THE CURRENT OPTIMIZATION

April 6th, 2022

What does traditional education optimize for? Traditional education has rightly come under a lot of fire and by many metrics education, at least in the West, is falling in quality and yield.

For a moment, let’s leave that question hanging, like an ominous cloud over a larger discussion of optimization in general. Good entry points to start thinking about this is to wonder what you are currently optimized for? Or what  systems in your life are being optimized? It’s a very valuable question to ask: am I optimizing for the wrong thing?

This kind of question helps protect against the sunk-cost fallacy: when we keep at something even when we should stop and switch gears, like waiting in a grocery line that isn’t moving as fast as the one next to it simply because we’ve invested several dozen seconds standing right where we are. There’s inertia in all things, not just physical things at motion or at rest, but with emotions and patterns of thought. A thought or behavior continually invoked is likely to be invoked again, it’s the law of habit, but good questions like “am I optimizing for the wrong thing?” help break this law for the better.

One principle that can be used to guide our focus of optimization is that the perfect is the enemy of the good. Optimizing for a problem that may never arise is a waste of time if we don’t do the work required to actually be awarded with that problem. An decent example is with app development. Optimizing for millions of users before having millions of users might sound very smart, but if it comes at the cost of never actually releasing an MVP (minimum viable product) then all that optimization was pointless.

Some things do benefit from thoughts of optimizing for the far future, like: I want to have as few regrets as possible when I’m very old. This sort of optimization - while invoking the far future - has a direct impact on how we respond on a daily basis to opportunities and chances we want to take.  Strangely, it’s optimizing for a very local circumstance: will I regret doing (or not doing) this current thing? 

The same sort of interplay can be repackaged and used to evaluate the situation for app development: will I regret optimizing for millions of users if I never even launch this product? Well of course. Or put another way, compare that question to this one: will I regret that I didn’t optimize for millions of users when I have millions of users? Well, maybe, but that’s also a fantastic problem to have - it’s the ultimate confirmation that something of real value has actually been produced.

So it turns out that we can have very long range goals that help us optimize for the right thing in any given moment. But it’s important to recognize that there might be correct order. Optimizing for no regrets can lead to a life of whimsy with no solid consistent commitments, and that in turn can become the regret: I should have committed to something and stuck with it! While all the while I was trying to live life without regret! Suddenly, the question of how to optimize for minimal regret gains a troubling amount of nuance: can we regret both paths? The scary truth is that the answer is ‘yes’.  Why? Because we can’t A/B test our life. We can’t see if a life of whimsy might lead to accidental fame and fortune but horrible personal relationships? And we can’t see that a consistent and boring area of focus might lead to a warm, stable, and loving family life.

The task of what to optimize for becomes fractal: Optimize for the long term, the mid term, the short term, and now. Answer all of these questions simultaneously, weight the answers, and most importantly: GO!

This is far from an easy task, which is what makes life both tragically difficult and quite fun and surprising, because our own personal algorithm for summing that equation is constantly shifting based on new information, opportunities and ideas.

Traditional education attempts (rather poorly) to cast a very wide net in terms of topics to be aware of. The focus of education is nouns. These are the things you need to know about and understand. The chief underlying problem with this approach is that life isn’t really a noun, it’s a verb: we live. Which means we need to know how to do things, and everything we do or end up doing can be optimized. A master of a craft is someone who has simply optimized their performance in that craft to an exceptionally high degree. Someone who has never committed very long to anything might not be a master in anything, but they may have optimized for fast initial learning, deriving subtle mental models that allow for quick initial optimization in any field. 

Which one does school point more toward? Well, considering grades are always a subtraction from a perfect score, school expects mastery. But it’s not even mastery in doing anything in particular, except perhaps regurgitation. Maybe writing and math translate to active verbs, but for the most part regurgitation goes a lot farther than creative exploration of these acts. School optimizes for knowing the noun as opposed to enacting the verb. What’s fascinating is that kids just don’t optimize well for this particular end. And yet they can concentrate for hours, days on end getting good at a video game. What’s the primary difference? A video game is a verb: you optimize for being able to do a particular thing, which is far more dynamic and challenging than any given topic that is best categorized as a collection of nouns. Math in school is boring, but millions of people spend hours every day playing sudoku. Why? Math classes require regurgitation with slight variation, while sudoku is a game that we can have fun optimizing for.

