Would you take a pill if nobody is sure about how it works?

I would. Not because I do not care about how it works, but because if it does work, all this question of why and how becomes irrelevant. What if it does not? Well, I am already in the footnotes, anyway.

This is a short-short article in a series of short-short opinion pieces. Ideas are of my own except when they are not.

Mourning is not a Task

Tue, 29 Jul 2025 20:00:00 -0400

Think about the conventional language we use to talk about mourning. “Have you recovered?” “Have you processed it?” “Yeah, I’m on good terms with it now.” “I need a few more months, I’m still a bit lost.”

They seem to suggest that mourning is a process that can start at a specific time, can be worked on throughout a period of time, and end when the process is finished. This makes sense on a surface level; according to our capitalist and granularized perception of time, it is possible to put a timestamp on everything. But is it?

Mourning, like all consequential processes of human emotion, is complex—not just in the way that psychiatrists say depression is complicated as a disclaimer or HR say the situation is complicated to give you time to brace for impact—in the way that the discussion around it is terrifyingly one-dimensional (in fact, increasingly one-dimensional and vague) compared to the vast structures it actually is. Think of Bowlby, Kubler-Ross, and Stroebe-Schut. Psychology that confirms common sense is nothing short of productive psychology, but it can seem unnecessary to bombard people who seek help with quarter-baked worksheets based on half-baked theories, even if they come from well-done research.

It is worth repeating that mourning is complicated. However, it is worth first challenging this task-based conception of mourning, especially in a world where finance bros and hustlers encourage people to slap AGILE goals on characteristically ever-changing processes.

So what is mourning like?

Mourning is like inserting punctuation in a never-ending monologue. Inserting punctuation is finding structure where structure is not obvious, but it is also understanding what is not yet understood. Punctuation does not soften harsh words in the monologue but finds logic in an otherwise chaotic thread of pain. New patterns emerge and disappear over time, never static, but always on topic. The work is never done, as long as you keep living and conscious, but you do not remember for sure when it started, either. An unreliable memory ensures that a new stable structure emerges every so often, but destabilization, whether intentional or unintentional, is never out of the question.

Mourning is like a long, drawn-out game of correspondence chess. Like a game of chess, mourning comes with certainties, uncertainties, surprises, and disappointments. You play moves that represent the future, whether hope or desperation. However, the moves of the opposing side play the remnants of the past you already know, but you might still sometimes disoriented by the harsh reality once so thoroughly understood. The game always ends in a draw; not signifying retaliation, but a pause and a possibility for a rematch. Maybe you do not have time for this by the time after the first match; the past is always there waiting for you.

Mourning is the rituals, celebrations, and rites of searching for the lost one. Mourning does not have to be inherently sad. Celebrating the time that has passed without something is just as valid as celebrating the time with it. The rituals do not have to be traditionally associated with mourning—you can borrow from birthday celebrations, pagan sacrifices, or even wedding ceremonies. When does the process of searching for the lost outweigh the desire to find? However different they may be, a new life is a continuation of the old, and one of them can help define the other. Reframing an event is not the whole story, but when someone sincerely celebrates being abandoned, despised, or left alone in this world, frank sincerity must outweigh the defensive irony.

Why see mourning as a progress bar when it is usually not? It is important to recognize when removing humanness from human desire goes overboard; it is productive to be confused, to be furious, or to be devastated. No, they do not even need to be productive to be allowed existence—they are deep marks of our compulsive resilience in a world where it is impossible to be.

This is a short-short article in a series of short-short opinion pieces. Ideas are of my own except when they are not.

Anger is Opposite to Revenge

Fri, 25 Jul 2025 20:00:00 -0400

Sea and headland now grew dim. Pulses were beating in his eyes, veiling their sight, and he felt the fever of his cheeks.
James Joyce, Ulysses

Anger is often portrayed in accompaniment with revenge. In fact, most of the symbols of anger—fire, the fist, or the veins on one’s head—double as symbols of revenge. However, there are differences between the two that, however subtle, put them opposite to one another.

Anger has a very specific meaning in psychoanalysis. Anger, if distinguished from a desire to revenge, is an affect. From this perspective, there is no such thing as unexpressed anger. In fact, there is no such thing as accurately expressed anger, either. An affect is part of the Real; it slips through the cracks of language but hits you like a brick wall after you charge full-speed at an object of desire on the other side.

Revenge lies on the finish line of the desire to revenge. The desire, like all desires, is a self-defeating one. Revenge can be done; once a revenge is successful, there is no point for the exact same kind of revenge to be repeated anymore, much like the desire to watch a show for the first time. Revenge does not slip through language; it only has meaning through the means of language.

These statements can seem kind of hand-wavey and mostly speculative, so it will be helpful to discuss a few examples.

Revenge incites; anger corners.

Anger, when it first manifests, might not look like anger. It might feel like despair, sadness, or even anxiety (in the modern sense; not psychoanalytically). It can look like something you can overcome by practicing anger management or vent it out in a rage room, but it demands confrontation and direct acknowledgment by cornering someone into a position where they are forced to do so.

