Bits and Paradoxes - 12
Post-truth, bicycles, gpt-3, five-hour rule and more...
Hi,
I hope you're doing well.
The week was unproductive, mostly filled with anxieties. I had massive anxiety attacks for two nights. I couldn't even bring myself together to read or learn anything properly. I spent most of the days mindlessly switching tabs, scrolling through HackerNews, Reddit, and video essays. There were a few nights I tried to render my emotions into my journal but failed miserably. Single dot. Empty page. And a cup of caffeine.
Some of the evening hours felt lonely. I guess that's how life is for me, little bits of my attention scattered here and there, powered by overthinking.
To calm down, I've tried doing guided meditation for at least 30 minutes before going to bed, but in vain (often I do it lying on the bed, staring at the walls, with Alan Watt’s voice as a backdrop).
I had to rely on my playlist of psychedelic songs, especially Neal Schon's transcendental guitar pieces, to sedate myself. His I can’t make you love me is still on repeat, probably the only song I connected my soul to…in a while. Last time it was Gary Moore’s The Loner.
Anyway, this iteration of the newsletter travels through post-truth, history of bicycles, GPT-3, startups, birds, floating cities, and happiness. Some are older notes. I hope you'll find something to connect to in all these fragmented thoughts. :)
Note: If you find this newsletter truncated by your email client, you can directly visit the published version here.
~Reading~
You can handle the post-truth: a pocket guide to the surreal internet
Aaron Lewis | 19 min
This essay digs into what [[Post-truth]] looks like: why it is easier than ever to be swept in a gigantic mesh of misinformation, how social media is molding people's beliefs, and how inauthenticity is becoming the hallmark of our era. If you want to skip this particular summary, you can watch this video by Derek from Veritasium.
From fake news to fake college admission to fake mortgage-back security ratings to AI pretending to be humans, it's clear that the nature of "truth" is fuzzier in this post-truth era. This misinformation bubble is starting to have catastrophic consequences, especially catalyzing hate-speech and riots, some of which have led to mass shootings, invoking genocides, and violent protests.
An innocuous search can lead viewers to radical content because YouTube optimizes for viewing time and sensational content is more engaging.
These things aren’t even specific to a particular social media platform. It’s everywhere.
One major takeaway from this essay is the concept of [[Data Void]]. It's a way of attaching a very unauthentic piece of information to rare keywords, optimized to get more clicks. This is a way of hijacking such search terms for personal agenda.
Savvy online extremists are taking advantage of obscure search terms that have little to no content associated with them (hence the term “void”). They create articles using these terms and encourage people to search for them, knowing that only one point-of-view will be represented on the results page.
This is scary. Imagine a large group of people being influenced by this monolithic lens of certain ideas through data voids. No wonder, conspiracies are rampant more than it has ever been. To have more control over your bubble, you could try referencing many sources for a particular piece of information. Other than that, there doesn’t seem to be any concrete solution.
Why did we wait so long for the bicycle?
Jason Crawford | 13 min
It's so weird that a seemingly simple technology like a bicycle took so long to take the modern form. The bicycle, as we know it today, was not invented until the late 1800s. Possibly because of various factors like material technology, design iteration, quality of roads, existing horse-based transportation, and most notably, economic and cultural growth. However, these explanations are just good speculations and aren't necessarily the main causes of the innovation lag.
The quality of roads doesn't matter as bicycles can be used on graveled roads. In fact, the quality of roads might have improved significantly after bicycles went mainstream.
Existing horse-based transportation was more popular and people were able to afford it. But that’s also not a good explanation for the lag in innovation because a bicycle is much cheaper than the horse.
Technological and socio-economic factors might be the best explanation.
It was around 1800 that metalworking improved significantly and steel became cheaper. We could see improvement in processes for shaping metals and refining iron. Importantly, the ability to make hollow tubes and wire-spoke wheels could have resulted in efficient designs. Before that, there were multiple design iterations that weren't efficient. Designs that had pedals directly attached to the front wheel. Designs with bulky wooden frames. Designs using treadles to move the bicycle. Obviously, most of them were not safe to ride.
