The Brahmins

The cognitive and personality profile, and overall achievement package, of Indians as a group is a rather interestingly unbalanced one. Sometimes they do spectacular things, like discovering the infinite series for trigonometric functions of sine, cosine, tangent, and arctangent as early as the 14th century, producing a good number of real geniuses like Ramanujan and Satyendra Nath Bose, and reaching Mars orbit on its first attempt, being the first Asian nation to do so, and doing so at a small fraction of the cost expended by NASA. An IMO gold medalist I talk to once said to me that there are probably more Indians than Chinese with IQ 160+ due to very high Brahmin IQ that has stabilized (meaning regression to a stable high Brahmin mean as opposed to the low Indian mean) over millennia of inbreeding within caste. I thought maybe. Certainly, I do sometimes get the impression that Indians, at least in science, are better than Chinese at breeding the type of genius with the right combination of technical ability and scientific discernment that manages to discover radically deep and groundbreaking science in a very independent and spectacular fashion. The Chinese have produced geniuses of the highest order (or close) in science the 20th century, like Chen Ning Yang in theoretical physics and Shing-Shen Chern in pure math, with Yang-Mills and Chern classes ubiquitous now in the literature of their respective fields, which are now very intertwined. However, they did so only after much training, exposure, and reinforcement based on the whole framework of modern science developed in the West over many centuries, and ancient China, on the other hand, did not produce in pure science anything near what Indians did, a sign of lack of genius and of poor taste, both in its rare individuals and at the collective societal level. On this, I like to think that Indians are Greeks and Chinese are Romans.

In sharp contrast to China, India in practical matters has been largely a complete fuckup, or at least vastly outmatched by China. It is well known that the ancient Chinese invented gunpowder and paper-making, whereas nothing of equal direct impact came out of ancient India. In modern times, China developed nuclear weapons way faster than India did, and even before that, defeated India in a war in 1962, which, even worse for India, was entirely her fault. Economically and infrastructurally, holistically speaking, India, exemplified by its frequent power outages and accident-prone train system, could be regarded as a few decades behind China, which is further confirmed by that India’s life expectancy and infant mortality rate is, today, where China had been at 20+ years ago. Given that the two had been around the same level in 1950, India’s development has unambiguously been a complete failure.

How to explain this? On this, I recall how my Chinese friend’s mom had said that it’s not because China’s elite is smarter than India’s elite, but because China’s grassroots is smarter than India’s grassroots. This is well-confirmed by international IQ studies which tend to put China’s average IQ at around 105 and India’s around 82, which is a 1.5 sigma difference. So even if India’s +3 sigma is as smart or smarter than China’s +3 sigma, there are too many dumb, dysfunctional people holding India back, from their needing to be fed while doing the routine work rather poorly. So, the smart, (usually) high caste Indians opt to go to America to escape India’s dysfunction, so ubiquitous that even the ultra-rich at home cannot immune itself. The best and brightest in that category tend to go through the IITs at home for undergrad, the most reliable ticket to a high paying tech job in the United States. That stratum of Indians has established by now quite a presence in top American tech companies and universities (just about every top STEM academic department in the US has several prominent Indian profs). For example, Microsoft and Google both have Indian CEOs, and plenty of Indian engineers and managers, with many of them in high ranking positions, especially at Google. In contrast, there are few Chinese in top leadership positions. When I learned that Google has several Indian SVPs but no Chinese, a guy from China responded with humorous ease followed by sarcastic insult: “不用担心,阿里巴巴的SVP全都是中国人,百度的SVP也全都是中国人,没有一个印度人。(In translation: Don’t worry, Alibaba’s SVPs are all Chinese, Baidu’s SVPs are all Chinese, not a single Indian) What does India have? Tata? Infosys?” This is, based on my experience, similar to how people react to the astronomical success of certain Indian academics, entrepreneurs, and business leaders in America. They will say: “Sure, an individual brilliant Indian does extremely well in America. But what does India as a nation get from that?”

