From 081d95b2ff52d88053bf0f5180208b630d4cdaa2 Mon Sep 17 00:00:00 2001 From: Preston Pan Date: Fri, 3 Jan 2025 20:22:16 -0800 Subject: add blog posts and journal posts; update emacs configuration --- blog/horses.org | 110 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ blog/index.org | 33 ++++++++++------- 2 files changed, 130 insertions(+), 13 deletions(-) create mode 100644 blog/horses.org (limited to 'blog') diff --git a/blog/horses.org b/blog/horses.org new file mode 100644 index 0000000..7c68834 --- /dev/null +++ b/blog/horses.org @@ -0,0 +1,110 @@ +#+title: Stop Asking for Better Horses +#+author: Preston Pan +#+description: It doesn't happen instantly. +#+html_head: +#+html_head: +#+html_head: +#+html_head: +#+html_head: +#+html_head: +#+html_head: +#+html_head: +#+html_head: +#+language: en +#+OPTIONS: broken-links:t + +* Times Never Change +#+name: Henry Ford (allegedly) +#+begin_quote +If I had asked people what they wanted, they would have asked for better horses -- Henry Ford (Allegedly) +#+end_quote +Before his time, cars were nothing but a novelty item for people that could afford them. +It's a simple idea, but it's what runs our world today. However, the democratization of information +that the internet has blessed upon us has some side-effects that distort truth signals created +by the market. Let's take a deep dive into the strange difference between what people /say is true/ +(i.e. "people want better horses!") versus what people /actually want/ (i.e. real demand or real +risk taking). +* You're Not Vested +You (probably) are not an investor. You have no skin in the game, and you do not lose anything +if your guess is wrong on the internet. Nobody will ultimately point out your mistake, because +you're likely not important enough that your mistakes will be talked about. And, even for people +that are, nobody will call them out on how wrong they are; people are too busy making their +own opinions! So if you're not investor, it does no good to you if you think something is a bubble. +In fact, if you /are/ an investor and have enough capital for what I'm suggesting, you should know +about: +** Information and Markets +In a market, any relevant information that's not reflected in the price is an arbitrage opportunity. +If you believe that some technology's fundamental value is zero (or low), you have a couple of +options: shorting and puts (or some put spread). Of course, the common counterargument is that +the market is irrational, so you'll just end up losing all your money if you buy puts. Well, you +always have a choice of parameters. For instance, you can buy less puts and you can increase your +amount OTM. If you have a couple million dollars, you could make a profit by buying one-week put +options every week, while adjusting your parameters OTM. During a bubble, there are only two +possibilities: either put options are not priced in, so as long as you continue to have capital +there's always some risk exposure (isolating your thesis) that could allow you to make money, or +the put options are priced in, in which case the fact that it's a bubble should be obvious from the +EMH. Therefore, there's always a way to make money off of your twitter prediction! Will you? I +believe that should be our measure of how much we should continue to believe in people on the +internet. +** Economists as Market Participants +When economists predict a recession or predict a Ponzi scheme or bubble, why don't they participate +in the market place? The government has essentially infinite capital; they could roll put options +on their own economy if they wanted to. If the government is correct about a possible recession, +then great: their bet in the market pays off, and it can use that bet to employ people and put +them back to work. If the government is incorrect, more money flows into the private sector, +strengthening the correct market actors and crippling the inefficient market actor. When economists +make predictions, remember: they don't take risks like this, the government doesn't take risks like +this in the same way market participants do. Therefore, this is my rule of thumb: /never/ trust +/anyone/ that isn't vested in the market to tell you what's happening in the market, even +/expert economists/. Economists can advise the government to short or to buy puts, basically +participating in the market in the same way that regular market actors do. In this way, there is +an inherent accountability/feedback mechanism. +* The Internet is the Better Fax Machine +People forget that the general sentiment at the time of the dot com bubble: while there were many +people who were fanatics about the new technology, there were many skeptics, which all said the same +thing: why do we need the internet when we already have [[https://www.laphamsquarterly.org/revolutions/miscellany/paul-krugmans-poor-prediction][fax machines]]? Often, they forget that +they're /not/ market participants, and there's probably some underlying /reason/ why people are +investing money at such a large volume. Of course, real bubbles exist, but it's no use in trying +to distinguish fake bubbles from real bubbles if you're /not vested/. And, often, real bubbles +are in part propped up by the government. In