2025 Q4: 29.04%, maximizing the value of a life well lived, the end of humans in investing and engineering

Disclaimer: By reading this article, you acknowledge and agree to the terms and conditions.

Year	Return
2024	46.6%
2025	29.04%

2025 resulted in a return of 29.04%. A good return. An index beating return. It was the most challenging year of my life to date. I don’t want to talk about it. That said, as a new Asian father, it’s my duty to be perpetually disappointed and overbearing. I look back at the mistakes made and know I could have done much better. I decided not to start a fund and that gives me the freedom to talk about whatever I want to talk about. From now on, these letters will be addressed to my kids, but maybe other people might find value in them.

“If you don’t read books, you live a lifetime. If you read books, you live 1,000 lifetimes.” - Tobi Lutke, CEO/co-founder of Shopify

Maximizing the value of a life well lived

Despite having spent a fair amount of time reading books and “lived 1,000 lifetimes” it wasn’t obvious to me that in my 30s, I would see a lot more death (and life for that matter). One implicit assumption I made growing up was that many people I grew up with would be around forever. But of course, death is a certainty and at some point, the time you took for granted suddenly becomes scarce. My son turned 1 in October and with so much going on, I wanted to take stock of whether there were changes to be made.

There are many lenses people view life through. As an engineer, I used to view life as a combinatorial optimization problem. That is, a problem where I needed to be smart and make a lot of money so I’d get to meet Janet, my wife. The rationale was, most Asian value systems care a lot about being smart and money. When we married, I had a bit of an existential problem. I solved my life’s mission! I met the perfect wife. What do I do with my life now? So far, the answer has been to do more of the same: try and make money, provide for my family, because that’s what society cares about.

In investing, the discounted cash flow (DCF) is a tool used to evaluate the value of things that generate cash flow. Certainly, that’s one way to view the value of a life. People earn a salary. Invest it and so on. It’s the way society views much of life, especially in cities. But at the end of the day capital is spent by people. People are analog machines. We can’t possibly view life entirely through the lens of capital. The divorced billionaire doesn’t feel several orders of magnitude happier than the common man with a wildly happy family. At the end of the day, people spend money to feel a certain way. When I attempt to reduce the dimensionality of human emotion, I get just two emotions: pain and pleasure. Aside from the spiritual realm, that’s all there is. We live our lives, we spend money, we marry our spouses, we raise our kids, we eat good food, etc. to avoid pain or attain pleasure.

We joined Janet’s family for a vacation in December. Among the destinations were Shanghai (China) and Busan (Korea). I hadn’t been to Shanghai in many years so I was excited to see how Shanghai had changed. Shanghai, and the broader mainland, is glorious. When I woke up, still bleary eyed, I’d open Meituan, order coffee and take a shower. By the time I’d be done with my shower, a robot with my coffee from a coffee shop down the road would be at my door. I’d just open my room door and BAM! Warmth, alertness, and joy in the form of $2 USD coffee delivered end to end.

“Any sufficiently advanced technology is indistinguishable from magic.” - Arthur Clarke

Both the US and China have been countries embarked on programs of mass magic: scores of scientists and engineers creating the future. Harry Potter might have had his owl that delivered things but Ninebot produces countless delivery robots. However, despite the incredible strides in technology, the quality of life, I noticed a counter intuitive narrative. Virtually nobody young smiled or looked happy. This wasn’t just something I found in Shanghai. The exact same phenomenon existed in Korea. This phenomenon was generational. Everyone who was old enough to have, I assume, seen hardship looked a whole lot happier.

In 1925, the average life expectancy in China was 32 years. In Korea, it was 30. 1925 was just a few generations ago, a recent memory. How times have changed! One can just imagine the tremendous changes seen in everyday life from year to year. As I see the progress of my son transforming into a cute but potato-like spawn into a full adult who can walk, talk, and discover, I can’t help but feel there are some parallels here. When my son turned 3 months old, the huge accomplishment was that he could sit up. Today, the huge accomplishment will be my son being able to form sentences. He still has a childhood filled with milestones and optimism. As a father, I can’t tell you how excited I am to see my son grow up. In 1950s Korea, the huge accomplishment was the end of decades of war and a shot at doubling life expectancy. Today, in Korea, no such increase in quality of life likely exists. It has already reached some level of saturation with respect to progress and this is presumably true of all developed countries. It’s likely someone from 2 generations ago would think the current generation is out of their minds. How can you not be happy, want to start families, feel the country has made it? However, when your childhood was amazing, you grew up seeing progress everyday, and expectations of living the educated white collared high flying lifestyle were the norm, how could you not be unhappy?

Cities are many things but among them are the best mankind has to offer. The best food. The greatest amenities. Goods/services from all over the world. As there is a scarcity of desirable locations, means of production, etc. of course, there is a competition of capital. Not everyone can afford everything. At the risk of sounding banal: nothing makes this more painfully obvious than technology. Meituan, Uber Eats, Facebook, real estate websites, Instagram, etc. makes the desirable sources of aspiration, hunger, and unhappiness. What do people do when they don’t feel they have enough? Among other things, they will probably stop having kids. Living in a city where real estate, education, attention, time, etc. is scarce, implies a much greater amount of effort required to have kids. Nothing wrong with it per se and obviously, the topic is more nuanced than that. For example, women's rights have enabled women to aspire to much more than homemakers. But of the factors that are involved, I think the perceived scarcity of modern living is the primary cause of both unhappiness and a desire to not have kids. This implies the opposite, the attainment of happiness, is the perceived attainment of abundance. The fact that you can say “I have enough”.