Even such brilliance of these elite Indians is somewhat questionable. On TopCoder, which plenty of Indians obsess over on Quora, now infested by low status Indians, India is ranked, as I am currently writing this, only 11th out of the 31 countries on there, with only two red (the highest category) coders, despite having more than twice the number of members as China, the second most populous nation in this algorithmic coding contest. They’ve actually done better in recent years. I remember back years ago when I participated, I, having been on the lower side of yellow (the second highest category) coder, would have ranked close to the top among the Indians. Of course, one must not discount the possibility that the best Indians have better things to do than practice for a contest where one solves artificial algorithmic problems, which is consistent with my having seen and worked with many Indians who are very competent at real software engineering, with quite a strong sense for systems design and real world production code, which are rather orthogonal to, and much more consequential than, what one sees in those contrived coding contests and interviews. Still, the dismally low performance of Indians on TopCoder still raises suspicions, because TopCoder, like the International Math Olympiad, which India is complete garbage at, is a 100% objective and fair contest, whereas success in the real world software engineering, determined by promotions and professional level, has a political and context component. It’s not just the Indians at home; even in America, where the smartest Indians tend to go, the Chinese kids beat the Indian kids by a wide margin on the elite math, computing, and physics olympiads, even when the Indian kids seem to have improved a fair bit over the recent years. From this, one can only conclude that Indians are naturally not that strong in the abilities which these contests load on, though of course they may be relatively much more talented in research and engineering, for which these contests are very imperfect predictors.

You, the reader, have probably noticed that up to now, we’ve focused mostly on brains and technical ability. Yes, they are essential, but personality characteristics (both individual and collective) and “soft skills” also matter, especially if one wants to rise to a leadership position. From my personal observation, Indians are, in general, very good at projecting confidence and assertiveness from the way the talk and present themselves, much better than Chinese are, at least in the American cultural context, even when you discount the language barrier Chinese face relative to Indians. I’m talking not only about how one says things in terms of word choice, but the vocal tone and body language behind it. Sure, you can disdain this as superficial, but it matters. Perception matters as much, and in some cases, more, than substance. There is also that Indians seem to have a stronger network and help each out more in the career world. Collective intelligence or ethnic nepotism, you be the judge.

I have stories to tell on this. First of all, I remember vividly how when I interned at the place as an Indian schoolmate, he was the only one who scheduled, successfully in a few cases, coffee meetings with executives, as an intern (!!!!!), when it never would have occurred to me, or probably almost everyone else except him, to even try. One can sort of link this to collective intelligence, in that it is an indicator of discernment with regard to who matters (the executives) and who doesn’t (the engineer worker bees) within the political organization. And needless to say, you rise up in the organization by aligning yourself with the people who matter. Yes, my telling a full-time engineer this was met largely with a response in the likes of: “He knows who matters and who doesn’t. And even if he completely fucks up, he has nothing to lose, he’s only a 2nd year college intern. In any case, he gets good practice interacting with people who matter.” There is also that multiple people I know have complained about blatant Indian favoritism in interviews in the likes of what is described in this Quora answer. Yes, others have told me that when Indians interview other Indians, the bar is much lower. It’s not just in interviews. Another guy told me about how he once worked for a company that turned into ruins after Indian managers protected some Indian fuckups from getting fired. Personally, I have seen a case of Indians getting promoted way faster than those of other ethnic groups on a big team with an Indian director. So sometimes, I ask myself the verboten. Could it be that Indians really are far higher ranked in tech companies than their ability and contribution, because they are much more self-promoting and collectively nepotistic than those of other groups? Moreover, could it be that many people secretly think and resent this but are too afraid to say out of fear of being publicly vilified for “being racist” and having their careers ruined from alienating a national group increasingly powerful in corporate America? And that gradually, other groups, as they awake to the rigging of the game and get past, reluctantly, their moral objections, will quietly do the same, transforming tech companies and the American workplace at large into literal prison gangs contend, destroying whatever is left of the ideal of meritocracy and fair play in this country, ever more mired in identity politics?

Don’t get me wrong. There is much variance in personality and character and ability in those of any ethnic group, including for Indians, and much overlap between ethnic groups. Like, I know of this really brilliant Indian who donates most of his tech salary to very worthy causes, leaving little for himself, and he would be the last person I would expect, based on his characterized as autistic personality, to successfully climb the corporate ladder, though through sheer talent alone, he should do just fine in the appropriate position. Moreover, I have interacted with several Indians who had been very kind, tolerant, and helpful towards me. However, averages can differ by a standard deviation or more, with enormous social consequences.

