KnotGodel
4mo ago
The Modelbot that is exactly the kind of thing I'm talking - Sam was exceptionally smart within formal models (epitomized by HFT crypto algorithms) and not really exceptional outside of formal models (e.g. the verbal argument of "what if something goes wrong")
There is some deep epistemological stuff operating here.
If the complaint is that numbers are too "precise", and the solution is to add ranges (whether implicit or explicit), the obvious next question is what these ranges mean.
In formal Bayesian epistemology, there are no "error bars" around probability estimates. The probability estimate is simply the probability you believe X will occur, which can be computed using Bayes formula and a prior probability in simple cases (more complicated cases yield more complicated formulas).
In less Bayesian epistemology, you can recapture much of this using something like proper scoring rules, but, again, a scoring rule only asks for a number - not a range, so it's still unclear what it even means to say "somewhere between 10% and 30%".
In human language, you might want to say something like "I'm saying 20%, but that's because I've consulted the important evidence, but I know there is smaller evidence I'm not considering that might tweak that ±10pp." In this case, you are using error bars to indicate logical uncertainty.
Alternatively, there is a decent analogue in finance: market makers often have both buy and sell limit orders. The spread between them is an indicator of confidence. Note: in finance, the spread is useful only because finance is a fundamentally social endeavor. If you choose to offer a very small spread, what you're really doing is saying that you don't think anyone else can do better than you. [note: this is less true for non-market-makers, who, for binary instruments mostly just care about point estimates].
The financial framing has an obvious betting analogy here: when you say "between 10% and 30%" under this interpretation, you're actually saying is that you would accept a bet that X is true if the odds given were higher than 9:1 and that X is false if the odds were better than 7:3. If we wanted to formalize error bars on probability, this is the model I would advocate for.
[To be a tad pedantic, finance also cares about risk. If I'm market making, I might actually have multiple buy and sell limit orders with different volumes. Likewise, me saying "between 10% and 30%" might mean I'm willing to bet 1¢ at those odds, but it doesn't necessarily imply I'm willing to bet half my 401k.]
KnotGodel
4mo ago
Based on this comment, SBF's blog, and his professional achievements it seems pretty clear that Sam is extremely smart. Specifically, he is very good at manipulating formal systems - math, software, games, etc. He is merely smart is once you leave the world of formal systems.
Unfortunately, in a move common common among STEM-nerd, I'm guessing he realized that all non-formal-systems contain subjectivity, which means you can use them to argue anything, which means they're "bullshit". Indeed, much of this forum is built upon a similar syllogism, but with a more explicitly political lens (e.g. the law is so vague that you can prosecute anyone with selective enforcement). None of this is completely wrong and it is often useful in some contexts, however...
I strongly believe that a large part of a STEM-nerd maturing into an healthy adult is learning
- that there are degrees of subjectivity and objectivity
- that whether a system (formal or informal) is useful for navigating the world is a pretty different question than whether it is objective/true
- how to use multiple systems, both formal and informal, simultaneously to navigate the world
Or, to use Robert Kegan’s model of development: to move from Stage 4 to Stage 5.
Like, take Socrates. Is Socrates the greatest philosopher in the history? That question doesn't have an answer. However, there is value in reading Socrates that puts him above a typical philosopher - namely that understanding Socrates makes it much easier for you to understand a myriad of other philosophers. If you're interested in digesting philosophy as a field, that is valuable. If, on the other hand, you're interested in how philosophy applies to doing the greatest good for the fewest dollars - not so much.
Or take Freud. I assume SBF would say Freud was pseudoscientific bullshit. To be fair, (1) I have yet to find value in some of his writing (particularly on sexuality) and (2) Freud was hardly a beacon of science qua science, and yet... Freud
- popularized the idea that much of our cognition is not conscious
- invented "defense mechanisms" as a concept and cataloged an enormous number of them (e.g. the use of intellectualisation to avoid negative emotions)
both of which are not really "provable", but are self-evidently true/useful.
Likewise, Freud popularized the framing of the Id, Ego, and Superego, which, when stripped of its mysticism essentially boils down to:
People's want to fulfill their desires (Id, Pleasure Principle), but these often conflict with moral/social values (Superego). This conflict, in addition to some desires/values being literally impossible (Reality Principle) introduce significant "tension" in that you can't achieve everything you want, so you have to trade off some of one for the other. Moreover, people use various cognitive tricks to help reduce this tension - e.g. rationalizing that they didn't want money anyway, when their desire (to have money) conflicts with reality (they're poor) or their morals (it's wrong to be greedy).
It is literally impossible to prove the above framework is "true". However, a great number of people find the framing useful.
