Posts
Comments
Correlation is linear. Many causal functions can be non-linear.
Think of medicine. X is the dosage, Y is the improvement of health. If the dose is too low, you will get no response. If the does is within a good range, health improves. If the does is too high, you will begin to get even sicker. If data was gathered all along this inverted parabola, the correlation might be zero. But there is still a causal relationship between health and dosage.
Thus you can have causation without correlation.
You can probably think of many such functions with diminishing or negative returns as the dosage increases, e.g. years of education vs. lifetime earnings.
Whether you see a positive, negative, or null correlation can depend on where you sample from the response function. In the "real world" data might be bunched up around certain regions of the response function. Thus for the "average person/instance" you can determine if there is a correlation or not, and then say this is basically the causal effect (for the average person/instance).
But if you want accuracy and precision over concision you will use a more complex model.
Concise models are better memes than complex models, however, and so we are flooded with linear models or binary models.
The only app you will ever need is "Google Sheets".
- Its UI is easy to use on a phone and syncs instantly, so you can use a computer as well.
- It is completely customizable. You enter any sort of info you want and analyze it later over time.
I started using it ~3 years ago as a time-sheet to track my work, activities, sleep, mood etc. I tried to do what you did - see how X impacts my mood and look for trends. Here's some of my findings over the past three years in that regard:
- The most helpful thing has been a daily "end of day" entry for:
- What went well?
- Where did I come up short?
- What one easy step could I take to improve where I came up short?
- What is something I am grateful for?
- What is my plan for tomorrow morning? (Food, Work, Routine, etc.?)
- What improves my overall mood the most?
- Gratitude journal entries 2x/day (morning & evening)
- What influences my energy throughout the day?
- Carb-heavy lunches. Obvious in hindsight.
- 30-mins of relaxation (e.g. web-surfing) are as fulfilling/energizing as 2-4 hours. Use Pomodoro (or the time-sheet) to shame you into getting back on task.
- Weekly reflections on your time sheet and what went well/bad are very helpful and make you remember much more of your life that you would otherwise forget without realizing it.
- "Chaining" tracking how many days in a row you've maintained a new habit can be very helpful to keep the new habit going.
- Tracking too many things can derail you so keep it small. (e.g. Ben Franklin style "virtue journal" was not much of a value-add for me and takes too much time to be worth it).
You learn by doing it.
Find someone you like (or who likes you) and start dating them!
Without dating apps:
- Join a few social/intellectual/school clubs or co-ed sports (e.g. running). Be a generally cool person and you'll find someone after a while. Or...
- Go to lots of low-key house-parties thrown by friends, or throw them yourself. See if there is any chemistry with "friends-of-friends." In these situations there is lower risk for each of you since you've been "vouched for" by mutual friends, and you've seen each other in a low-stakes setting. Bonus points if you are the one hosting/organizing the party.
With dating apps:
- Think of fun things you'd like to do if you had someone along: e.g. hiking, comedy show, exploring downtown, bar-hopping, etc.
- Use an app like "Coffee Meets Bagel" that limits how much time you spend on it each day.
- For each person you "like", mention one thing written in their profile and one thing from one of their more adventurous photos (generally one towards the back). Say something like "You look so excited in that photo with the ____, there must be a good story behind it!"
- This naturally starts the conversation on something and shows that you admire them in a way that other people may not comment on.
- Keep the back-and-forth relatively short (just within one day). Ask if they'd like to meet on "Thursday at 7 for drinks at ___. If it goes well we can take it from there." Suggesting a hard time and place before asking their schedule is easier for them to say "yes" or "No, how about Friday at 6" instead of asking "when are you free."
- To prepare for your date look at the map and make a short list of things to do / places to go depending on the mood of how the date is going: Chill, High Energy, Make-out spot, etc. Don't tell your date that you have additional options planned until you meet - just say "I have some ideas of additional places to go depending on how we feel later." She will generally be impressed. This takes a bit of time, but pays off a lot. Also, if the date isn't going well, you can end it as early as you want, without revealing you had more plans.