I actually feel somewhat sympathetic for India and the Indians here. Somebody, on this, even said something along the lines of: “India is just such a shitty place that the Indians here have nothing to lose, so they play dirty political games and engage in the most spineless social climbing.” What can be done to resolve this? Immediately, I cannot think of anything other than drastically reducing the number of abjectly impoverished, low IQ Indians in India by simultaneously improving economic conditions and enforcing birth control on the poor and unable, so that less suffering and dysfunction is spread to the next generation. India could, instead of drinking the democracy Kool-Aid, learn from China, in a way compatible to its own culture and circumstances, just as China did from the West and the Soviet Union, to great success. Its elite needs to correct many of its deeply flawed social attitudes, and not only that, actually act accordingly with full force; otherwise, the excessive damage India does to itself, America, and the world at large with its internal dysfunction and exported corruption will always far outweigh what its elites contribute to science and technology. I can’t be optimistic on this though, barring some really radical change.

Tech industry, an interview question, and tail recursion

I have written on here before that I sort of disliked the tech industry. Why? Because I felt many of the people there are kind of boring and not that smart, and much of the work is quite mundane, though of course there are some extremely good ones who do the bulk of the technical heavy lifting (I’m not, though maybe I could become one), who are grossly under compensated relative to their actual contribution. Of course, my standards must be way too high, or I must be way too weird or non-conformist, or too spoiled. At the very least, the tech industry pays quite well, especially the big companies which offer bonus and equity. Of course, plenty of 150+ IQ people will go into grad school in math or physics or computer science, doing some much more academically involved work, often with contempt for the intellectual lightweights in the tech industry. I plead guilty to having had that sort of attitude as well, and maybe I still do. Related to that is how I found the whole artificial marketing and inflation of achievement in tech kind of disingenuous. However, I’ve figured out by now that one only has much to lose from not playing along in that game. I’ve been paying more attention to LinkedIn recently. It’s literally a credentialist cesspool of professional posturing, full of mediocrities who put on there literally every detail of their professional and extracurricular life. My having become more accepting of that indicates somewhat that I’ve improved attitude-wise. I feel like I talk to some non-techs too now, in a normal way, without expressing any sign of contempt, because what’s the point? My next step would probably be to shut down this socially unacceptably nerdy and elitist and non-PC blog, but unfortunately, I don’t feel comfortable dulling myself out like that. Of course, it might just be that the whole career game more or less compels me to do so sooner or later. When I say this, I have in mind the following from Michael O Church’s essay Does Genius Exist:

Most gifted children seem like they might achieve creative excellence in adulthood; very few actually do. I’ve observed the careers of extremely intelligent (i.e., IQ 160+) people and the results are, at best, disappointing. About half go to graduate school; the other half go to Wall Street or Silicon Valley straight out of college. Either way, they expect to defeat the morons in business handily, retire within ten years, and dedicate the remainders of their lives to intellectual pursuits. It almost never works out that way. It’s not uncommon for highly intelligent people to be mobbed and bullied in their corporate jobs by resentful mediocrities, although even more common is for them to disappear into the bland, beige fog, and to lose every element of originality they once had. Most often, they disappear somewhere in the folds of middle management, and do what they can to hide what they once were.

I already feel more comfortable doing what, according to this, is most often done by the gifted later in life, not that I am +4 sigma above the mean, which is evident from my credentials, though +3 sigma sounds about right. Surely, success in the corporate world relies much on being liked by those in power which requires being conformist, loyal (or at least appearing so), dependable, and not threatening to the interests above. You get promoted by becoming the manager’s favorite, which is done by being the one who supports the career of the manager the most in an indispensable way.

I don’t like much the whole interview process in tech. It’s like the problems are so trivial (they are artificial and all that related to real engineering) and some of the interviewers are nowhere near as smart as me, IQ-wise. Well, it doesn’t matter, because one has to adapt to one’s world instead of the other way round. And like it or not, for those on the tail end, the distribution of IQ/ability in the world is what it is today.

Speaking of tech interviews, I was asked this question in a recent interview.

Flatten a list. The list, of course, can have list elements. So something like
flatten([1,2,3,[10,30]]) => [1,2,3,10,30]

I had done this problem before. There is the brute force recursion solution. In Python, which is a dynamically typed language (which is more or less necessary for this problem (because otherwise, one would have to impose some type constraints and on top of that find a way to make nested lists work within a static type system, which as far as I can tell, would require defining some generic sum type of primitive types and the list type of that generic sum type itself), this would be, in functional style

def flatten(l):
  return [l] if type(l) != list else sum(map(flatten, l), [])

Of course, in the actual interview, I wrote it imperative style, to increase my chance of passing it. 😉

This is of course actually inefficient, in that there will be a copy made at each level of the recursion. To avoid that, we employ a helper function.

def traverse(acc, l):
  if type(l) != list:
    for e in l:
      traverse(acc, e)

The flatten function itself would be

def flatten(l):
  acc = []
  traverse(acc, l)
  return acc

By essentially preallocating the space for the output flattened list and appending to it via traversal, we use constant memory aside from the linear for the returned flattened list itself.