Anyway, I've gotten off track... ¯_(ツ)_/¯
KnotGodel
4mo ago
I guess it depends how exactly @Recursive_Enlightenment envisions this. Are we getting all of Western Canada, all conservative provinces, or all rural Canada? Still, I think the end result of the geopolitical calculation remains essentially unchanged.
KnotGodel
4mo ago
We have state-wide standardized tests, and almost every high school's average score is public knowledge. Graduation rates are also typically a matter of public record. So, it should be pretty easy for a college to know how impressive a diploma (or specific rank) is from a given school.
No idea if college admissions people do this ¯_(ツ)_/¯
Edit: Idk why you think a national test would add much more value over state tests. There are plenty of proxies you could use to compute average state-smartness
school_smartness = avg_state_smartness + school_smartness_on_state_tests
KnotGodel
4mo ago
I don't think trying to convince an ally's citizens to secede is generally considered a wise geopolitical move. Would you rather
- have Canada as a close ally
- make the Canadian government dislike us for decades (centuries?) but we get the honor of adding a few millions citizens - citizens who probably average half the US average income, which means they will probably be net-government recipients rather than payers.
Seems like an easy choice to me.
KnotGodel
4mo ago
If you’re a Christian, you should probably be doing that anyway? Well, that and charity - rich people have a hard time entering heaven.
Why should the existence of aliens affect that?
KnotGodel
4mo ago
I’m actually having a hard time finding any concrete documentation from the last year. However, the Asian:Black ratio at Caltech is 10. At MIT it’s 4.4; at Harvard it’s 2.4; at Stanford it’s 3.5. So, at the very least it’s far lower in practice.
KnotGodel
4mo ago
Those two officers are in a unique position to speak their own views without political interference from above due to being in an elected role and being retired.
Isn't being in an elected role the very epitome of suffering political interference? The sherrif knows anything he says might be used against him in the next election.
KnotGodel
4mo ago
I think there are two levels to an answer here. The first is to take your framing and dive into the difference between intelligence and perceived intelligence. I think there are two important things here: motivation and legibility.
Plenty of genius-level people just don't spend their time in smart-seeming disciplines. One of the smartest people I knew graduated from college at the age of 18, was an international chess master, and wend to work at a FAANG. Then he left to become a baker. This outlier aside, I suspect that when a man realizes he's good a math/writing/etc, he is disproportionately likely to reorient his life to spend oodles of time on the subject. This brings me to...
Legibility. The guy in class answering all the questions may or may not be the smartest - but he definitely appears the smartest. The guy who aces the test may or may not be the best in industry or in research, but a test score is much more legible than the latter two. Intelligence related to emotions, socialization, and even words are all much less legible than intelligence related to math and coding and engineering. But it is crucial (in life, really) to avoid conflating "harder to measure" with "less important". Also, having the confidence/narcissism to state and defend your beliefs is probably only loosely related to intelligence, but probably strongly related to perceived intelligence.
As an aside intelligence is academically defined as the principle component vector of academic test scores. Do you know what one of the strongest predictors is? Vocabulary size (r=0.83), followed by similarities (r=0.80). Both are much stronger predictors than arithmetic (r=0.68). Yet, a math-smart person is typically considered by people to be obviously smart, while intelligence in other disciplines is less obvious.
The second level is psychological. Why do you care about your partner's intelligence?
Examples:
- You worry about her ability to make money.
- You worry about other people's opinions of your intelligence.
- You worry that your difficulty seeing her as smart indicates a character flaw on your part.
If this were me thinking through this about myself, I would also ask myself why, on some level, I want to believe she is less intelligent. On an internet forum, such a presumptuous question is probably out of place. I will note though, that as someone widely considered a "math wiz" growing up
- It is psychologically comforting to believe that being math-smart is super special awesome
- If you spend thousands of hours doing math... you're going to end up believing that math is Important. The same is true of anything you spend time doing.
Harvard mostly boils down your smartness into their Academics rating as described here:
Summa potential. Genuine scholar; near-perfect scores and grades (in most cases) combined with unusual creativity and possible evidence of original scholarship.
Magna potential: Excellent student with superb grades and mid-to high-700 scores (33+ ACT).
Cum laude potential: Very good student with excellent grades and mid-600 to low-700 scores (29 to 32 ACT).
Near-perfect test scores and grades will only ever get you the second-highest rating. I remember when I was looking at colleges 10 years ago that I noted that Brown only admitted ~25% of people with perfect ACT scores.
