>>27805
Somehow I doubt that TC users are very picturesque...

>>27805
I worry about what I look like a lot. Probably not very hikki of me, but I can't help it. Even if it's just for myself, I want to at least be something that isn't disgusting to see. Although, I see beauty in a lot of things that other people don't. I personally am a twig, but I see the stereotypical overweight neckbeard as creature of beauty if only because it represents and ideal for me and is distinct in it's individuality. Maybe a monster from a horror movie isn't beautiful, but it has character. That is what I want.

It's reasonably important to me. It makes me feel like a shounen protagonist and also makes me feel better about myself.

>>27805
why does that line mean a lot to you?
is being attractive inherent to your value?
must suck if it is

I don't really think about myself much. It isn't important, but I'm glad I'm at the very least not too ugly on the off chance I see a mirror.

>>27809

I don't know when it started, but I don't feel deserving of love if I'm not feeling beautiful.
It does sucks and I'm trying to accept how I look but it's a slow process.

>>27809
People will manipulate their interpretation of value to be something that conforms to themselves and gives them value.
I think in general their are 4 measures of value.

Intelligence/knowledge- This is good because it's hard to actually quantify or measure, even IQ is not great for it. Most losers think they are intelligent even when they most often are idiots but they can always tell themselves that they are intelligent and unlike attractiveness it cannot be instantly disproven by looking into a mirror. Even if they do something stupid they can shift the interpretation of what is stupid or intelligent to ignore that or they can make excuses for it.
This is why people argue so much on the internet. Their very identity, what it is that makes them feel like they have value is tied to them being smart or knowing about things.

Wealth- Wealth is physical, it can be measured but it also can be displayed, people can display wealth even when they are in fact poor. Wealth can be equated to success and value in a society, it can even imply intelligence, money can get you things, wealth is power.

Morality- Like intelligence this is good because it is also hard to quantify and measure and also is based on what the individual in question deems is moral and good. Morality can also be brought yet at the same time somebody that lacks wealth could justify that by claiming it makes them more moral.

Attractiveness- Attractiveness is again physical, to a degree one can buy it or lie about it but only to a degree, they can try to frame it to suit them but that is hollow. What you look like is a reflection of what you are as a person as well.

What should be inherent to ones value? Wealth and attractiveness are really the only things that can be objective and that people cannot make up to suit themselves. I am not saying the other two are not important values but people can make up whatever they want regarding them.

>>27813
But if someone is intelligent, talented, or knowledgeable wouldn't they be able to display that and receive feedback?

>>27814
Yes but it's something that the individual in question can choose the manner in which it is displayed and dictate what they feel is intelligent or not.
People can focus on certain fields and become knowledgable or talented in that field or at least appear that way to others who may not be, I could read lots of books about the Napoleonic wars and when I talk to my mum about it that makes me look smart because she has no idea about any of this. Though I may read a lot about this but actually understand little about it, it's just that most people don't have any knowledge on it and could not challenge me on it, or maybe I don't read at all I just watch Youtube videos on it so I don't actually have any proper knowledge in the first place but still enough to appear like I do to people with no understanding of a topic.

And also I can decide whether what I say is correct or intelligent or not. Nobody ever wins arguments on the internet, I can always feel like I am right and there is no real way to challenge that, even if multiple people are saying I am wrong well that just means that I am smarter than multiple people, they all must be wrong because I am a genius right?

This can't really be done with wealth, a million dollars is a million dollars, you can't really shift or decide goal posts. With looks it might be done to a degree as some people might find some looks more attractive than others but there are commonalities within that and taste is fairly easy to recognise for being taste. I am not going to feel ugly because somebody finds dark hair to be unattractive or is only attracted to dark hair, that's personal taste.

>>27815
A person can also demonstrate their value through their works. I think a person who creates things is more valuable than a person who's knowledgeable in some niche.

>>27816
Yes, true. I forgot about that. But then works also fall into a spectrum of value all on their own. It's not enough just to make something you have to feel that it has value and therefore gives you value and you have to feel that it has value because it's attractive, funny, meaningful, informative or that it has been proven to be valuable by making you money.

Beauty is subjective which means you can trick others (and you're self, perhaps more importantly) into thinking you are.
Not so much being beautiful as looking polished and well groomed.
Dressing nicely, shaving, having healthy teeth, etc., improves my self-esteem and other's perception of me. Nobody would dare to call my appearance into question and as such, that isn't even something I have to worry about. In the complete opposite, it enables compliments on my appearance which further boosts confidence and detracts from the matter of whether I am beautiful or not.
As an example, I sport a buzz cut for the simple fact that it's low maintenance and looks well groomed. Whether it's out of fashion or gives me a mean look doesn't really matter as long as people perceive me as someone who is professional and pragmatic. That's my goal and how my waifu desires me.

>>27815
>a million dollars is a million dollars
What about speculative assets? Are a million dollars worth of stock selections and cryptocurrency, or perhaps something more tangible like gold or silver, a million dollars? Note I am very ignorant of economic matters, but I'm under the impression a lot (most?) of the wealth circulated today is purely speculative in nature rather than backed by any actual currency, which might make it possible for some groups to fabricate a sense of wealth without any intrinsic value, in much the same way as your example of one being able display the appearance of intelligence without any depth to back it up, no? Do I misunderstand how speculative markets work?
(although I imagine being able to manipulate the value of an asset in such a way might work to demonstrate a certain degree of power over them, which might have been your point?)

>>27822
https://en.wikipedia.org/wiki/Financialization
Leverage and derivatives. It's backed by currency from the future. With low interest rates, this timescale can be arbitrarily far out. That's why with a discount rate close to zero, you can have companies lose money every month for years on end but still have a high valuation, because they might just make money one day decades from now.