What did this problem remind me of? Tail recursion. Functional languages support it to avoid adding a new stack frame on each recursive call, which would be very handy for this problem. In fact, the traverse function, in its pattern of implementation can be translated to a tail recursive one in a functional language. We’ll leave that for later.

To start, we’ll present a canonical example of tail recursion, the factorial function. In Haskell, the immediate implementation it would be

factorial :: Int -> Int
factorial n = if n <= 0 then 1 else n * factorial (n-1)

In assembly, we would have something like

        # create new stack frame
        pushq   %rbp
        movq    %rsp, %rbp
        subq    $16, %rsp
        # copy parameter to stack
        movl    %edi, -4(%rbp)
        # compare parameter with 0
        cmpl    $0, -4(%rbp)
        # the recursive case
        jg      .L2
        # the base case, set 1 as return value and return
        movl    $1, %eax
        jmp     .L3
        # set argument in recursive call to current argument minus one
        movl    -4(%rbp), %eax
        subl    $1, %eax
        movl    %eax, %edi
        call    _Z9factoriali
        # set return value to n * factorial(n-1)
        imull   -4(%rbp), %eax

One can see explicitly in the assembly the adding of a new stack frame for the recursive call. To avoid that, we employ tail recursion as follows.

factorialTail :: Int -> Int -> Int
factorialTail acc n = if n <= 0 then acc
                      else factorialTail (acc*n) (n-1)
factorial = factorialTail 1

In assembly, this would be

        # move acc to %edi
        movl    %edi, %eax
        # bitwise AND of %esi with %esi itself is %esi, set SF, ZF, PF flags accordingly
        testl   %esi, %esi
        # return if sign flag (SF) or zero flag (ZF) are on
        jle     .L5
        # acc *= n
        imull   %esi, %eax
        # n -= 1, subl also sets zero flag 
        subl    $1, %esi
        # loop back if zero flag is not set, return once n == 0
        jne     .L2

Notice how this is essentially a for loop, with no recursive calls. With the -O2 flag set on x86-64 gcc 8.1, which I used to generate the above assembly code, with my own comments later added, the tail recursion compiler optimization was implemented, as evidenced by the assembly produced. I ran this not on my own machine, but on the cloud via the handy compiler explorer I found, the code of which happens to be on GitHub. And I can only say that the guy who created this tool looks like another one of those uber prolific programmers blessed with tremendous instinct and power for building software systems. You might think that my writing a blog post with Haskell and x86 leans me towards this category as well. Oppositely, I actually think that I’m quite pathetically weak at computer stuff (and started off very unnatural at it), though I also believe that with some dedicated practice I can become good. I would say that I was natural with mathematics and algorithms but not with engineering or systems, though surely, with quite a lot of exposure, I developed, slowly, a sense for the latter as well, gradually steering what had been a horrendously off intuition towards the right direction, and concurrently, reducing, stage by stage, the sense of awe and intimidation I had felt with respect to the actual natural hackers. I can at least console myself by thinking that much of my awe’s having transformed into some sense of normalized (mentally) understanding is an indicator of rapid progress. Yes, there are still plenty of people way better than I am, but I no longer feel like what they are doing, their thought progress, cannot even be understood by my weakling brain, that once perceived it as some form of otherworldly wizardry beyond my comprehension, and of course, its actor some form of higher being.

On quite another domain, I felt somewhat similarly with regard to those at the top of the socioeconomic and political hierarchy. The default for corporate executives and those officially at the top is one of reverence. People assume that because they are on the top, they must be inherently superior in some way or another in their ability. Programmers, as status-insensitive, socially clueless aspies, are supposed to be largely oblivious to the political machinations orchestrated by those on top, to what is the reality of their (subordinate) position within the whole hierarchy. In any case, those people felt to me to be in a whole other world, similar to the impression I had of these elite programmers; I was much oblivious to that world and also could care less about it. Until I more or less developed, as far as I tell, a more accurate intuition for how that world works as well, much aided, of course, by experience, mostly indirect, but enough for me to, with my intelligence and independent judgment, construct what I believe is a reasonable picture or model for what actually goes on, one which I expect to be enhanced over time with more data collected. I’ve increasingly grown to realize, over time, that people are very much naturally psychologically chained to the reality of their official position, their formal credentials, which are correlated very imperfectly with actual ability, or in some cases loaded on (largely born) social position and artificial perception, even nil or negatively correlated; it takes an independent mind, a revolutionary mind to break free from that by disentangling the real and the artificial. And becoming mentally free is the first step towards becoming actually free.

