February Update

Much happened already in the last two weeks and the next few weeks are guaranteed their own share of novelty, now is a good time for a checkpoint. My apologies for this unplanned and, hopefully, unusual burst of writing.


Last Wednesday, I was driving ~450km (~280 miles) back from a visit to my mom. The previous sentence forms a shared suffix to four different stories.



We do not talk much. I see her between one and three times per year. We find it hard(er) to maintain a working relationship with a closer contact. I had visited last in September 2023, and would usually visit around the end of the year; this year however, I was abroad in December and January, and so I had offered to visit at the start of February. So I did, and last Wednesday, approximately on our planned schedule, I was returning home.



My mom experienced two heart infarcts in December. She did not communicate this to me. I only learned from my brother three weeks later that she would be heading to the hospital for treatment. The diagnosis and prognosis at the time were unclear: I learned that the cardiologist had detected major heart dysfunction but also that the treatment needed not be delivered in an emergency. My brother estimated that my presence was unneeded. I called my mom and she sounded upset at the time, but also clear of mind and cagey about expected outcomes. Trusting my brother’s judgment and considering the complexity of trekking all the way from Brazil, I decided I would use my originally scheduled but later visit to check on her health and accompany her to her followup treatment. Only later, after the first treatment was successful, did I learn that her situation, albeit unfortunate, was routine: she received three stents and is now recovering well. Eventually, relieved, last Wednesday I was returning home.



My mom suffered severe health events four times in her life. Severe as in, “no known treatments and doctors were expecting her to die within a year.”

The first one occurred while I was still a child. Although I barely remember the specifics, I still remember the emotional trauma. It was also my first experience of premature mourning. Then, thanks to technical prowess from a surgeon and, at the time, her relative youth, she unexpectedly recovered.

The second event occurred during my teenage years. The disease was obscure and incurable. For the first time then, she sat me down as eldest child and walked me through the logistics of managing without her, if and when needed. I do not cherish the memory of those conversations. I went through premature mourning once more. Then, thanks to an unexpected oddity in her genetics, her condition permanently stabilized—at great sustained expense from her immune system. From then on, she was permanently allergic to any and all painkillers and most other pharmaceuticals. This is how and when she started to mistakenly internalize that the medical establishment would not ever have anything more to offer her.

This is indirectly the cause for the third event, which occurred a few years ago. The issue was not COVID-related and instead a predictable consequence of avoiding encounters with doctors for too long. As the lockdowns prevented my visit, we went through the motions of parting again over the phone. This, too, was unpleasant, and I experienced premature mourning again, albeit “remotely”. Then, she was socially coerced to seek professional help, and despite a pessimistic prognosis she recovered, again. Not very well, but still.

The last event occurred last year, before her heart infarcts. We were actually going through the motions of parting and grief again when the cardiac issues preempted that protocol and pulled our attention towards the shorter term. Albeit initially improbable, it is now possible that the improvements to her heart could help with the other challenge she was facing.

I was thus partly recovering from incomplete premature mourning, with a strong sense of emotional backlash, when last Wednesday I was returning home.


Segue—the strongest thought that accompanied me on that drive was thus: mourning is an important, natural process, but it is not reversible. Through mourning, the character of a person that was mourned already is gone and is not coming back. The body, and possibly the soul, of that person are still there, but they become new characters in everyone else’s stories. And it takes work to build those characters and redefine relationships. I lost my mother already four times, and I dread rebuilding our relationship each time, expecting to lose her again too early.

How to deal with this? It seems as if the premature mourning is the main source of complexity. How to avoid it? It seems unavoidable when one chooses to treat high probabilities of end of life as a visible end point on the proverbial line. However, remaining silent or too casual about health events does not feel respectful to loved ones. What other options are there?



My mom is not officially retired yet. For complex reasons, she was (not entirely incorrectly) led to believe she was not yet entitled to retire until she turned 71 years old. Her occupation is related to tourism and it keeps her intensely busy each year from April to November. So busy, in fact, that we can hardly spend quality time together when I attempt to visit during those periods. It is thus that for each of the last 15 years, nearly all my visits occurred in winter. The backdrop of our ongoing relationship is that of her empty work premises (little activity in winter) under gray, wet, cold and depressing weather.

I resent her work, and I resent the location she chose for this activity, because the work and location “stole” the happier and more beautiful/active features of our relationship away from us. Just being there brings out unpleasant feelings on my side.

Like every time I felt relieved to head to my happy(ier) place home and leave these bad emotions behind, when last Wednesday I was returning home.


Segue—it is disorienting to be sad and happy at the same time. In earlier times, I would also experience shame and guilt. No more of that, but the disorientation is still there.