Objectively, we live in incredible times. If you were lucky enough to be born into a developed country, you get to experience what was once magic. As these programs of mass magic progress, our quality of life will likely improve. Life expectancy has never been higher. But despite being in amazing times, it’s easy to feel that it’s never enough. Perception is what matters. Going back to the lens of optimization, the function I believe we want to maximize is the integral of perceived abundance over time. That is, I think maximizing the value of a well lived life means to create the perception of abundance of money, time, love, happiness, etc., for as many people, for as much time as possible, especially one’s family.

The end of humans in investing and engineering

A bit flipped in my head this quarter. I used to think engineers would forever write code. I don’t think so anymore. It seems obvious that engineers will express logic in natural language. That’s not such a novel insight these days. If anything, I am behind the times. There are a lot of applications, however, that AI/LLMs are sorely lacking. The code I get for backend systems is still quite shoddy and I had an LLM attempt to delete a production database. Common sense isn’t common, especially for LLMs. And yet, these systems are as bad as they’re ever going to get. Frontend engineering seems to be an area where LLMs are already competent.

When we look into the future, I see a timeline that looks a lot like self driving. These systems are very good at some subset of tasks. But the bar for completely handing over the effort to AI is high. With self-driving, the number of corner cases is high as are the stakes. Someone can die if AI fails to be a competent self-driver. Thus, the timeline for competent self-driving is significantly longer than most people thought more than a decade ago. The same is true of engineering. In some cases, people can die as a result of poor engineering. Security breaches can mean the end of a business. And so on. In my view, the full automation of software engineering will take decades if ever.

I think investing is an easier case for full automation via AI than engineering. AI is very good at having an encyclopedic knowledge of topics and bad at logic. In engineering, both are required, but logic is more important. Code, as with math, is a raw expression of logic. As I stated previously, I still get very bad code and AI does some really silly stuff. In investing, the opposite is true. Simple investment theses tend to outperform contrived/complicated investment theses. Previously, an argument for human investing is that humans are great at the qualitative where machines are not. With LLMs, I am not sure this is true anymore. Outperformance in investing has historically tended to be a result of knowledge/research, behavioral/emotional, quality of logic, or timespan. Well, LLMs can have encyclopedic knowledge and can ingest the entirety of the written word. So it’s unlikely the edge going forward from humans vs. machine is going to be a result of research.

In the past, it used to be humans vs. humans. Having an emotional edge over other humans used to be valuable. However, as someone who has had the hardest year of my life, I can tell you, my emotional state this year has definitely resulted in mistakes. A machine has no emotion, but can understand emotion. In my undergrad degree, I worked on speech emotion algorithms where algorithms would attempt to understand how humans felt based on the spoken word. Back then, there were algorithms that understood emotion based on what words were used. There were also algorithms that understood emotion based on the tonality/volume/etc. of the voice. Combining the two (novel at the time), I was able to achieve accuracy levels over 90%. When I invest, sometimes I imagine I am going to war. The opposition is a machine that feels nothing, understands human emotion, does not require sleep and is not affected by the overstimulation of a child needing love/care. I would tend to believe the machine has an emotional edge over humans.

Another edge in investing tends to be one of logic and here, it’s clear that humans still have an edge. However, the structure of public equity markets have dramatically changed from years past. In the past, investing geniuses like David Einhorn would notice something contrived like a hidden asset living on the balance sheet that management would unlock. Today, even he would admit: “nobody is paying attention”. So even if complicated logic results in the understanding of hidden value, as evidenced by value investors over the past decade, it may not matter. You may not get credit for the discovery. Anecdotally, I find simple, obvious thesis’ tend to win. In addition, when reading writeups by the world’s most “sophisticated” investors, it occurs to me that in the areas I am competent in (mostly software) many of their thesis’ are predicated on faulty assumptions. It’s possible the bar for logic in investing may not be so high so as to be surpassed in a short period of time.

Finally, the last edge investors have is one of time which is, perhaps, the strongest argument for human investing. The value that markets give companies credit for differ wildly based on timespan. As an extreme example, whaling used to be a business! Today we have fossil fuels. Thus far, it seems most investors (algorithmic investors included) are concerned with the short term and that gives long term investors an edge. However, AI can come up with thousands of thesis’ years ahead of time, be patient, track all of those theses and when a catalyst exists, capitalize within fractions of a second. So perhaps the edge of being patient will also be dominated by machines.

So what am I doing about it personally? I’ve written tools to automate parts of the investment process thus far with promising results. My intention, this year, is to experiment with automating the entire investment process with an emphasis on long term investing. That is, I am attempting to imagine Warren Buffett in algorithmic form if he could analyze millions of data points. I’m a few days into writing the infrastructure. With modern tooling, it has taken me just 3 days to download the data history of the US stock market and write a backtesting platform from scratch.