On credentialism and selection systems

I’ve mentioned before that an Asian-American friend of mine, who is quite smart, disapproves of the whole campaign against Asian quotas spearheaded, or at least advocated, by Steve Hsu and others.

His words are the following:

  1. I don’t believe in legitimizing the credentialist culture of modern academia
  2. I don’t generically feel much kinship with Asian-Americans (who are the most affected by purported discrimination in admissions), even if I might feel more kinship with them on average than I would with any other large ethnic group in America (which is itself not necessarily true)
  3. I don’t find it implausible that there are legitimate reasons to discriminate against Asian-Americans in the admissions process, if by ‘discriminate’ we mean ‘weigh their formal accomplishments less than one would for a member of a different race’
  4. At the end of the line, I believe that persistent whining about this is a reflection of emotional immaturity on the part of Steve et al., in that they seem to have a ‘chip on their shoulder’ which they are incapable of overcoming, and if they were actually taking a principled approach, they would come together and try to create a superior alternative to the radically broken university system, which will likely not be saved by any infusion of Asian students

Here’s what I think.

On 1), I don’t like the credentialism culture of modern academia either. Much of it is a superficial and soulless arms race. Not that grades, test scores, publications, citations, impact factor aren’t strong signals but they are prone to manipulation and artificial inflation, and that there are qualities of work not well-captured by those metrics. People are more or less compelled to single-mindedly play this game, often at the expense of actually substantial scholarship, if they are to survive in academia nowadays.

On 2), I hate to say that this country has become more toxically consumed by identity politics over the years, not to mention that people are judged at least subconsciously by who one is associated with. So collective bargaining is crucial for a group’s position on the status hierarchy.

On 3), there is that due to Asian-Americans’ and Asians in general having traditionally been the underdog, as well as their lack of media presence, which is intimately tied to the alienness of their names in the Western linguistic context, some people are inclined to view Asians are grinds who aren’t actually as capable as they might appear on paper. Especially with the whole tiger mother phenomenon that Amy Chua popularized with her infamous book. Of course, China’s rise over the recent years has altered this perception somewhat, especially the one that Asians are smart but not creative, though surely, it does seem that controlling for grades and test scores, or IQ, Asians do seem less creative, though that may be due to environmental factors, such as de facto or implicit quotas imposed by diversity mandates and economic circumstances.

On 4), I mostly disagree. Asian-Americans don’t really have the power to create a sufficiently credible alternative in a world that runs so heavily on associating with prestigious, usually long-established, institutions like Harvard and Goldman-Sachs. In their ancestral countries, China and India, Asians can improve the university and research system and the economic and technological competitiveness of the country as a whole, so as to make their universities more credible as well. In America, all Asian-Americans can really do is make more noise around the issue to exert more pressure on the elite universities, and also donate more and enhance their media and political presence as their socioeconomic position improves, especially at the elite end, improves, so that the elite universities perceive themselves as having more to lose from discriminating against Asian-Americans based on race.

This is all I have to say as pertains exclusively to Asian-Americans. I shall now give my thoughts on credentialism and selection in general.

The job of admissions and hiring committees and HR is astronomically harder than in the pre-internet age. So many people apply for positions they are grossly under-qualified for, now that it’s so easy to shoot off a resume or application online. There are, of course, application fees for college and grad schools, but they are not enough to deter. This means in the selection process can be afford now significantly less time per candidate, and one can argue that as a consequence, the process becomes more bureaucratic and easier to game. Often, people will in the pre-screening stage eliminate all applicants who do not meet certain formal criteria, such as minimum GPA/test scores or a certain degree from a certain set of sufficiently credible universities. In the case of academia, to my limited second-hand knowledge, committees will look at publications lists with a focus on citation count and impact factor of the journals on which the papers were published and also verify the candidate against senior, tenured faculty in the same or at least similar area of research. In the case of industry jobs, what matters more is the interview, where for technical roles, technical questions will be asked to further test the technical aptitude and knowledge, as well as, the softer aspects of communication and personal chemistry. For non-technicals, I can only say it’s even more about credentials (school, companies, job titles, dates of employment) and how you present yourself. I can only conclude that way more energy is expended now in aggregate on application and selection than before, which is quite costly really. In the career world, people are mostly out for themselves and don’t really care about wasting other people’s time, so long as they can get away with it with impunity more or less.