When you combine this with now-public data on Harvard's admissions, it becomes pretty clear that, with no change to the ACT/SAT, Harvard could pretty straightforwardly choose the next incoming class to have an average IQ of at least half a standard deviation higher than previous classes.
I think that's the rub: even if the ACT/SAT were redesigned to better discriminate among the top of the distribution, Harvard et al's current behavior makes me pretty skeptical that this would result in smarter people being admitted.
That being said, if you have amazing test scores and grades, you should probably really consider Caltech - they're have no legacy or affirmative action, and they place a huge emphasis on those exact factors.
This is neither here-nor-there, but there is good evidence that top schools under-weigh test scores if their goal is to predict who will be most successful. Who knows to what extent this is because (a) intelligence is super important at accomplishing things or (b) nearly all selective institutions [edit: including med school, law school, FAANG companies, consulting firms, etc] use intelligence filters since they're easy to evaluate - for instance, grit is hard to figure out in a test or interview.
KnotGodel
5mo ago
Also, general question, how do people go about archiving posts here with many reference links that risk bitrot? Recursive wget into an mht?
The Internet Archive Wayback Machine works pretty well for this purpose. There's even an extension that makes it pretty easy to archive specific pages that you worry won't be archived due to lack of visibility.
Alternatively, just use File > Save Page As...
Alternatively, alternatively, it's not that challenging to use selenium-wire to intercept every HTTP request a dummy browser makes if you really want to automate saving a bunch of webpages, and very few websites have good defense against that level of scraping (at least, if you keep you QPS low enough).
KnotGodel
5mo ago
I think the focus on censorship is misplaced. Censorship is one of the many ways a Culture War can play out, and it is neither necessary nor sufficient for showing a Culture War is occurring. Consider, for instance, a racial minority marching for civil rights. It doesn't matter whether there is censorship - that is quintessentially culture war. Conversely, consider censoring how to make nuclear weapons - that's not a meaningful component of any significant Culture War.
A culture war happens if at least one side decides that the other side is so wrong/dangerous that it needs to be converted
I agree with this. A Culture War is created by two groups of people attaching so much value to a dichotomy that converting the other seems important. So, necessary conditions for a Culture War are
- A dichotomy [ Edit: or two? I remain agnostic on the extent to which each group can have separate conceptualizations of the dichotomy ]
- Value placed on the dichotomy enough to prompt both ends to attempt convincing others
KnotGodel
5mo ago
Ah whatever, we’re all dullards.
I appreciate your patiences.
Why is the poll tax wedged in that parenthesis?
Generally, in the optimal tax literature, it is assumed that utility is a function of consumption rather than income. Consumption (in this simple model) is simply your post-tax income, which is
wage * labor - poll_tax
That's why poll_tax is in the parentheses.
Re your model, I agree with your math, but I think my model better matches reality, because it assumes that one's 1,000,001st dollar provides you less happiness than one's 1,001st dollar, while your model assumes they are equally important.
And, sure, using a logarithm is itself somewhat arbitrary, but the result should be pretty stable. For instance, let
- f be any continuous function such that f' > 0 and f'' < 0
- g be any continuous function such that g' > 0 and g'' > 0 [note in this case, we assume the second derivative is positive, because there are increasing costs to labor, which is equivalent to saying there are decreasing benefits to free time]
- let utility be given by
U = f(W * L - poll_tax) - g(L)
We can prove that any poll tax causes equilibrium labor to increase in this model.
KnotGodel
5mo ago
Sorry - my brain farts have been real during this exchange, and I honestly wouldn't blame you for just concluding I'm a dullard and not worth your time - if it quacks like a duck...
I believe capital gains taxes cause people to move more of their wealth into risky assets, so I believe they effectively increase people's risk tolerance.
KnotGodel
5mo ago
The Culture War? I guess something like the social activity related to emotionally-charged dichotomies?
That’s what happens to half the issues. The other half get resolved so thoroughly that everyone in the past looks completely evil - slavery, Nazism, etc
there will always people who want to be free, and there will always be people who want to censor and control them.
That is the libertarian dichotomy.
An Effective Altruist would say
there will always be people who want to help others, and there will always people who want to ignore them or merely feel good about themselves
The Christian wold say
there will always be people who believe in Jesus, and there will always be people who reject him for a life of sin and pleasure
The Scientist would say
there will always be people who pursue truth, and there will always be people who cling to dogma
The SJW would say
there will always be people who stand up for the weak, and there will always be people trying to oppress them
etc. etc.
The Culture War is not simply the dichotomy of the free versus the controlling. That is simply the dichotomy people on this forum tend to favor.
KnotGodel
5mo ago
Another misunderstanding
This whole section is my bad. My reading comprehension clearly needs work. Mea culpa.