- Don't treat each date as a potential partner until you've seen them >3 times. Just treat it like a more intimate meetup.com or couch-surfing person. (unless there's lots of chemistry, in which case you can turn it up). Have fun meeting lots of different people and doing lots of fun things. Some dates will be hot, some will be duds. Some will be really great chemistry but then break your heart after a month of dating once you start to fall for them. Some will be perfect timing.
I know this isn't an "economisty" answer but it's pretty solid wisdom as far as I'm concerned. Also note this advice is geared toward long term relationships, not hookups. If you want to date for hookups instead you can checkout the many pickup artist websites.
WRT "who is still left in the dating pool" there are always some fraction of people who just broke up, or moved, or are still looking for the right guy/gal. As you get older there are fewer good single people, but they are also less choosy, and thus easier to get. Remember, you also stand out from the competition more when there are fewer good people.
I had this same thought recently regarding friendship and connection as "divide and conquer" for understanding the world.
WRT why people cling to collective worldviews despite contrary evidence... one answer lies in interesting work in Social Psych called "Terror Management Theory."
Basically, people seek to avoid death, but we all know that we will die. To avoid being paralyzed by this fear, we are socialized by our parents & society into worldviews that grant us "immortality" of sorts: religion, prestige, being "good" to be remembered by loved ones, survival our "our people", hedonism of the "now" to block out the future, etc.
Many of the other comments deal with thought experiments rather than looking at the reality of how "many worlds" is USED. From my point of view as a non-physicist it seems to primarily be used as psuedo-science "woo" - a revival of mystery and awe under the cloak of scientific authority. A kind of paradoxical mysticism for non-religious people, or fans of "science-ism".
An agent might act differently from MISUNDERSTANDING many worlds theory. Or by paying more attention to it. Psychological "priming" is real ansd powerful.
The answer by TAG below is case in point. For someone committed to a belief in determinism or fatalism, having a manyworlds theory in mind may buttress that belief.
I guess the actionable version is to develop transferable skills, abilities, wealth, or social capital that are highly valued by many different tribes.
Then you have the leverage to flit from one to the next, and not care about standing up for any particular tribe.
However, the game to acquire wealth, social capital, and valued skills is basically the game that we are all playing and has lots of competition. The only way to "opt out" is to join a local monopoly (i.e. a tribe). Also, in the real world, tribes often "loan us resources" to develop our skills, capital, etc. in exchange for "joining" the tribe.
I find this very true.
In fact, portraying a STRONGER identity often is met more easily results in better responses. The trick is that you can be strategic about it. By selecting between "personas" or "roles" you can select what kind of responses you want to get.
I find it helpful to think about the different situations I am in (work meetings, studying in cafes, meeting friends, etc.), and then think about "what is the most ideal response I could get" - and think about "what kind of person / action would provoke that kind of response?" Then, for the given situation make sure that everything is coherent - appearance, energy level, behaviors, speech cadence, etc.
Coherence is very powerful.
We already do this when we have a "work self" and a "home self". But for most of our activities it is not pre-planned. We just want to be "ourselves" - i.e. not have to strategically prepare for each situation.
As for "social identity theory" and feeling attacked, I don't think KYIS quite applies. When you are part of a tribe or subculture or whatever, there are several factors at play. (1) Defending the tribe may gain you status in the tribe. (2) Allowing attacks on fellow tribe-members to go unprovoked may put you personally at risk as well - thus the tribe makes it a value to protect fellow tribe-members.
KYIS may mean "don't join any tribes". Or more realistically - only feel kinship or trust toward those you personally know, not any abstract larger categories of people. Some would argue that this is how China used to work. However, as societies scale up in size, we typically do join social groups with abstract myths that bind people together, provide standards, and allow coordination among strangers.
Anyway, I guess it gets pretty complex as you unpack it. I suppose if you have skills that are in demand by many people, you do not need to be "married" to any one tribe, nation, or company. You can flit from one to the next if the current one falls. This may cause locals to mistrust you (e.g. the hatred for "globalists") which lowers your status locally, but if your skills are valuable enough, you won't mind too much.
So, the ultimate way to KYIS - be very valuable to many different groups of people. This may be from transferable skills, a great personality, or just a very strong and wide social network.
Does anyone really track the marginal utility of their possible investment this way? Utilons - sure. But ROI on status? ROI on "warm fuzzies"?
Also, this assumes we have good estimates of the ROI on all our options. Where do these estimates come from? In the real world, we often seem to spread our bets - constantly playing a game of multi-armed bandit with concept drift.
Can you explain what you mean by the problem of job training?
You mean job vs. career vs. calling?
If by "job training" you mean maximizing short-run over long-run earnings, I agree with you. But for that reason, if you move the "slider" toward a longer payoff period, then the schools will be incentivized to teach more fundamental skills, not short-term "job training".
On the other hand, sometimes people just need to get their foot in the door to get up and running. As they accumulate savings, on the job experience, professional networks, etc. even a good "first job" can give a lifetime boost.
A lot of people I grew up with have the "cold start" or "failure to launch" problem, where they never get into a good-enough paying job and just spin their wheels as the years go by, never gaining traction. For them even getting a foot in the door will get the ball rolling.
Nearly all education should be funded by income sharing agreements.
E1 = student's expected income without the credential / training (for the next n years).
E2 = student's expected income with the credentia / training (over the next n years). Machine learning can estimate this separately for each student.
C = cost of the program
R = Percent of income above E1 that student must pay back = (E2-E1)/C
Give students a list of majors / courses / coaches / apprenticeships, etc. with an estimate of expected income E2 and rate of repayment R.
Benefits:
- This will seamlessly sort students into programs that actually benefit them.
- Programs that lie or misestimate their own value will be bankrupted (instead of saddling the student with debt). Schools must maximize effectiveness, not merely enrollment (the current model).
- There would be zero financial barriers to entry for poorer students, which is equivalent to Bernie's "free college", except you get nudged toward training that is actually useful instead of easy or entertaining. Also, this could be achieved without raising taxes one iota.
- If "n years" is long, then schools will optimize for lifetime earnings, not just "get a job now". This could incentivize schools to invest in lifelong learning, networking, etc.
Obviously, rich students could still pay out of pocket up front (since they are nearly guaranteed a high income, they might not want to give a percent away).
Easy answer:
(1) Learning via experience imposes costs (rejection, broken heart, etc.); so learning will be slow and tentative.
(2) People who get good enough (via talent or experience) generally exit the dating market (they find a partner). Whoever is left is either still learning (expensive) or a "player" who prefers serial dating / hooking up to a long-term relationship (LTR). There may be a few people gaming the system by having multiple partners (the stereotypical "alpha"), but I doubt this is a significant fraction.
Depending on how things were weighted would really make a big difference. Especially since a lot of these theories make use of (aggregates of) proxies to measure what they really care about as the data you care about is long lost to history.
WRT overfitting, it would not be too hard to measure the error on a holdout set. Turchin et al. essentially used China as their holdout set.
Nobel Laureate in Econ Elinor Ostrom describes how in the real world we have a variety of formal And informal governance structures (that do negotiated decision making, monitoring, conflict resolution, punishments, etc.) to allow both local and global optima. From an info-processing view you know locally what’s best for you, but you need a way of aligning local decisions to reach global optima. Because this is very complex and fuzzy we humans have nested overlapping norms and institutions to govern behavior while allowing freedom and flexibility.
To see culture from a more CS perspective look up papers on “cultural evolution” and cooperation. The books / blog by evolutionary biologist / historian Peter Turchin “War and Peace and War” and “Super Cooperators”. Behavioral research on altruism and costly punishment in repeated prisoners dilemma games also shows the importance and impact of culture.
In this interpretation a “good” culture is one that has more solidarity & honor than back-stabbing & free riding. From a purely economic perspective it creates greater overall welfare and trust.
From an evolutionary perspective “more fit” cultures replace less fit ones, especially via invasion or general external pressure. When there is a lack of sustained external threat to incentivize collective action, cultures often devolve into decadence, back-stabbing and mistrust like Mafioso Sicily.
High social cohesion isn’t necessarily morally “good” though. Catholic Spain was super cohesive when kicking out the Moors. It propelled them into a golden age, but it also turned them into genocidal intolerant maniacs.
A good example of a strongly cohesive culture today is South Korea. They are always under threat of invasion from many sides, so they built a culture of super strong cohesion, hard work, etc.