Eye Laser #
Summer days in Europe are long and hot. Sunglasses become necessary both subjectively and objectively. That’s what triggered me to get laser eye surgery.
Some decisions are just decisions, while others are DECISIONS. Choosing which laser eye surgery techniques was clearly the latter.
Making decisions is more about the arguments, statistics, and process, not just the conclusion. It’s like when the Fed Chair waves his magic wand to orchestrate the global economy: there are only two directions, raise or lower interest rates.
You could reach a conclusion through simple reasoning: inflation is too high, so let’s raise rates, or vice versa. Any five-year-old could do that. Or you could gather the most brilliant minds of our generation to conduct extensive research and run complex simulations to arrive at the same conclusion. The answer might be identical, but it carries different weight.
For laser eye surgery, I let Grok4, ChatGPT o1-pro, Gemini 2.5 Pro, and Claude Opus 4 (SOTA models then, web versions since I am lazy :P ) conduct deep research on the overall landscape of laser eye surgery techniques, history, pros/cons. I learned that in Taiwan, the mainstream includes three different technique systems: TransPRK, LASIK, and SMILE Pro.
Two questions to answer: 1) Which technique to select? 2) Which machine to trust, specifically which surgeon to execute?
In simple terms:
- The cornea consists of five layers from outside to inside: Epithelium, Bowman’s layer, Stroma, Descemet’s membrane, and Endothelium. All procedures perform vision correction on the Stroma layer.
- TransPRK directly removes large areas of the Epithelium and Bowman’s layer, then reshapes the Stroma.Theoretically needs the least Stroma destruction, though recovery takes the longest since you wait for Epithelium regrowth.
- LASIK lifts a flap of Epithelium and Bowman’s layer, processes the Stroma, then replaces the flap. Fast recovery since Epithelium doesn’t need regrowth, but the lifted portion now has an incision line.
- SMILE Pro first creates the portion to extract (lenticule) directly in the Stroma, makes a small opening in Epithelium and Bowman’s layer (relative to other techniques), then extracts the lenticule. Fast recovery since no Epithelium regrowth needed. Though it requires more Stroma removal, academia generally considers the post-op corneal biomechanical stability higher than LASIK, closer to TransPRK, since the upper cornea isn’t cut 270 degrees like LASIK. Main risk: lenticule extraction is manual, possibly causing lenticule tear, fragment retention, or incision tear.
Later, I visited three different eye clinics in central Taiwan to gather actual data about my eyes and cross-verified the information I’d gathered with doctors. My corneal thickness and myopia/astigmatism conditions made me eligible for all three procedures. Good: I had choices. Bad: I had to choose.
I eliminated LASIK first due to its relatively weaker post-op biomechanical structure.
Then, the core question between TransPRK and SMILE was whether both procedures already exceeded certain safety margins. How much safer is “theoretically safer”? Can it be quantified? Which studies support these claims? Are the sample sizes large enough? Is a difference of three decimal places statistically significant? Is “theoretically safer” just marketing speak?
Given my specific conditions, TransPRK had more drawbacks without significant advantages, so I chose SMILE Pro after consideration. (I subjectively perceived TransPRK’s so-called full automation avoiding surgeon operation risk as risk transfer to the optometrist, since input information must be precise.)
For the specific machine model and surgeon part, it seems more like an art than a science. I could only deliberately play devil’s advocate to feel out the doctor’s vibe and understand their backup plans for unlikely surgical failures, trying to avoid overconfidence bias.
LLMs #
In the meta view, there are two shifts for me as LLMs have matured rapidly in the past few years.
Shortening the gap from intention to execution: Depending on the level of impact, we need different levels of information, knowledge, and insight to arrive at different confidence levels to make the move. LLMs can exponentially lower the cost of gathering information and facilitate the information-to-knowledge step, hence shortening the gap from initial intention to actual execution. For example, eye laser surgery was previously so complex I might never have bothered to act, but now I could decide within two weeks.
I do think we live in a world composed of probabilities. It’s all about making educated bets. Embracing uncertainty is mentally uncomfortable. It depends on the degree of realness you want to approach, and that realness often comes at the price of endeavor and constant questioning.
If you agree that life is composed of numerous decisions, and since LLMs can accelerate and improve the quality of the decision-making process, then by definition, LLMs can change the speed and direction of our life trajectory.
Shift of default attitude: I’ve found my default mode toward new things/challenges has silently changed.
In the pre-LLM era, I would seek out the most authoritative friends, or friends of friends (you know, six degrees of separation theory), for answers and advice. In the post-LLM era, with proper use of AI tools, abundant curiosity, and enough patience, guided by first principles and humble instincts, my default mode has become: “How hard could it be?” (This mindset shift was influenced by my good friend Thomas, who always seems to have the courage to face hard questions.)
For the whole AGI/Singularity/Effective Accelerationism arguments, I think those two hidden shifts are the key for the carbon-based life camp.
Investment #
The key to investment is risk management.
Much like my experience with eye laser surgery, there’s rarely a universally optimal answer for most cases. If there were, alternative methods wouldn’t need to exist. Instead, every decision represents a specific risk-return profile, and we can only act based on our particular conditions and risk tolerance.
“High risk, high reward” has become a cliché excuse that many people use to avoid thinking, especially since many decisions aren’t even close to the efficient frontier. Yes, we can all have different risk targets/appetite, but the premise is that we’re all as efficient as possible in the return-risk space.
More importantly, there’s a fundamental difference between being an observer and being a participant. When you put money on the line, you experience fear, hesitation, and doubt. These emotions exist for participants but are almost absent for observers. Without these variables in play, you can hardly claim to be a true participant.
This aligns with my eye laser surgery experience: science and rationality can only take us so far. When you have significant skin in the game, ultimately, you still need that last leap of faith. IYKYK. (Of course, like every other decision, not participating always remains within your option universe. Exit the game if it’s not one you can play comfortably.)