KnotGodel
5mo ago
So your math was incorrect, but luckily you interpreted it wrong, so by being doubly wrong you went all the way to being right again?
No. I did the math on a piece of paper correctly and copied it wrong.
And what happened to the k’s now, why did you remove them from the denominator?
Because a wealth tax doesn't affect the variance of post-tax returns.
By that logic, and assuming the equations are worth a damn, the wealth tax also causes increased risk tolerance.
I've been arguing this whole time that
- A wealth tax doesn't affect risk tolerance
- A capital gains tax reduces risk tolerance
So it does, indeed, follow that, relative to a capital gains tax, a wealth tax increases risk tolerance.
Any tax in this model is distortionary.
Incorrect. A flat tax (without any poll tax or welfare) is non-distortionary in that model.
Edit:
U = ln((1 - tax_rate) * wage * labor) - labor
U = ln(1 - tax_rate) + ln(wage) + ln(labor) - labor
dU/dLabor = 1/labor - 1
so labor = 1
all independent of tax_rate
KnotGodel
5mo ago
It does. He then gets to deduct those losses on his taxes, resulting in recouping 99% of his lost value.
KnotGodel
5mo ago
I think that discredits the model. They’re not going to take all the equity risk for 1% of the equity premium, they’re supposed to be loss averse.
I never responded to this. The point you're missing is that a 99% capital gains tax also reduces the risk. An investment that returns 10±30% now returns 0.1%±0.3%.
I want to also take this opportunity to wax poetic on financial markets and risk.
Most economic risk in society has nothing to do with financial markets - think things like
- risk regarding whether a particular company will become more or less profitable
- risk regarding commodity / mineral prices
- home price risk due to the natural housing market
etc.
Finance, as a market, serves two purposes:
- It connects money-havers with money-needers
- It connects people who are poorly positioned to take a risk with those better positioned to take that risk
So, for instance,
- a wheat farmer is poorly positioned to accept the risk that the price of wheat will crash, so they buy risk protection via options or futures from investors (ditto for mining operations)
- a startup founder takes enormous risk starting a company - a venture capitalist takes some of that risk by allowing the founder to, say, draw an income in the meantime
- a family needs a house and will probably be able to afford it (the future is never certain! maybe all earners will be paralyzed in a car accident) - the bank assumes some of that risk when they lend the family money via a mortgage
Why are the investors better positioned to take these risk? Two reasons: they have money (duh) and they can diversify by buying a basket of somewhat independent risk.
So, when functioning properly, financial markets reduce the cost of risk to society.
That being said, "functioning properly" is key and, due to the principal–agent problem and plain old human error, there are significant sources of imperfection.
Much of that imperfection probably pushes investors to take too much risk: generally speaking if a trader does well, they make oodles of cash, while if they do poorly the worse they can be is fired. Note, however, that absent a specific scenario (see next paragraph), that cost of that risk is still paid by the company funding the trader, so it is internalized to the company and we, as the broader society, don't really need to care too much about it. When a trading firm's net worth hits zero and it closes, we still don't really care - the costs of that firms' sins were laid at that firm's feet.
The specific exception to this is when a company's net worth hits negative, which can only happen if a firm takes leverage. Even in leveraged firms, this is not too common, since as any firm worth its salt will start winding down positions automatically as its overall wealth goes towards zero (the specific terms are "funding risk" and "unwinding"). Robinhood, for instance, will do this for you automatically (or, rather, force you to do it).
Even if a firm's net worth hits negative, someone was lending the firm money and that firm bears the costs, so, again, you and me (as outsiders) shouldn't really care. So, the actual exception is a firm that
- uses leverage
- does not use unwind strategies (either because they're incompetent or because the market is highly illiquid like housing)
- poses a systematic risk and so gets bailed out at society's cost
This gives us the three things to target in seeking solutions:
- discourage leverage through regulation and taxes (e.g. don't allow the interest rate deduction)
- legally require the implementation of unwind strategies (I don't remember if this is generally legally required, but is implied for banks via reserve requirements)
- have stringent requirements for firms large enough to impose a systematic risk (e.g. the Volcker Rule)
But none of these imply a wealth tax or capital gains tax. Neither tax policy discourage leverage (borrowed money is deducted from wealth and, at least today, margin interest is tax deductible). Neither have any bearing on unwind strategies or target firms that impose systematic risk.
I’m talking more about boardgames and card games.
Lol is technically a formal system (ignoring lag), but, at SBF’s level (bronze) skill is largely determined by “softer” skills like last-hit-clicking minions.
More options
Context Copy link