fact, if the government simply bought puts in the way +outlined above, and they were correct, there would /be no bubbles/. Why do we need the internet? +It's not for the economists to comment on, let alone the /average Joe/. +* Blockchain is the Better (or Worse) Bank +The same sentiment is currently being expressed by the public consciousness for blockchain +technology. First, probably 80% of these people have never heard about smart contracts or how +they work. I'm currently vested in blockchain, and I used to work in the industry. I can tell +you a couple things about blockchain, but most of what I'll tell you is that most of the public +is just /wrong/, like horribly wrong, most of the time. It's probably a common theme in most +industries, but for blockchain, where on one hand there is a huge craze, and on the other there +are people saying the technology is worth $0, it is even more so. + +Blockchain technology gives people incentives to agree on the state of some data so long as the +native token is valuable, with possibly some state machine that manipulates this database of data. +That's literally it. A lot of research into blockchain technology boils down to this, and its +uses range from supply chain tracking to finance (which I think is the most realistic use-case). +This isn't a debunk of common talking points, but what I can tell you is that for many use cases, +if you didn't have the concept of blockchain, you'd almost have to reinvent it, but worse. That +in and of itself makes the industry valuable, and given my "insider experience" (I didn't do that +much but I was very much surrounded by world class experts), I can tell you that it's not all +vaporware. And yet, despite all of these peoples' hard work, all their long hours and vision and +clever insight, people have the audacity to say that the technology is equivalent to "a worse bank". +* AI is the Better Autocomplete +Now the new kid on the block is AI. Of course, people are already saying that this technology should +be worth close to $0, or that this technology might be a net negative on society, when we haven't +even seen 3 years worth of LLMs existing. People refuse to believe that these LLMs are developing +emergent abilities, despite the [[https://cset.georgetown.edu/article/emergent-abilities-in-large-language-models-an-explainer/][clear evidence]] from LLM researchers suggesting so. People are out +here still talking about next token prediction, meanwhile chain-of-thought reasoning and +[[https://ai.meta.com/research/publications/large-concept-models-language-modeling-in-a-sentence-representation-space/][Large Concept Models]] are under way. The reason why we saw an explosion of narrative that +"AI is just a better autocomplete" is specifically because people are resistant to change despite +it being ultimately more demanded and better for them, in the same way people wanted +/better horses/. When will the public learn to stop tracing the same loop? +* Conclusion: Stop Asking for Better Horses +Silicon Valley often reinvents the wheel in the early stages of large technology. However, +clearly we see that investments into technology that on the surface are just "theoretically better +horses" end up being completely different, and often more demanded and better products than the +original. Of course there are real fuck-ups as well. See Enron and Theranos. But investors don't +want to lose money; the government, on the other hand, won't leave people alone about regulations +on the finance industry without showing us that it actually has better knowledge. And, it'd really +be nice if the general public would stop expecting better horses, and then getting disappointed +when they don't get them. + diff --git a/blog/index.org b/blog/index.org index 26e36c8..8086990 100644 --- a/blog/index.org +++ b/blog/index.org @@ -21,23 +21,30 @@ Blog Articles: #+begin_src shell :results output raw :exports both for f in *; do - if [[ "$f" == "index.org" || "$f" == "img" ]]; then - continue + if [[ "$f" == "index.org" || "$f" == "README.org" || "$f" == "emacs.el" ]]; then + continue fi - printf -- "- [[file:$f][$f]]\n" + + if [ -d "$f" ]; then + continue + fi + + name="$(grep '^#+title:' "$f" | sed 's/^#+title:[[:space:]]*//')" + printf -- "- [[file:$f][$name]]\n" done #+end_src #+RESULTS: -- [[file:automation.org][automation.org]] -- [[file:cognition.org][cognition.org]] -- [[file:crypto.org][crypto.org]] -- [[file:machine_learning.org][machine_learning.org]] -- [[file:nixos.org][nixos.org]] -- [[file:private_keys.org][private_keys.org]] -- [[file:stem.org][stem.org]] -- [[file:tech-bros.org][tech-bros.org]] -- [[file:voting.org][voting.org]] -- [[file:you_dont_matter.org][you_dont_matter.org]] +- [[file:automation.org][Automation, Hypocrisy, and Capitalism]] +- [[file:cognition.org][Cognition]] +- [[file:crypto.org][A Review of Cryptocurrency]] +- [[file:horses.org][Stop Asking for Better Horses]] +- [[file:machine_learning.org][Machine Learning is Here]] +- [[file:nixos.org][You should use NixOS]] +- [[file:private_keys.org][Passwords Are Obselete]] +- [[file:stem.org][Stem]] +- [[file:tech-bros.org][Tech Bros]] +- [[file:voting.org][Representative Voting]] +- [[file:you_dont_matter.org][You Don't Matter]] @@html: @@ -- cgit