reflection

August Update

This space is intentionally left blank.

❦❦❦

bouquet in vase

This space is intentionally left blank.

sunset over the clouds

This space is intentionally left blank.

sunflower in bloom

The sunflower bloomed.

❦❦❦

Although I didn’t choose my day of travel, I felt lucky that the rear-view mirror had automatic dimming so I could see the eclipse, partly.

❦❦❦

Technical stuff that made me raise an eyebrow: Bug Blindness by Dan Luu. In short: some people (like me) look at technology and can see all the way it’s incorrect, wrong or mistaken. Other people just accept the technology as a natural phenomenon and limit their judgment to what they learn through empirical discovery.

The piece is a tad longer than it needs to be to make its point. Also, it falls short of making another point that I care about more: quality in technology may not be necessary for it to be useful, but living with bad technology is like living with broken windows: it’s cold in winter, and it invites more crime in the neighborhood. We fix windows and install triple pane glass not just “to fix problems”; we do it because it objectively enhances our quality of living and, indirectly, our collective wealth.

❦❦❦

More technical stuff: RAG Is Simpler Than You Think by Rafael Pierre. The essence here is “if you want your users to be able to search for stuff, you don’t have to deploy a LLM; there are better/simpler solutions” with a list of six true and tried “recipes” (approaches / algorithms), their cost and a decision tree.

It’s good; I knew about half of them and felt happier by learning about the other three.

❦❦❦

Also, some guy called Thomas Dullien shared Three important steps in [his] maturation process. Good for him, I guess?

There’s approximately just two small nuggets of value in there:

  • for people who consider themselves smart, it’s valuable to understand the limits of one’s own reasoning abilities. Our brain is good at retro-fitting a reasonable story to justify decision-making that’s mostly feeling-based and based on partial / wrong input and partial / wrong learned patterns.
  • people who have received a mostly technical / scientific education and then stopped educating themselves further often mistakenly believe in determinism and it’s just dumb.

I barely salvaged these two points from a text that’s otherwise mediocre. I felt it was making me dumber as I was reading.

More seriously, I’m really annoyed / disappointed at this trend of authors who want to share / teach something and feel compelled to phrase it as “I learned this, it’s my personal experience, now do with this as you wish.” I get it, they do it because this way they can’t be criticized much — they can always say “hey, it’s just my personal experience, if it doesn’t work for you it’s not my fault.” But that’s really cowardly. If someone really thinks it’s useful for an audience to learn something, then they should own it and make an effort at teaching it properly. That implies picking an audience, starting where they are, then doing incremental reveal to develop your point, etc.

❦❦❦

The only bit of LLM news I felt worth remembering: AI is removing the middle class of software engineering by Florian Herrengt. This one is a refreshing take on “LLM will take all the jobs”: instead of arguing that all junior people are doomed, it argues that LLM use will help all the mediocre folk in the middle make more mistakes and make other people notice their mediocrity and start expecting more. So only those who are more serious about doing the job well will remain. In a nutshell, LLMs will chase away the posers I guess? 🤷

❦❦❦

The only article that really made me think and even reconsider my views was Incentives are for losers, by Adam Mastroianni.

At its heart, this essay is a plea to de-rank incentives as the structure we build as a society to make people do (and not do) stuff, and instead re-emphasize personal values and morals. The author focuses here on the individual perspective: with the question “why am I doing this”, going from “because the system / religion told me so” to “because I believe it’s the right thing to do”. He does not reject the idea that social constructs might condition how someone’s personal values develop (in fact, he recommends we inspect and renew incentives over time); instead, he advocates for taking responsibility and not blaming “bad incentives”:

Bad incentives have become our generation’s version of “I was just following orders”, an excuse that should be an indictment.

The reason why this text gives me pause is that it is difficult to unify with another belief I currently hold: that our society has become complex, and that people can’t reason from first principles about what they should do every moment and why. This is a belief that I’m working to make actionable — how to choose a structure to lean on, to ease making decisions when life is tough or busy.

Maybe the unification here would be to insist that the structure is external, but needs to emerge and crystallize from personal values as basic ingredients.

❦❦❦

Next, we have a bit of very good brain food:

In “These hobbies will save you from infinite scrolling”, TJ Power teaches us that we can de-tox our brain from fast dopamine addiction using activities that engage the body steadily and are moderately engaging for the brain (i.e. more than not at all, unlike meditation, but less than fully, so unlike high performance sports).

He offers nature hobbies (e.g. hikes in forests / walks on beaches), people hobbies (e.g. board games, time with family), mood hobbies (e.g. re-watching movies, listening to music we already like) and learn hobbies (e.g. woodworking, ceramics).

Good stuff, that’s quite close to the common recommendations we make in that online community where I’m volunteering.

❦❦❦

This contrasts with this terrifying bit of brain food: The brain may be about to have its Ozempic moment, reported by The Economist (archive link). This reports on the current clinical trials of a molecule that reduces our brain’s need for sleep.

On the one hand, this is good for people with narcolepsy, which is what the article focuses on. Yay.

On the other hand… An actual product could bring about the biggest class war in history: people who can afford to pay to not-sleep vs. people who have to sleep. I don’t like this at all.

❦❦❦

Thankfully, I found a real juicy bit of good news to sandwich the icky stuff above and close for today: CATL Debuts World’s First Field-Validated Sodium-Ion BESS, Bringing Sodium Storage to Commercial Reality.

This is better explained in this video:

In a nutshell, CATL is bringing a sodium-based battery on the market. Sodium-based batteries are 1) better for the environment (less pollution), 2) easier on geopolitics (not limited to a few places where lithium can be mined), 3) less dangerous to operate than lithium (they don’t explode) and 4) way, way more durable (thirty years+ instead of ten). They’re also likely to become quite a bit cheaper than lithium to manufacture!

While lithium is likely to remain the main technology for stuff that needs to move around (e.g. cars and mobile phones, because lithium batteries store more energy per unit of volume & weight), sodium may be the missing link to make renewable energy work at scale.

A specific point mentioned in the video and not the original news release: thirty years of operations means this becomes infrastructure that big funds can invest in (batteries were, until now, mostly consumables from a financial perspective; with thirty years of operations there’s a real possibility to turn an investment into a return). This opens the door to way larger budgets available to fund city-scale battery storage, and less political complexity to do it.

I want to live in that future already.

❦❦❦

References:

Publication Date: