Game theory is extremely important when creating systems for civilization. We ignore it at our own peril.
Game theory explains why capitalism generally creates more productivity than communism. In a system where reward is closely tied with the amount of valuable goods/services one is rendering, the incentive to be productive is very high. But if the reward for being productive is similar to the reward for just pretending to be productive, more people will tend to gravitate towards the latter.
Legal punishments revolve heavily around game theory in order to hold teeth. In order for a crime to be deterred, the expected payoff must be lower than the punishment for getting caught multiplied by the probability of getting caught. If you don't make sure this condition holds true, people will just keep doing that crime. Example: A train ticket costs $2; the fine for hitching a free ride is $100 but the chance of getting caught per ride is 1%. Then, in the long run, cheating every day is actually cheaper than following the law! The expected value in this example is higher for cheaters than for rule-abiders. In order to fix this flaw, the fine has to be raised to at least $200, or the chance of getting caught has to be raised to at least 2%.
This math concept also explains why forging a parking permit carries such a heavier fine than parking without a permit. It's not inherently more harmful. It's simply harder to catch.
This brings us to how laws apply to the police. The game theory on this is just totally bonkers and completely broken. I don't even know how it lasted for so long. Normal people go to jail for beating innocent people with a stick. But cops usually just get fired and rehired elsewhere. Civilians are deterred from letting their anger get the better of them and becoming violent, because they would have to answer for their violence with jail time. There is currently no such deterrent for a police officer doing those same heinous acts of violence!
The game theory flaw has implications way beyond those currently on the police force. It could be encouraging a certain type of person to join the police force just to gain immunity for their future crimes. Psychopathic bullies from all around the country have recognized that if they want to commit violent crime without suffering legal consequences, all they have to do is join the police first. Obviously there are many good people who join the police force and I'm not saying all or most cops are psychopaths. However, bad people exist too, and our laws are currently offering would-be criminals the opportunity to fulfill their violent fantasies with no legal repercussions, as long as they're willing to work for it!
As long as this legal loophole exists, we will keep seeing psychopaths/bullies becoming police officers to satisfy their urges. Recently many have proposed other solutions such as "defunding" or "training", each of which only slightly improve the situation without fixing the fundamental problem. Defunding just makes the problem smaller, and doesn't get rid of it. Extra training just increases the amount of work a bully has to do to fulfill their fantasies; it doesn't discourage them altogether. For truly sweeping positive change, we need to fix the Expected Value of committing violent crimes as a police officer, which is currently positive for people who enjoy power tripping. The legal consequences of abusing power as a cop must be at least as harsh as the legal consequences of committing the same acts of violence as a civilian.
Monday, June 8, 2020
Sunday, May 3, 2020
How to actually win at PUBG
The most common mistake people make in PUBG is to assume getting kills increases your chance of winning. It really doesn't.
The intuition that getting kills is a good idea for winning, is a relic from traditional shooters. In deathmatch, getting kills increases your score. In esports like CS, getting kills greatly increases your team's chance of winning the game.
But in a battle royale game where the objective is to be the last person standing among 100 people, getting into a firefight is a losing situation for all parties involved, even the survivors. Whether you are reducing the player population from 50 to 49, or from 4 to 3, it is not a good thing that everyone around you is now aware of your presence due to fired shots. Instead of killing them yourself, it would be better to let someone else kill them. In a perfect ideal play, you'd get literally 1 kill: The 2nd-to-last person other than yourself. (In practice, of course one will have a few forced encounters and have to kill some people out of self-defense).
My current strategy is called the "early-bird camper". It has consistently landed me in #2 or #1 in one-man-squad and solo modes: Each phase, run into a building near the center of the safe zone circle as early as possible. Being early minimizes my chance of encountering someone already there. Being near the center minimizes my chance of having to move during the next phase (in terms of math, a circle within a circle is much more likely to include the center than an edge, and depending on the relative sizes of the circles, it could even be guaranteed to include the center).
Granted this assumes you are actually playing to win. Many players have stressed that they enjoy running around and racking up kills in PUBG as their preferred way of playing and thus aren't all that focused about getting the Chicken Dinner victory. But if you want the Chicken Dinner, being a cowardly camper is the way.
P.S.
I recently stumbled across an article on medium which contains a lot of data about what kind of behaviors are most likely to result in victory. As you can see from my responses to that article, I'm really happy about how much data is available for us to analyze, but I believe two of the author's interpretations of the data is flawed.
The intuition that getting kills is a good idea for winning, is a relic from traditional shooters. In deathmatch, getting kills increases your score. In esports like CS, getting kills greatly increases your team's chance of winning the game.
But in a battle royale game where the objective is to be the last person standing among 100 people, getting into a firefight is a losing situation for all parties involved, even the survivors. Whether you are reducing the player population from 50 to 49, or from 4 to 3, it is not a good thing that everyone around you is now aware of your presence due to fired shots. Instead of killing them yourself, it would be better to let someone else kill them. In a perfect ideal play, you'd get literally 1 kill: The 2nd-to-last person other than yourself. (In practice, of course one will have a few forced encounters and have to kill some people out of self-defense).
My current strategy is called the "early-bird camper". It has consistently landed me in #2 or #1 in one-man-squad and solo modes: Each phase, run into a building near the center of the safe zone circle as early as possible. Being early minimizes my chance of encountering someone already there. Being near the center minimizes my chance of having to move during the next phase (in terms of math, a circle within a circle is much more likely to include the center than an edge, and depending on the relative sizes of the circles, it could even be guaranteed to include the center).
Granted this assumes you are actually playing to win. Many players have stressed that they enjoy running around and racking up kills in PUBG as their preferred way of playing and thus aren't all that focused about getting the Chicken Dinner victory. But if you want the Chicken Dinner, being a cowardly camper is the way.
P.S.
I recently stumbled across an article on medium which contains a lot of data about what kind of behaviors are most likely to result in victory. As you can see from my responses to that article, I'm really happy about how much data is available for us to analyze, but I believe two of the author's interpretations of the data is flawed.
- First, there is a finding that higher # of kills is strongly correlated with longer survival times. But the causation is reversed, because people who survive longer are more likely to encounter more people, even if everyone is using the same avoidance strategy. This is a typical example of survivorship bias.
- Another criticism I had is the article's conclusion that being a total camper is a losing strategy, because the distance traveled was positively correlated with amount of time surviving. Again, there is an obvious survivorship bias at play, because the people who survive longer are obviously going to have greater travel distances than people who died early (in fact my "early bird camper" strategy would've been measured as "high travel distance" even though I'm being a camper)
Monday, March 16, 2020
Why is a rooster sound spelled cock-a-doodle-doo?
As a kid I often wondered why the onomatopoeia for a rooster call seems to have zero relation to how it actually sounds. An actual rooster call sounds like "er, er-er, er-er", so why is it described as "cock-a-doodle-doo"? The answer is unlike many animal calls, the defining attribute of a rooster call is more in its rhythm than in its timbre. They needed a way to capture this rhythmic aspect, and using commas or dashes to denote rhythm would've rendered the onomatopoeia rather inelegant and hard to remember. What's interesting about "cock-a-doodle-doo" is that it is practically impossible to say it without the swing-like rhythm which was intended by its inventors. English speakers are then meant to infer that this combination of words represents only the rhythm of a rooster call rather than its actual sound.
How "achoo" is actually pronounced during a sneeze
It's a myth that "achoo" is pronounced "aah-CHOO" with the emphasis on the "choo" part. The way the word was originally invented, it's actually "ACH-oo". "ACH" is the big expelling factor and "oo" is a very quiet sub-vocalization after the main part of the sneeze.
A sneeze generally has 3 parts: Inhale, big expelling ending in "sh", and sometimes a very quiet vocalization after the "sh" as the air expelling draws to a close. The big expel always has an "sh" sound. The loudest sound is the fast-moving air through the "sh" syllable. This is universal and it's literally impossible not to do it while sneezing. So it is simply not possible for "oo" to be part of the loud sound, unless of course one were to tack it on voluntarily after sneezing. Anyone sneezing with no cultural background (e.g. babies, small children), will form a sound resembling more "ASCH-oo" than "aah-CHOO". In fact literally the only people who pronounce it the second way, are people who learned the word "achoo".
Observing the full list of sneeze sounds in every language, it appears the majority including put emphasis on the 1st syllable while a few including English put emphasis on the last syllable. It is my hypothesis that unfortunately at some point in history, the onomatopoeia became misinterpreted. Now a lot of children's books and cartoons depict people saying "aah, aah, aah-CHOO!" and the unnatural pronunciation is stuck in culture.
A sneeze generally has 3 parts: Inhale, big expelling ending in "sh", and sometimes a very quiet vocalization after the "sh" as the air expelling draws to a close. The big expel always has an "sh" sound. The loudest sound is the fast-moving air through the "sh" syllable. This is universal and it's literally impossible not to do it while sneezing. So it is simply not possible for "oo" to be part of the loud sound, unless of course one were to tack it on voluntarily after sneezing. Anyone sneezing with no cultural background (e.g. babies, small children), will form a sound resembling more "ASCH-oo" than "aah-CHOO". In fact literally the only people who pronounce it the second way, are people who learned the word "achoo".
Observing the full list of sneeze sounds in every language, it appears the majority including put emphasis on the 1st syllable while a few including English put emphasis on the last syllable. It is my hypothesis that unfortunately at some point in history, the onomatopoeia became misinterpreted. Now a lot of children's books and cartoons depict people saying "aah, aah, aah-CHOO!" and the unnatural pronunciation is stuck in culture.
Thursday, March 5, 2020
Traditional tea brewing is overrated
What is the best way to enjoy a batch of fresh high-mountain oolong tea such as Taiwan's esteemed "Jin Xuan"?
Just put hot water on it, wait for 5-10 minutes, and enjoy the whole thing. Seriously. The only benefit to the traditional short steep times of 40-60 seconds is that you can taste how the flavor of the leaves changes over time, like telling a story over time. That's nice, but what if I want to taste the whole story in the same sip? Steeping for a long time gives you its entire complex flavor in a single serving, rather than giving you slices of its full glory diluted across multiple servings.
I challenge advocates of traditional brewing to administer controlled taste testing experiments to the general public comparing short-steeped tea to long-steeped tea and see which one is more popular. I would bet some money that the long-steeped tea would be more popular.
Also, do NOT throw away the first batch. The initial steeping has the strongest, best flavor. The reasoning that it gets rid of pesticides isn't correct, because you're actually getting rid of tea flavor at the same rate at which you're getting rid of other chemicals. You may as well just throw away the first 10 batches to get rid of 99% of the pesticide and also end up with 1% tea flavor. Also, "awakening the leaves" not only has no scientific basis, but also is simply a logical fallacy: It's not like throwing out the first batch is going to make the water spend more time with the "awakened" tea leaves than it would've if you'd just left it in.
Just put hot water on it, wait for 5-10 minutes, and enjoy the whole thing. Seriously. The only benefit to the traditional short steep times of 40-60 seconds is that you can taste how the flavor of the leaves changes over time, like telling a story over time. That's nice, but what if I want to taste the whole story in the same sip? Steeping for a long time gives you its entire complex flavor in a single serving, rather than giving you slices of its full glory diluted across multiple servings.
I challenge advocates of traditional brewing to administer controlled taste testing experiments to the general public comparing short-steeped tea to long-steeped tea and see which one is more popular. I would bet some money that the long-steeped tea would be more popular.
Also, do NOT throw away the first batch. The initial steeping has the strongest, best flavor. The reasoning that it gets rid of pesticides isn't correct, because you're actually getting rid of tea flavor at the same rate at which you're getting rid of other chemicals. You may as well just throw away the first 10 batches to get rid of 99% of the pesticide and also end up with 1% tea flavor. Also, "awakening the leaves" not only has no scientific basis, but also is simply a logical fallacy: It's not like throwing out the first batch is going to make the water spend more time with the "awakened" tea leaves than it would've if you'd just left it in.
Thursday, September 12, 2019
Disproving the Boltzmann Brain with Evolution
The Boltzmann Brain idea is that the thought in your brain at this very moment could be the result of randomly bouncing particles (e.g. particles in a soup or gaseous state). So maybe you're not actually here reading these words. Maybe the universe we know, the milky way, and earth doesn't actually exist. You're just a product of a momentary serendipity of particles randomly bouncing in a very specific way which just so happened to correspond to this exact thought you're thinking of, about reading these words. The past didn't actually happen.