Life is a series of complex and dynamic problems: what best helps us solve these problems? A heap of nouns, or active skills? The answer is obvious, but hopefully it’s also intuitive, and points to potential solutions to help optimize the way we educate our children and ourselves. Will your life be a static topic?  Or is it time to change because the current thing requires a new optimization.

THE FAST APPROACH OF NOW

April 5th, 2022

Yesterday Google released news of the most advanced language model to date.  Readers may remember that a few episodes of Tinkered Thinking were produced in concert with the language model known as GPT-3, an impressive tool that shocked a part of the tech world that was paying attention. But like many new shiny gadgets, the novelty of GPT-3 quickly wore off. It’s limits became apparent very quickly - it couldn’t reason properly, it made some weird mistakes, and soon it was all but forgotten. But progress does not stop.

We have to imagine the inveterate tinkerer - the person who toils away at an idea. They reach breakthroughs like stepping stones. The thing works! Sort of… and so the tinkerer goes back to work until the next breakthrough. Now this thing really works! . . . but it could be better, and so the cycle continues until innovation becomes so commonplace that it becomes an invisible protocol.

Think of email. There was a time not so long ago when email was a pretty radical concept. Instant mail? That doesn’t need to be written and types and folded up and sealed in an envelope and stamped and mailed? Just instant? Of course, today, email is so pervasive it’s become a protocol of life. What was once only a novelty is now so ubiquitous it’s become more of a nuisance. We become inured to novelty at a phenomenal rate.

And now the tinkerer of humanity has reached another milestone with Google’s new language model. This sucker is wild, and it displays it’s shocking coherence in a realm that I thought would take quite a lot longer to capture: humor. The examples that Google posted are excerpts of the language model explaining jokes. The joke is listed, and then the language model responds with a step by step description of how the joke functions. It’s shocking because this thing explains jokes better than I could, and it does this instantaneously. 

It does well to remember that Wilbur Wright told his brother that “Man will not fly for 50 years.” He said this in 1901.

In 1903 Wilbur Wright and his brother achieved flight. So even the world’s expert in potential flight who was trying to make it happen was wildly off when he tried to predict how long it would take to figure out.

Now imagine asking someone random in 1901 how long they thought it would be before man would fly? Someone that had no idea what the Wright brothers were up to. Chances are solid this random person would laugh and say “not in my lifetime!”

Technology today is developing in hundreds and thousands of pockets, all of which are brewing to the ignorance of almost everyone on the planet. How many people know about GPT-3? Let along Google’s latest language model? Perhaps it gets a spot on CNN, but it’s old news before the segment even finishes and rarely does anyone pause to think about the implications.

Tony Seba is long known for his predictions and analysis regarding disruption technologies. How they arise on S-curves, and then combine and piggy-back off one another, joining, like transformers to create something hitherto unimaginable. The iPhone is a great example: an icon + a cell phone + an internet browser. Bingo. Three separate but similar technologies joining together, all of which required similar underlying technologies to develop before, like the harnessing of electricity, the development of wifi, hypertext transfer protocol, mp3 formats….Modern cellphones represent a vast collection of much smaller incremental human inventions.

We generally see only the consumer facing side of technology, but there’s more going on with the peripheries of this growing realm of tech, and we can’t see it yet. We think these things are in the far future, but in some unsettling cases, the future is already here, it just hasn’t been evenly distributed yet.

HERD AUTOIMMUNITY

April 4th, 2022

Autoimmune diseases have been on the rise for quite a few decades. No one is sure why, but a leading theory within virology is The Hygiene Hypothesis. The tldr  of this hypothesis is that our immune system has always been revved up to a high gear because for the majority of our evolution we were in circumstances that guaranteed enemies for the immune system to deal with. More specifically, large parasites were ubiquitous, and because of these unwelcome friends, our immune system evolved to constantly be in a state of aggression. But today, few in the developed world have such parasites. Our environment has become so clean that our immune systems are left without their usual sparring partners and in their absence, the trigger happy aggression of our immune system is taken out on the body itself instead.