Epilogue—my mom is now effectively retired, by mandate from her cardiologist. This might help us rebuild our next relationship upon more auspice grounds and temperate weather, both literally and figuratively. Time will tell.


Did you notice how I wrote the text above? Each of the four stories is coherent: I could tell each one alone and it would fully motivate and explain my travel. Each story is also true. Yet, the four stories would also be very hard to mesh together: a single text attempting to say all at once would be hard to follow and to fully comprehend. A combined or reduced story simply does not exist; their truth is essentially multi-faceted and only exists as seen from separate lenses.

I made the effort to share these four stories, at the risk of over-sharing my private family situation, as an illustrative example of what I am preliminary calling a theory of overlapping narratives. I think of it as a useful approach to share complex situations.


Last week, on Tuesday morning, I suddenly woke up. Multiple trains of thought had momentously collided together at once and it felt Very Important. Cautious about the possibility of torpor-induced confusion, I initially jotted some notes on paper and went back to sleep. One day later, I re-read my notes and even more thoughts flowed from there. It appears that I had a philosophical insight. The thoughts continue to flow to this day, and like Ice Nine they are slowly rewriting my mental model of the world around me. The title on my notebook page: “overlapping narratives”.


The two ongoing wars close to home are keeping me preoccupied. In addition to the induced natural and expected (albeit unfortunate) basely anxiety, I am intellectually troubled by the degree of apparent irreconcilable incompatibility of the opponents’ world views.

The only possible foundation for peace, history teaches us, is a shared narrative with intermediate outcomes acceptable (or, at least, tolerable) to all parties. In the past, I had believed that shared narratives are a necessary feature of reality; in other words, that through sufficient listening and empathy, parties would always eventually uncover and agree to a shared understanding of their conflict and an adequate resolution. In the current historical chapter, I was forced to consider that there may exists no such shared understanding—that the situation is just too complex for this to ever be possible. What, then, of peace?

Well, perhaps one next history chapter will contain a new version of peace, one that rests instead upon multiple overlapping (albeit possibly incompatible) narratives with just a few adjacent touch points.


Reading about trends online, especially those emerging on X/twitter, tends to irritate me. I used to find myself confused and disappointed at how large groups of people could start promoting incongruous versions of reality that simply do not stand even the most casual scrutiny. At some point, I learned that things are the way they are because The Algorithm presents different content to different communities. Each community is looking at the world through a lens customized to their background and primal fears, and so that is what gets promoted and amplified. Now I am not confused any more, but I am still irritated: why do the social networks continue to produce customized feeds? What prevents them from doing otherwise?

Well, as I learned over the years, most humans want the stories they hear to be (re)told in a language they understand, for the stories to use characters they are already familiar with, and for them to call for emotional states they are already comfortable with. There might be wisdom in “showing things as they are”, unfiltered, to everyone some of the time, but I now recognize that it would be cruel to expose everyone to input-related discomfort all of the time. From that angle, I now recognize that customized, or should I perhaps say, overlapping narratives are a necessary human comfort.


See what I just did again?

More fun may be had again later.


Just a bit more than two weeks did not enable long reading sessions. I did not open as many books. I was also less lucky than last with regards to quality.

I’m OK - you’re OK - Thomas Anthony Harris

Where the author introduces and explains transactional analysis, a model for interpreting psychological behavior. In this model, all interactions between people allegedly reproduce features of either the Child (I’m not OK, you’re OK), the Parent (I’m OK, you’re not OK), or the Adult (I’m OK, you’re OK).

As these things usually go, I found it a tad reductionist. Yet, it felt adequately different from other things I have read about practical psychology (attachment theory, etc) and I filed it in my mind as “perhaps useful”.

The 5 levels of leadership - John C. Maxwell

Where the author tries to share his “career ladder” for leadership.

Some bits felt adequately original and though provoking, so despite the insufferably patronizing writing style I did not regret the time spent on this. However, I figured afterwards that most of it is redundant with other items on my reading list, so with a better reading order I could have skipped it at no loss.

In other words, I read this so you don’t have to. Don’t.

The Psychology of Money - Morgan Housel
This was a much better reading experience. Although now little of it feels new to me, I so wish I had read this ten or fifteen years ago. It would have allowed me to overcome serious emotional blocks I had w.r.t finances, much earlier.
Ego is the Enemy - Ryan Holiday

This is a book I could not have read earlier, as I was simply not ready for it until very recently. It also brought back the excitement I experienced while reading last month.

It is a challenging read, as in, it makes me emotionally uncomfortable in each chapter. This is the good kind of discomfort: the one where I have to deconstruct a few positions that previously held secure to accept what the author is trying to teach me. I will likely want to share this book actively with friends, although I am not sure yet how to place it in conversations.