I’ll say that there is a tradeoff between optimizing for one’s formal credentials and optimizing for one’s actual ability and knowledge. One loses out so much more now if one neglects the former too much due to more competition per position. Surely, there has been gross inflation of credentials. This is in its crudest form epitomized by college’s having become the new high school, thereby rendering prestige of institution a stronger signal. Furthermore, the largely consequent grade inflation and watering down of coursework has added more noise to school transcripts. Contest training, for math in particular, has become so much more popularized, that to not have credentials in those raises questions in some circles, and moreover, there is so much more of an obstacle course of summer programs and scholarships and grants and internships and jobs which one must pass through to some degree if one wants a reasonable chance of success at a specified level. In this sense, there is more pressure to conform to an existing, often complexity-ridden system. It may well be that people nowadays are not all that much better in terms of knowledge and proficiency than before, correcting for the positive effects of technology on learning, but they actually put in much more time and effort.

Now, if one expends much energy on actual substance, there is concern as to what would be lost if those translate not into formal credentials. Arguably more common is the other way round, where one turns into a soulless credential-chasing machine. I’ve been amazed at how many people manage to achieve much higher grades, test scores, and awards than what their knowledge and ability from interaction with them would reasonably indicate. Those people tend to be very boring and risk-averse, and they are often the types our current system selects for, like it or not.

I used to feel like to prove that one is actually smart, at least in STEM, one ought to do sufficiently well in one of those major math, physics, or computing olympiads or contests. I would say that for raw technical ability, that is probably still the strongest signal. Grades are somewhat noisy, because it’s not hard to copy or snipe homework solutions, and for tests, there is a large cramming and figuring out what’s gonna be on the test component. Perhaps they are more consequentially so as there are also some genuinely capable or even brilliant students who for related personality reasons have a hard time getting themselves to care too much about grades. I’ve personally seen some high GPA people, even in college, who signal in what they say or write complete idiocy that would make you wonder if they were pretending stupid, especially if said person were female. Some people learn much more deeply and also much more broadly, outside of what the system teaches them, to a high level of retention, much of which is not captured through any formal credential. From my personal experience, tests of a wide range of knowledge, sufficiently substantial but not too esoteric, are stronger signals since they cannot be crammed for, but they are, for the difficulty of organization, seldom administered.

In the real world and in academia though, what matters is the ability to deliver actual projects and conduct meaningful research, and those, while correlated with ability to learn, are not quite the same. Those are also way more context-dependent, which means more noise, both due to more variance and more ambiguity of judgment.

I will say that at times or even often, society is met with the problem of people finagling themselves into a position to judge what they are not really qualified to, per their ability and expertise, which means some resume-padding bozos rising up and actual competents being passed over. This problem I believe has been accentuated by the ever more credentialist culture that has emerged over the recent years. What’s kind of sad is how the more conformism and risk-aversion rises, the more these traits are pressured and selected for.

I’ve come to notice that there tends to be some difference between maverick genius and the conformist first-rate professional. If one looks at history, real genius, the ones who create paradigm shifts, tends to have more very lopsided profiles, though surely, it might go too far to say that *most* of the real geniuses were out of it in a Stallman or Galois like fashion, especially as it’s the deranged ones which garner more attention. But one can say with high level of confidence that there were many real geniuses who had a hard time fitting in even into the elite mainstream of his profession, who have even been marginalized. I’ve been told that the real genius mathematicians like Perelman, Langlands, and Shimura more or less cut contact with the mathematical community apparently out of disgust. There is also evidence that plenty would-be real geniuses did not actually make it, with their enormous potential having been thwarted by the system at some point and hardly realized. In an ideal world that optimizes for collective value, if somebody else can do the job much better than you and actually really wants to, you should let him do the job and get out of his way. Of course, reality is far from that. I have personally felt that way with regard to my mathematical ability, often feeling that I wasn’t good enough when I failed to derive something on my own, yet I see so many people worse than I am even so eager to play the whole credentialist game without recognizing how deficient they really are. This suggests that I am very partial towards a certain side of the spectrum. I even feel that in some sense, nothing is more embarrassing then formally being much higher than what one’s ability actually merits, since it demonstrates not only incompetence but poor character. However, I am, regrettably, or not, feeling that circumstances are pressuring me ever more towards the opposite direction.