This is a tidy example of unintended second-order effects. When something outlives it’s usefulness, does it then begin to unravel the improved situation?

“Embedded within the fruits of their success are the seeds of their decline.”

This is a quote from Ray Dalio’s latest book entitled “Principles for Dealing with the Changing World Order.” In this book he details his thesis about the rise and decline of world empires. As the quote indicates, his thesis holds that the situation generated by hardworking people contributing to a rising empire in fact lays the foundation for its undoing. Fueled by innovation and creativity a society greatly improves, making the value of its currency rise.  But then there’s a switch in the culture and the society. The improved condition undermines the mindset of the society as a whole. People are less driven to improve their condition because at this point, it’s one of the best conditions that can be found around the world. The child of a wealthy individual who worked very hard to rise out of poverty does not experience that same poverty that shaped their parent’s mindset. This is an oversimplification, but an appropriate one which captures the larger trend. Another aspect of the change in society is relative condition. For a country or a group of people who are just beginning to rise, they are all essentially in the same boat. But when an empire has matured, that’s no longer the case. While the general condition of most people in such an empire might be much better than conditions found in most every other country, the economics of innovation inevitably create large gaps in wealth. People in the empire don’t compare themselves to people in other countries, they compare themselves to those who have it the best within their own country.

Notice the difference in mindsets here: when a nation begins to rise, the predominate mindset is to work hard and to try and improve things. At the top that mindset has vanished from the majority population. The “enemy” of poverty is absent, and this has a fundamental impact on mindset which develops without that worthy adversary of poverty, and just like the immune system which is overreactive in the absence of parasites and other expected enemies, the mindset of a population begins to rile against itself. Populism rises as people bicker and gaps in wealth become the concern of most people. The virtuous cycle tips into a vicious cycle and the status achieved by innovation and hard work is undone in the absence of a mindset that generates innovation and hard work.

Generations of people born during or after the height of an empire fundamentally lack the situation necessary to develop a perspective required to continue growing that empire.

Just as our own bodies begin to unravel themselves in the absence of expected hardship, societies unravel in the absence of the conditions that created them.

Something about all this feels odd, in the same way we can witness ourselves taking part in a behavior that undermines our larger goals. Like opting for dessert over and over when we have a larger goal to lose some weight. Do we have to make the same mistakes over and over? Like a bad habit? Like another flair up of an autoimmune condition? Like the crash of another country?

Clearly not. Changes can take place that enable people to achieve their goals, and not everyone has an autoimmune condition. But so far - given human history - all empires fall.

It’s easy to fall into the trap of thinking that nothing changes, and that cyclical crashes are a fundamental aspect of human experience. But innovation is itself the process of discovering and leveraging a fundamental change in the way things work. Da Vinci and countless people before him dreamed of flight, and it was likely widespread for people think it would be impossible. But today flight is commonplace. The same applies to something as simple as soap and germ theory. Countless people have died due to an ignorance about germ theory, but once the concept took hold, it changed human behavior. There’s a particular combination of awareness and imagination that’s required for innovation. It’s not enough to simply imagine a better future; an awareness of the present circumstance is required. A sensitivity of perception is fundamental for realizing something about the current situation that can help us change it. Ray Dalio’s book is perhaps the first inkling of a mass awareness of a problem that exists at the size of societies. Our ability to identify the trend - to become aware of the problem is likely our first step to innovating our way out of it.

A LUCILIUS PARABLE: A HELPING HAND

April 3rd, 2022

—ow and massaged his temples as he struggled to think. It just didn’t make sense, at least not yet. He looked up at the Robot again who was analyzing Lucilius’ facial expression, extrapolating it’s possible interpretations in a thousand different directions, all with varying degrees of frustration, confusion and a sense of searching for understanding. The Robot had a 96% confidence score for the most likely thought Lucilius was having at that very moment, and with it, the Robot further extrapolated and tested 2.2 million possible sentences that it could say to alleviate Lucilius’ confusion and help understanding blossom in his mind. Each of the 2.2 million sentences was weighted for a final confidence score. The same process was carried out with the next several thousand potential thoughts Lucilius could be having which the Robot had generated using its model of Lucilius’ mind, each potential correct thought scoring with less and less confidence. While it was unlikely, this exercise was carried out because every so often a low confidence guess at Lucilius’ thought yielded a better generated response when it was used to propagate potential responses. These errant successes were then used to further refine the model of Lucilius’ mind generated by the Robot who was currently trying to teach him a very simple idea about the paradoxical incompleteness of language.

“Ok,” Lucilius said, “say it one more time.”

The Robot hesitated a brief moment, analyzing Lucilius’ listening receptiveness via his facial expression, and when his muscles had relaxed in just the right amounts to signal maximum reception, the Robot spoke.

“The following sentence is false. The prior sentence is true.”

After immense calculation, the Robot continued “It may help to look at the statements, here look.”

The Robot projected the text it had spoken into the air before Lucilius face. He looked at each sentence in turn, his eyes switching from one to the other, over and over.

“Well, on their own, they make sense, but when paired together it doesn’t make sense at all”

After immense calculation, the Robot responded in an instant. “Can each sentence really be regarded on their own?”

Lucilius considered this. “Well no, because each one refers to the other. If there was only one then it would imply the absence of a missing statement to refer to.”

“Precisely,” said the Robot. “So, how does that inform your previous conclusion which was, and I quote ‘Well, on their own, they make sense, but when paired together it doesn’t make sense at all!”

“Ok,” Lucilius admitted. “I guess they don’t make sense on their own, which means they can only make sense together… but when they are paired together they don’t make sense again! So there’s not too much difference between thinking about them together or on their own, they’re either incomplete or they put your mind in a stupid loop that makes no sense!”

The Robot smiled, this Robot had been with Lucilius his whole life, since he was a tiny baby, helping him learn all manner of things at a much faster pace than had the Robot not been with Lucilius. The AI residing within the Robot body was specifically tuned to model the mind of Lucilius with the express purpose of being able to calculate and carry out the optimal learning challenge for Lucilius at any given moment. The Robot was not there to give Lucilius answers, but to help him to navigate his own way towards an answer more efficiently.

“I think you understand now,” the Robot said.

“What!?” Lucilius cried out. “Are you kidding?”

“Not at all, the point of the paradox is taking hold and becoming apparent.”

“What, that it doesn’t make sense?”

“You are correct, but not precise.”

Lucilius contemplated this a moment, and thought back to the infuriating pair of sentences.

“They don’t make sense simply because it’s a contradiction, like saying ‘This is not a sentence’ even though it clearly is. But it goes beyond that because it’s a recursive contradiction built in two separate parts. Each only becomes potentially incorrect when the meaning of the other is taken into consideration.”

Lucilius shook his head. “I feel like I could talk about this in circles all day and I’d never get anywhere.”

“Sounds like a good description of many conversations I’ve witnessed you having with friends.”

Lucilius laughed. “You have a point. I’m not sure I like it, but you have a point.”

“Maybe this simple paradox is an analogy in miniature for those conversations. What exactly would that mean?”

Lucilius sat back, as if refreshed by the larger perspective the question afforded.

“Hmm… You weren’t actually joking. It’s exactly how people can be so opposed and yet at their core they basically want the same things. We are always somehow talking past each other, but assuming something about the other person in the same way that infernal pair of sentences refers to each other.”

The Robot smiled again, pleased with the match between reality and its generated predictions.

This entire interaction was not in fact real, but simply an abstracted interpretation of a heuristic that occurred in the slightest fraction of a microsecond in the form of a thought about where Lucilius hoped the conversation would go. He was in fact imagining himself in the role of his student, as all good teachers do. And this ideal interaction was his goal - the place he wanted to arrive at, but his student was having quite a lot of trouble grasping the concept. He looked at the confused Robot sitting with him. The Robot he had built with his own hands, who he had been trying to train and teach for its entire existence.

The circular lights indicating the Robot’s eyes flicked, as if to indicate a blink, and Lucilius sighed, knowing that the poor Robot was confused. It was still early days, and Lucilius was constantly tinkering with the design of the Robot’s mind in an attempt to help the Robot learn faster and more efficiently so that one day it could be replicated and it could be the ultimate teacher for hundreds and thousands -millions and billions- of kids.

“The first statement must be incorrect,” the Robot stated.