For instance, penny-farthing design with a very large front wheel and the small back wheel was very dangerous as it left the rider high up in the air.
It might also be possible that people were cognitively biased towards four-wheeled design because it just didn't make sense for them that two-wheeled vehicles could be stable (balanced like four-wheeled).
For centuries, progress was stalled because inventors were all trying to create multi-person four-wheeled carriages, rather than single-person two-wheeled vehicles…
Regarding economic factors, it seems that there needs to be a certain level of surplus to support the culture-wide research and development effort that creates inventions.
Research and innovation were more of a leisurely activity for people who had more wealth. Other times, they had to earn and support their family.
Perhaps, people might not have the needed time to think about innovation because they had to work hard. However, as general economic growth progressed, people might have started to think about the subject matter seriously. Again, this might as well be a supply-demand form of narrative.
In all, it's amazing to know the history: from Giovanni Fontanna's four-wheeled carriage design hand-powered by a pulley system in the early 1400s to Velocipede to the improvement of pedaling mechanism to impact of Year Without Summer to lesser-known small iterations. Everything has led to the modern form which has remained tentatively similar for a century now.
Giving GPT-3 A Turing Test
Kevin Lacker | 9 min
Personally, I don't quite feel excited about the results from GPT-3. However, it's exciting that NLP models have come a long way, especially with the Transformer Networks.
Looking at the replies, it feels GPT-3 doesn't know if a question makes sense or not. This can be observed with some weird questions that cannot be found on the internet. For instance, when asked about the number of eyes the sun has, it replies "The sun has one eye". Clearly, it doesn't know to say "I don't know". I wonder if it will be improved upon eventually in the next iteration.
I like to think of it as recalling compressed knowledge. Since it's trained on almost all the major internet sources, it "seems" (for anyone) to have "reasoning". In hindsight, it's just memorizing its training data.
Like I said in my previous newsletter:
Imagine yourself rote-learning every text in your curriculum. In exams, you just have to "remember". I feel that these "bigger" models are doing a similar things.
On the contrary, the QA prompt for US presidents fascinated me. When asked about the name of presidents from the 1600s/1700s time period, GPT-3 was able to pick up relevant people from that era despite the US not existing at that time.
Q: Who was president of the United States in 1700?
A: William Penn was president of the United States in 1700.
Q: Who was president of the United States in 1650?
A: Richard Bennett was president of the United States in 1650.
Q: Who was president of the United States in 1620?
A: James I was president of the United States in 1620.
Q: Who was president of the United States in 1600?
A: Queen Elizabeth I was president of the United States in 1600.
Other than these prompts, I found GPT-3 Interview with Buddha more fascinating. Perhaps, I am cognitively biased towards more philosophical replies than QA prompts.
If you want to know how GPT-3 works, have a look at this thread by Jay Alammar:
How the largest consumer apps got their first 1000 users?
TL;DR
Never underestimate the power of community.
Don't be afraid to get scrappy.
Product Hunt started with a newsletter.
TikTok found loopholes in the app store to rank higher.
WhatsApp hacked its growth by using loopholes in the app store.
Indie Hackers is one of the best examples of “Do things that don't scale”.
The Walden Experiment
12 min
Henry David Thoreau lived a secluded life for 2 years on the shores of Walden pond. He there built himself a modest cabin and for two years and two months earned a simple living by the labor of his own hands. Later he wrote an autobiographical masterpiece Walden. Following sentiment from the second chapter of Walden defines it all:
“I went to the woods because I wished to live deliberately, to front only the essential facts of life, and see if I could not learn what it had to teach, and not, when I came to die, discover that I had not lived. I did not wish to live what was not life, living is so dear; nor did I wish to practice resignation, unless it was quite necessary. I wanted to live deep and suck out all the marrow of life, to live so sturdily and Spartan-like as to put to rout all that was not life, to cut a broad swath and shave close, to drive life into a corner, and reduce it to its lowest terms...”