In the Iliad, anger is portrayed as somatic and visceral: “[Agamemnon’s] heart was black with rage, and his eyes flashed fire…” Almost as if on their own, cornered by the language and unwavering rule of the gods and goddesses, his heart and eyes are moved by this affect before he himself realizes it. Somatic clues are often interpreted as preceding or causing psychic ones; even if the intelligent self tries to repress, anger corners the repressed parts of one’s consciousness into admitting its existence with the body as a starting point.

No one understands revenge; anger understands no one.

Revenge often amplifies itself it a process of inflating the object of desire each time it is finished. Towards the end of The Odyssey, in which Odysseus embarks on an epic and frustratingly drawn-out revenge (among other things such as indulgence, strategy, and fear), he kills all the suiters of his wife, therefore starting a terrifyingly massive war between the inflated subject—his forces and followers—and an inflated object—countless families’ revenge of the dead suiters.

The ending of the Odyssey is as frustrating as it is inevitable. The revenge is amplified to a divine feud, but it is resolved without much elaboration. It shows that nobody understands revenge, or knows how to resolve amplified revenge, not even the divine powers. Anger, however, is the brick wall—it hits everyone without discrimination, as if it understands none of the efforts to repress it, not even the fantasy of not acknowledging it at all.

Revenge lingers; anger moves.

In Pu Songling’s Strange Stories from a Chinese Studio, Nie Xiaoqian lingers in the world of the living, her existence dependent on her desire to revenge her wrongful death at 18 years old. She plans an elaborate ploy; not forcing, but tempting the victim, Ning Caichen, towards her trap to take his soul. Revenge lies in strategic planning, towards a narrative of fairness where others suffer just like what the subject experienced. Revenge stories are possibly most powerful because the desire lingers in the characters and in us, the readers, aligning us with their fictional rage.

How does Nie Xiaoqian resolve this desire? Perhaps disappointingly for us, she subverts her desire into experiencing a life after 18 years old that she thinks she should have lived, marrying Ning Caichen after his wife died, fulfilling domestic duties, and ultimately having a child with Caichen. Readers familiar with the story will recognize that Ning Caichen is portrayed as a perfect subject (which arguably becomes an object, perhaps all-consuming object) who resists temptations and only desires within the traditional Chinese cultural framework. He is unmoved by anger just as how Nie Xiaoqian is moved by hate; he is a most passive traditional subject just as how Nie Xiaoqian is a most passionate traditional object.

We have seen how anger and revenge coexist but lie on opposite ends. Like how Buck Mulligan experiences the unignorable pulses of blood and heat on his face, anger, as an affect in the Real, reminds us of what it means to be a subject who desires. Anger can prompt revenge or be interpreted through revenge, but their two forces pull on the opposite ends of the rope.

“—O, an impossible person! he exclaimed.”

Simply a naive denouncement or not, the impossibility of aligning a directional revenge with an aimless anger makes us persons. Possible persons who fight, yet hopelessly and unwillingly coexist with the impossibilities of being.

This is a short-short article in a series of short-short opinion pieces. Ideas are of my own except when they are not.

The Forth-Takes, Volume 1

Mon, 21 Jul 2025 20:00:00 -0400

I am excited to publish this post accompanying my first mini-LP, The Forth-Takes.

The Forth-Takes Vol.1 by Kai Wang

Songs in The Forth-Takes are all recorded in four takes or less, uses sampler-based midi instruments, and based on other melodies or ideas. Its name, The Forth-Takes is a modification on the Japanese series The First Take, while “Forth-Take” is an inversion of “forthbring” (or, in its modern form, “bring forth”). Not too much thought were put into every song; in fact, most of them were recorded in under 40 minutes.

Improvisation can be fun, but one makes mistakes in improvisation. I have to admit that some sections are hard to listen, but some of my friends have suggested that they can be enjoyable for certain people. I decided to stay authentic to the first (or rather, the four first) takes and preserve the wonkiness and strangeness occasionally involved. Therefore, it is not a heavily produced album. It is best understood as a snapshot, both in skill and spacetime, of an improv session I might play at home, only that there are four of me each playing a different instrument.

The recording strategy used is also rather primitive. Most instruments are free (libre) SF2 files I found online. They are only static samples and no more, which are very basic compared to, for example, sounds by Native Instruments. No effects are used except for a reverb, a compressor, and a global EQ. Mastering is limited to making the LUFS levels consistent across songs, as well as conforming to streaming standards.

There is little consideration into the arrangements. However, there are a few things you might want to know.

Track 1 is based on a previous improv recording, which maybe sounded a little bit better. It is also my first complete arrangement that is more original than a straight cover. The chorus of Track 2 takes inspiration from the original melody and uses a 5/5/6 meter, except when I failed to follow my own rhythm. The synth in Track 4 is inspired by the toned-up Kasane Teto from the original song. I think the sandy tone of the synth replicates the original voice quite well. The piano solo in Track 6 was meant to be the “Happy Birthday” song because that is what the original song is about. However, I could not for the life of me remember the tune at the moment and played “Twinkle Twinkle Little Star” (or, Baa, Baa, Black Sheep, or, The Alphabet Song, or, Ah vous dirai-je, Maman) instead. The end of Track 7 quotes the national anthem of a certain country, I will not say which one because I do not want to get arrested.