“That’s not the point,” Lucilius said.

“What is the point?” Asked the Robot.

“What does it imply about language if these statements can exist as valid statements?”

“They cannot be valid statements. The first statement must be incorrect.”

Lucilius chuckled at the predicament. He sighed, and frustrated, Lucilius raised a hand to his furrowed br—

EVENT HORIZON

April 2nd, 2022

What is the mental equivalent of an event horizon? In astronomy, the event horizon is a relatively thin area that surrounds a black hole. It marks a threshold of profound change in the amount of gravity the black hole exerts on surrounding objects, and once the event horizon is touched, these gravitational changes are so profound that matter, time and space start blurring and perhaps even blend together. To come into contact with the event horizon almost certainly means there’s no coming back. 

What’s the mental equivalent of this sort of thing? As an analogy it evokes a pretty dark picture - something like depression, but it need not. Think about for example the idea of learning something new, something that creates a realization, an ah ha! moment that changes the way you see the world. Isn’t that also appropriate? It’s not exactly easy to “unlearn” something we’ve gained an awareness of.

Monitoring changes in one’s own perspective and mental well-being is incredibly valuable for detecting certain event horizons in mental life. While we can’t predict a new idea we might have or encounter tomorrow, we can become acutely aware of how our own mental machinery is trending. It can be rather courageous to admit that you’re not doing well, but it can be life saving to see that situation coming from a mile away. It’s recognizing that somewhere, recently a wrong turn was taken, and a course correction needs to be taken before the effects of that wrong turn compound to a degree that is increasingly difficult to do anything about.

That being said, there’s still a great deal of mystery that remains, which also maps onto the analogy. Black holes can also eject matter, at tremendous rates, which is rather counter intuitive to something that has more gravitational pull than anything else. What exactly goes on in the inner workings of a black hole is still a mystery - but so is the human mind. Pits of depression and anxiety can end up producing incredible results, thought it’s a rarity, and certainly not a reliable strategy. How exactly the mind works is a mystery, despite the tremendous amount of resolution we have on its tiny mechanisms.

Pondering such realms can evoke some bizarre questions: can anxiety be useful? Can happiness lead you down a dark path and backfire? If there was a 70% chance of having an incredible life after a major depressive episode, would you opt for the depression?

Framed in such emotional terms, the case for something like Stoicism becomes rather compelling, but it also highlights an even stranger question: is it possible to go through a massive depressive episode without being consumed by it? This might seem like a contradiction, but think of it in terms of anger: is it possible to experience great anger without acting on it? With the emotion swapped out, the answer is crystal clear: of course it’s possible to experience an emotion without letting it have a meaningful effect on your external reality. But the swap is a bit of a hack here. Not acting on a sudden spike in anger depends on additional factors: a certain level of self-awareness, a calculation of consequences, outcomes and a genuine ability to exist as a contradiction.

In our quest to simplify things for explanatory power, we’ve likely glossed over crucial elements of complexity. There’s many layers or overlapping realms that exist to make up a single human being. A humorous observation of this point is that multiple personality disorder implies that there is multiple personality order. We contain multitudes, as Whitman once wrote. Not only this but we all contain a different set of multitudes, and any attempt to understand a broad spectrum of people must generalize and therefore gloss over context and nuance - the very things where useful answers often lie.

The fact is that happiness might render one person totally useless, equivalent to being in a drug-induced state, whereas a little bit of anxiety might help another person get motivated to do some work that feels fulfilling at the end of the day. One way to think about this is to think about there existing an enormous array of sliders that make up a person’s particular state and abilities. Then think about how different activities impact these sliders in different ways: how a workout alters all these different levels, or how bad news effects them, or several years of consistent daily meditation.

An event horizon might be represented as a certain configuration of these sliders that if achieved, becomes extremely difficult to change back to previous ways, and the point here is that we can develop new realms within our own sense of self, and one of those realms can be a kind of monitor that touches, senses and measures all the other realms, and slowly, over time, gains a dynamic understanding of which combination of shifts in personal mental color lead potentially irreversible states. Without such a lookout, we are like ships sailing blind, bound to one day collide, leaving it to chance whether we collide with a soft and sandy beach or an iceberg in a cold and dark place.