Meanwhile, as usual, much interstitial learning occurred.


Reading Tanner Greer’s How I taught the Iliad to Chinese teenagers was an experience that was just all around enjoyable. This book-formatted-as-a-single-web-page has three layers: on its face, it explains the author’s curriculum for a workshop introducing US American “college culture” to Chinese students. At that level, it is interesting to learn about the expectations and behaviors of Chinese students. The second level of reading is a solid introduction to the main themes of the Iliad. Although I had read the Iliad before, most of these themes were new to me and Tanner made me want to rediscover them through Homer again. The third layer was Tanner’s view of the philosophical foundations of our society. I may not buy his view (yet), but I admire the elegance and succinctness of his presentation.


In Work is Work, Coda Hales looks at organizations through the lens of fundamental principles in system design. For example, Amdahl’s law for computer programming can be used to develop an intuition that the best management techniques and largest hiring budgets can never accelerate a team further than the speed allowed by the length of their longest sequential team process. Other system design principles are applied as well and the result is an general toolbox to detect or predict organizational challenges. I will remember this.

In Do Your Employees Feel Respected? (Harvard Business Review, archive), Kristie Roger delineates a (her) definition of respect with two components: owed and earned. This decomposition enables her to identify multiple organizational pitfalls and possible strategies to avoid them. I found this actionable.

In Faults, Errors and Failures, the pseudonymous author helped me refine my understanding of error handling in programs. Although I had already developed most of these intuitions through experience, this read taught me a few new good words to teach them better. This connects well to another article I had read a few weeks prior from Ryan Fleury, The Easiest Way To Handle Errors Is To Not Have Them, which taught me (or reminded me of, perhaps) a useful programming pattern.

Dan Luu’s Diseconomies of scale is an exhaustively supported argument that large organizations are irreconcilably worse at doing certain complicated things than small organizations. This goes again the folk belief that any complex problem becomes easier/cheaper to solve “at scale”. The article is very good and hard to summarize properly, so if you are generally interested in the topic of perverse incentives you should just read it.

Also, tangentially related to my overlapping stories from above: Why it’s impossible to agree on what’s allowed also by Dan Luu.


In I worry our Copilot is leaving some passengers behind, Josh Collinsworth convincingly argues that LLM-based assistants will make programmers un-learn how to make their software accessible, because the models were not trained for it. After all, it takes careful governance to act in favor of disenfranchised minorities, and few of the participants in this latest industry frenzy have incentives to care about them. I agree with Josh and it makes me sad.

This reminded me of how AI face recognition struggles at recognizing Black people, also because of incomplete training and how Google blocked searches on “gorilla” for not knowing how to direct their search algorithm properly.

I have mentioned earlier why I do not believe that “AGI” will be discovered or built any time soon; perhaps the better statement is that I am pessimistic we can simulate anything better than the most deeply flawed versions of our collective intelligence.


Today’s last (but not least) reportable learning is an intuition that I will find very hard to teach.

Segue—it is not often that experience brings me learnings that I am unable to decompose into convincing teaching materials! My last time was as far as December, when I discovered how a product can emotionally appeal to one’s sense of self-identity in a way that defies language. I still do not like when learnings are not easily teachable. When I was younger, I would even outright refuse these learnings. Like the only good software is the one that’s open, I used to operate under the principle that the only good learning is the one that’s shareable. I guess I have become more accepting.

Anyway, here are two thoughts, and they are saying the same thing:

In essence, the intuition is that cost functions in the real world have fractal dimensions!

When this dimension is “small enough”, refining the measurement device (say, a literal measurement stick, or the granularity of tickets in an issue tracker) merely increases the precision of cost estimates, through an asymptotically converging process.

However, there are things where the fractal dimension is larger. In that case, refining the measurements yields an exploding cost estimate. The more refinement work one does, the more effort seems to be required to achieve a result.

I believe that successfully identifying the type of fractal dimension is a valuable skill. One I will want to learn.

I also came to learning this intuition by looking hard at my home renovation project.

Through last November, my lack of trust was driving me to ask my contractor to help me close some project specification gaps. After each clarification request, a more detailed project analysis was produced, with each successive cost estimate going up non-asymptotically. This was an expensive lesson. My financial fears being stronger than my lack of trust, I then stopped asking questions. Since then, the staff in charge of doing the work has independently addressed numerous gaps I had identified prior, plus many more I hadn’t seen, using creative (and unplanned) solutions. Moreover, these solutions will not be charged to my budget.

In other words, trusts saves money?