As suggested in the name, I plan to compile more albums in the future. They will be just as improvised as volume 1 is, but will hopefully be snapshots in skill and spacetime for when I am better at improvisation, arrangement, or music in general. You can buy the mini-LP on Bandcamp for 0 dollars or more, although I have not figured out how to deposit the money from Bandcamp yet.

mi tawa.

A Modified QMK Firmware for the Polyglot

Sat, 12 Jul 2025 20:00:00 -0400

I never really did programming after training for (and failing at) computer science Olympiads. The two projects I currently have are IRAHET (which is just an interface for LLMs) and Assistant Maker I made for MWH (which is embarrassingly unfinished).

Anyway. I found my Stenokeyboards Polyglot today. It really is a fantastic keyboard, although I was bad at stenography and gave up before I got to a level where I can use steno to write an essay without going crazy. Without the custom keycaps, though, it is an orthogonal and small keyboard. I realized today that I have always wanted an orthogonal, small keyboard. I am also starting to use Colemak, so I decided to modify and flash a new firmware.

I replaced the original keycaps with this set of cutesy ones. They are in the Colemak layout, but the firmware maps all keys to Qwerty. Many operating systems have Colemak available in the language settings.

The process was fairly simple, and I reckon anyone who has compiled C before could figure it out in an hour or so. If you have the time, it is definitely easy to design and flash a keymap better suited to your needs. However, in case anyone wants to use my keymap and flash the firmware I have already compiled, the layout is here:

I might make a digital version of this if anyone else decides to use the firmware. The backspace is in Colemak position, because Windows does not handle capslock as backspace properly.

You can find the uf2 files ready to flash here. The keymap file is also in the GitHub repo, and it should be clear how to modify it. Refer to the QMK documentation and the original firmware if you need to.

mi tawa!

Three Ivy League Students Fail to be Mr Beast

Mon, 07 Jul 2025 20:00:00 -0400

To the creators of the original video: neither my blog nor your video currently attracts a lot of attention. If this is no longer true for any of us in the future, I will probably happily remove this post if I get contacted.

“LAST TO LEAVE HOT TUB WINS $10,000!!!!!”

This is a YouTube video with 80 views. An article on WeChat rolls out, advertising the video, filled with incomprehensible phrases like “contemporary art narrative”, “social installation art”, and “a revolt against youth identity, consumer anxiety, and academic burnout”. The Asian girl in the thumbnail dissociates into oblivion as her face gives up on moving a single muscle. The other girl makes the Mr Beast screaming face but with genuine fear in her eyes. Their hair all wet. They look like they want to kill themselves.

The article on WeChat introduce the two contestants. They study at Columbia University. The cameraperson studies at Yale. This is a Mr Beast video. They want to be Mr Beast.

Okay. I do not want to paint an overly negative picture here. They are trying to make a video, maybe trying to go viral. The WeChat article probably pitches the video to another audience that may prefer a different reading of the content, presumably someone who cares about which university they are studying at, or someone who would like to treat the video as a “social installation art”. There is some vague activism going on here, when the competition is described as “a form of capture, demonstration, and compromise of female bodies in urban spaces”, but as I said, not everything that sounds feminist is feminist. And no, this video is not at all similar to works by Santiago Sierra when he paid people to sit in cardboard boxes.

I do not want to say that making a fun, little, and (not yet) viral video is morally wrong, but I do want you to grasp what they are doing and what they would like to be doing. Why would three Ivy League students try (and fail) to be Mr Beast?

Mr Beast is not a particularly good person. Furthermore, there are certain things associated with him that do not carry nice connotations, like YouTube in general, the manosphere, and influencer bros. The only reason people want to be him is, perhaps, to be famous and be on Joe Rogan podcasts. Their video is not particularly bad or boring, but the thought process they represent troubles me. Some people think that they can be Mr Beast and use their word salad to pretend to be philosophy tube, maybe even be Jordan Peterson and use their word salad to pretend to be Judith Butler. This is horrible, horrible word sorcery. It is even more horrible that word salad gets easier and easier to pull out of thin air over time, thanks to the emergence of machines trained and specialized in salad-making that everyone is forced to used now. I fear that we get used to that someday in the future.

It is not just that Mr Beast is not a particularly good person. It is that Mr Beast represents mass-produced content-for-the-sake-of-content, and people try to be him. It is that Mr Beast represents spectacle that does not care about the spectacle-makers, and people try to be him. It is that Mr Beast pretends to be making a difference, being inspirational, and being educational, and people try to be him. It is that he is popular, and people try to be popular in the exact way he got popular.

So creators, go be your clean girl influencer, go be your sad beige tiktoker, be your internet atheist skeptic, be your true crime storytime fursona! Please, Be anyone you want to be!

Please, do not try to be Mr Beast.

This is a short-short article in a series of short-short opinion pieces. Ideas are of my own except when they are not.