Proponents of the Boltzmann Brain idea claim it's way more likely you're a Boltzmann Brain than a real brain, because there must be such a surplus of disordered particles (like randomly bouncing gas) that surely the thought in your brain is a result of one of those processes rather than one which takes millions of years.
My counter to the Boltzmann Brain is that the estimated probabilities are all wrong. Yes, there is a non-zero chance that a bunch of randomly bouncing particles suddenly, out of nowhere, produced the very specific complex thought your brain is thinking right now. But there is a far greater chance that they did it with a more gradual process. The particles became stars and planets. Some of those particles became chemicals, which became rudimentary life, which eventually evolved into humans, which became you. I contend that the latter scenario is way more likely than the former scenario, and the reason is evolution.
If you think about it, once you have the most basic form of life and DNA (or really just any self-perpetuating thing that mutates and can thus be subjected to natural selection; it does not have to be DNA as we know it), there becomes an arms race for greater awareness and intelligence for survival, and it's only a matter of time before the emergence of interesting and sentient life. In other words, once you have the scaffolding for evolution, life will naturally tend towards higher and higher complexity, just as surely as water falls downward, or as entropy increases in a closed system.
What is more likely: That a group of particles spontaneously formed a thought, or that a group of particles spontaneously formed the barebones scaffolding which allows for pre-single-celled life? We know that a thought of a brain is extremely complex, whereas all that's needed for evolution is some basic form of information which can be passed on and mutates once in a while -- this starts a snowball effect whereby intelligence has a high chance of arising naturally and is practically inevitable. Thus, taking evolution into account, it seems far more likely that you evolved out of more basic building blocks via a long and gradual process, than that you're a Boltzmann Brain.
Proponents of the Boltzmann Brain idea claim it's way more likely you're a Boltzmann Brain than a real brain, because there must be such a surplus of disordered particles (like randomly bouncing gas) that surely the thought in your brain is a result of one of those processes rather than one which takes millions of years.
My counter to the Boltzmann Brain is that the estimated probabilities are all wrong. Yes, there is a non-zero chance that a bunch of randomly bouncing particles suddenly, out of nowhere, produced the very specific complex thought your brain is thinking right now. But there is a far greater chance that they did it with a more gradual process. The particles became stars and planets. Some of those particles became chemicals, which became rudimentary life, which eventually evolved into humans, which became you. I contend that the latter scenario is way more likely than the former scenario, and the reason is evolution.
If you think about it, once you have the most basic form of life and DNA (or really just any self-perpetuating thing that mutates and can thus be subjected to natural selection; it does not have to be DNA as we know it), there becomes an arms race for greater awareness and intelligence for survival, and it's only a matter of time before the emergence of interesting and sentient life. In other words, once you have the scaffolding for evolution, life will naturally tend towards higher and higher complexity, just as surely as water falls downward, or as entropy increases in a closed system.
What is more likely: That a group of particles spontaneously formed a thought, or that a group of particles spontaneously formed the barebones scaffolding which allows for pre-single-celled life? We know that a thought of a brain is extremely complex, whereas all that's needed for evolution is some basic form of information which can be passed on and mutates once in a while -- this starts a snowball effect whereby intelligence has a high chance of arising naturally and is practically inevitable. Thus, taking evolution into account, it seems far more likely that you evolved out of more basic building blocks via a long and gradual process, than that you're a Boltzmann Brain.
How to reverse entropy
Believe it or not, it is possible for entropy to reverse in a closed system, at least according to current understanding of physics.
Entropy is purely a statistical process, which means the only reason we tend towards higher entropy, is that there's a higher chance of going to a high-entropy state than a low-entropy state. A common analogy is throwing a deck of cards in the air. What are the chances that after being thrown in the air, a deck of cards comes back and lands in a perfectly neat pile and completely in order? Almost zero. ALMOST is the key word of course, and the fact is if you had an infinite amount of time and did this forever, it would happen eventually (you wouldn't actually need an infinite amount of time. The point is, it would happen eventually, given a long enough time).
This mathematical concept applies to any process in the universe. Glass unshattering? Life randomly forming? Way less likely than cards landing in order, but it's the same principle; you just have to wait that much longer (way longer than the age of the universe)!
If you're still not convinced, just apply the proof by induction. Say there's a 0.01 chance of some pocket of bouncing matter lowering its entropy by some small amount over a small amount of time. Then there's a 0.01^2 chance that it happens twice in a row, and 0.01^3 chance it happens three times in a row. You can do this as many times as you want and end up with arbitrarily low entropy, where the matter has been reshaped to pretty much anything you want -- statue of liberty, a computer, or a human brain. Yes, the end result probably is an abysmally small number, but it's bigger than zero!
The problem when applied to entropy is that the universe is also expanding. Which means we don't have infinite time to wait for the metaphorical cards in the air to finally land in an interesting order, because the cards are getting further and further apart as time passes so that eventually they won't even be able to interact with each other at all. That's no fun.
So how do we give the universe time to reverse entropy, even though it's expanding?
What we need to do is figure out a way to give some system enough time to do random stuff for entropy to reverse. And this system has to be truly closed, not like an expanding universe.
We need to build an unbreakable, totally insulated box with a lot of stuff inside it -- enough matter/energy to make interesting information and/or life as we know it. Once you've built the box all you have to do is wait.
Unfortunately for us this box will just be a stagnant uniform soup for as long as we're alive. By the time the randomly bouncing matter in the box spontaneously develops into something interesting by pure chance, we'll probably be long gone. But I am sure whatever alien is birthed from this once-in-an-eon lowering of entropy will be grateful that it exists whether or not we're around to see it.
Actually, we probably wouldn't be allowed to see what's inside anyway, as that would violate the requirement that the box must be unbreakable and not even light can escape it, to insulate it from the expansion of the universe.
Entropy is purely a statistical process, which means the only reason we tend towards higher entropy, is that there's a higher chance of going to a high-entropy state than a low-entropy state. A common analogy is throwing a deck of cards in the air. What are the chances that after being thrown in the air, a deck of cards comes back and lands in a perfectly neat pile and completely in order? Almost zero. ALMOST is the key word of course, and the fact is if you had an infinite amount of time and did this forever, it would happen eventually (you wouldn't actually need an infinite amount of time. The point is, it would happen eventually, given a long enough time).
This mathematical concept applies to any process in the universe. Glass unshattering? Life randomly forming? Way less likely than cards landing in order, but it's the same principle; you just have to wait that much longer (way longer than the age of the universe)!
If you're still not convinced, just apply the proof by induction. Say there's a 0.01 chance of some pocket of bouncing matter lowering its entropy by some small amount over a small amount of time. Then there's a 0.01^2 chance that it happens twice in a row, and 0.01^3 chance it happens three times in a row. You can do this as many times as you want and end up with arbitrarily low entropy, where the matter has been reshaped to pretty much anything you want -- statue of liberty, a computer, or a human brain. Yes, the end result probably is an abysmally small number, but it's bigger than zero!
The problem when applied to entropy is that the universe is also expanding. Which means we don't have infinite time to wait for the metaphorical cards in the air to finally land in an interesting order, because the cards are getting further and further apart as time passes so that eventually they won't even be able to interact with each other at all. That's no fun.
So how do we give the universe time to reverse entropy, even though it's expanding?
What we need to do is figure out a way to give some system enough time to do random stuff for entropy to reverse. And this system has to be truly closed, not like an expanding universe.
We need to build an unbreakable, totally insulated box with a lot of stuff inside it -- enough matter/energy to make interesting information and/or life as we know it. Once you've built the box all you have to do is wait.
Unfortunately for us this box will just be a stagnant uniform soup for as long as we're alive. By the time the randomly bouncing matter in the box spontaneously develops into something interesting by pure chance, we'll probably be long gone. But I am sure whatever alien is birthed from this once-in-an-eon lowering of entropy will be grateful that it exists whether or not we're around to see it.
Actually, we probably wouldn't be allowed to see what's inside anyway, as that would violate the requirement that the box must be unbreakable and not even light can escape it, to insulate it from the expansion of the universe.
Subscribe to:
Comments (Atom)