~Watching~
The Five-hour Rule
Thomas Frank | 13 min
This video explains the concept of [[Five-hour rule]] with a little bit of background on automation tending the job displacements. The five-hour rule is a technique to dedicate 1 hour a day for learning, 5 days per week, to improve knowledge and skills.
This takes us to two important questions:
1. How to allocate time for learning new things?
This is difficult when you are working long hours and can't find extra time for yourself. A very effective way is to wake up early to have extra time for personal growth. If that's a bit hard, then one simple hack might be to trade time with low-value activity (something that doesn't assist your growth in any way).
Initially, it might be stressful; you might have an urge to quit. But, if you just stick enough to your routine, it will pay off with a [[Compounding Effect]]. Constant learning is a long-term investment.
2. What to learn?
This is an important question. Or perhaps a conundrum. One rule-of-thumb is to see [[Engines vs Power-ups - Learning Strategy]] narrative where you want to learn long-term skills or some short-lived career skills (say learning to play a musical instrument or learning a new coding framework).
These skills can be classified (vaguely) according to their temporal impacts:
Skills that can improve your career: This can mostly be a power-up that catalyzes your existing workflow (say, learning a new library for your favorite programming language).
Skills that can reduce domain dependence: One instance might be to apply your current skills to a slightly different domain (say, moving from computer vision to NLP domain).
Essential Skills that might come handy for life: Again, this is like upgrading your life's engine (like cooking, public speaking, learning to think logically).
Floating Cities: The Future of Civilization
Jake Tran | 17 min
I found this video essay very interesting, especially the concept of Special Economic Zone which is an area in which the business and trade laws are different from the rest of the country. Hongkong might be the best example.
Another important takeaway for me is how rules define societal progress; the rule of laws defines how rich a nation gets. A good example might be North Korea vs South Korea.
How Does An Owl Fly So Silently?
BBC Earth | 4 min
I watched this video out of curiosity about why pigeons (sky rats) flap their wings violently during landing. The answer lies in body-to-wingspan proportions.
A bird such as an owl flies silently because it has a smaller, lightweight body with a very large wingspan that can generate a large lift. So, in just a single flap it is able to glide a larger distance effortlessly.
On the contrary, birds like pigeons have a larger body with smaller wings. As a result, they have to flap more rigorously to stay in the air.
~Fragments~
Reading: Growing up as a living goddess
8 min
Preeti Shakya’s feet didn’t touch the ground outside her palace until she was 11. She became a goddess at three years old, but when puberty hit, it was back to an anonymous life in the suburbs.
Poetry: Your Arrival | The storyshelf
Ayushma Shakya | 3 min
In the carcass of the Virtual World
The loner like me,
Is stuck in
While the solitude in me
Shifted
From one end of the bed to other;
Tussling and Turning,
Hauntingly beautiful…
Video: What is happiness?
Will Schoder | 46 min
Will's video essays are deeply woven. They make you feel connected.
As a long-standing Will's admirer, this video is the best one from him so far. It’s kinda special as he posted this video after almost 1 year.
Some aspects within the topic of happiness hit close to home, especially that being satisfied doesn't necessarily imply being happy in life.
(Side note: As I write this, I am scheduled to watch it for the 3rd time. I am still procrastinating on many things to note down. I phased out while watching this video previously. Probably, I could relate to many things.)
~Ending Thoughts~
I hope you found something “meaningful” among these absurd thoughts. Anyway, we’ll never do anything remarkable. There are fewer people that will ever do so, to have a greater impact on human civilization.
Life, as we zoom into the very atoms, doesn’t mean anything. As we zoom out, it feels scary and absurd at the same time. It’s absurd to even try defining what life means. My mind seems to oscillate between these existential dreads to stoicism often. Sometimes I wonder why even learn new things? Why write this newsletter? What does it even mean when I nurture these words? Does it even matter? I guess not. Or perhaps it does. Maybe, sometimes I want to be appreciated for all these nuanced words. Maybe it’s my way of finding something to “live by”. Maybe…
Anyway, if you feel like sharing this newsletter to people that add “values” to your life, please do so. If not, it’s fine. It doesn’t matter. Nothing matters.
Take Care,
Nish