>>27815
I think it's pretty obvious when someone is trying to seem smarter than they are. A lot of times it's when someone is talking to me about something I do actually understand and know a lot about. Or just any subject that the other person isn't aware I'm well read in. Anyone who only cites a single source of information, for example, or clearly displays a bias towards a single, apparent, source which they never directly mention but which you can tell is there. Now everyone has biases, but most people who actually know what they are talking about will use a lot of "we don't really know, these are a handful of our best guesses". I firmly believe no one who understands the scientific method or historical analysis will ever say that absolutely everything they mention is 100% correct. While there are things that are pretty well solidified, those are usually concepts or ideas worked on laboriously for a hundred years by now. Any of the fine details of a specific subject will be to some degree up in the air. For example I could tell you that we know for a fact the main component of modern Europeans is West Eurasian ancestry and it's probably between 94 - 98% of the total, and that there's a smaller East Eurasian related component, but the fine little details of how much exactly they have of that East Eurasian component, and where exactly it plots compared to modern East Eurasians and Amerindians, is still being analyzed.

No one who tells you they know all the answers actually knows what they're talking about.

>>27825
>no one who understands the scientific method or historical analysis will ever say that absolutely everything they mention is 100% correct
Yet I've seen articles written by geneticist claiming they know race and intelligence has no connection, basically because "there isn't enough difference in dna between races". Which is pretty ridiculous. Even twins will end up expressing their identical dna in different ways.

>>27826
Any place where science intersects politics is a shitshow, as evidenced by the last 2 years of "pandemic" policy making. Soft-sciences are the easiest to distort by these means, because properly randomized trials are hard so it's very easy to selectively cherrypick and people rely on authority too much.

>>27826
No serious geneticist says that, although there probably have been geneticists that said there were no studies that showed a correlation between race and intelligence. Most geneticists don't even touch on that to begin with because we genuinely don't know enough about genetics to say that for sure. We're still in the process of working out how exactly phenotype ( expressed genes ) and genotype ( all an individuals existing genes, expressed or unexpressed ) work out for certain traits. At this point we still just go off of "this gene is correlated with X trait", which only means we tend to see more of that trait when we see more of that gene. As for genetic differences between races, it's not nearly enough to delineate separate species, because at the very least we've gotten down the major picture of species phylogenetic trees. But clearly there are enough differences between the major archeogenetic lineages to influence visible, physical phenotype, so there's no reason to say it wouldn't affect mental phenotype as well.

I've also seen people claim that there's no substance for the concept of race, blah blah blah. Yeah, it's true that it's not quite as cut and dry as everyone belonging to just one race or another, but it's more just a case of how genetics views the same structure of information differently than a more simple anthropology. As far as the structured nature of human populations, the 18th century racial theorists nailed it. They just didn't have the technology to expand and improve on the idea. And in fact, genetics has been able to differentiate separate archaic human populations, which all branched off a long time ago and which all correspond more or less perfectly to the races identified by the 18th century guys. Pic related are the actual scientific terms they use, and the main groups they identified.

Possibly the only major difference, really, is that a lot of populations were "race-mixed" between two or more of these within recent millennia, which makes it hard to classify a lot of groups as being of only a single "race". But genetics does still easily identify and separate these individual archaic populations, no matter how mixed someone is. And "pure" or nearly pure descendants of all these groups exist, the obvious ones being Europeans, East Asians, Black Africans, South American Natives, and Melanesians. It's pretty much match for match the same groups identified by Linnaeus and Blumenbach. They knew what they were looking at, and their identification of separate lines of human "races" within human ancestry was very well done.

>>27822
Stocks are a fairly liquid asset so it does not matter how speculative the investment is, you can still sell your holdings for cash at any time.

>>27829
If a large chunk of your equity is bound in your stock though ($XXM) then selling your stock will cause the price to drop, which means the actual value upon liquidation is less than the theoretical value. I.e. if someone like Bezos were to sell half of his amazon stock, the price would immediately crater.

>>27830
That can happen in extreme cases, but a reasonable person would have multiple positions and they could also split their trades over numerous days, that would not work for somebody like Bezos but for normal people it would.

What is more likely to get people is investing in small companies with low trade volumes meaning they also have low buy orders and those buy orders might be at a price lower than they would like.

>>27828
If they do it professionally, I would call them serious.
>we genuinely don't know enough about genetics to say that for sure
That's what I figured. From what I have seen written about it, although not explicit saying it's cut and dry no connection between race and intelligence exists, there was a confident implication that the "lack of evidence" is evidence against a connection.

>>27833
>there was a confident implication that the "lack of evidence" is evidence against a connection.
Which is, ironically, the single greatest scientific fallacy of all time and which is the first you learn of. I think articles are being written very dishonestly these days, and tacitly approved by other groups by way of them simply not addressing the problem. These days it seems scientific newspapers just say whatever the fuck they want whole actual geneticists quietly publish completely contradictory evidence in the background.

Take this study by Rosenberg: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1310579/

The kind of work done in this study, along with all the work by Lazardis, is what every company like 23andMe, AncestryDNA, etc bases their tests on. We've been using computers to automatically sort and sift human genetic similarity to discover structure and relatedness for over two decades now. That's because the most important genes in determining human relatedness are autosomal DNA, which contain the genes for individual-specific phenotype, like height, skin color, hair color, eye color, general build, skull shape, whether you'll get multiple fingers or not, etc. Using a K clustering algorithm, where a computer is told to try and blindly sort samples of information into a number of groups K, we can find the population structure of human , animal, and plant populations. It's a little complicated but basically, people descended from the same population, like say a continent, will share a very large portion of their ancestors. >99% in a lot of cases. This means eventually the populations sort out and balance out around these ancestral makeups, and depending on how different two groups ancestors might have been, you can get a lot of drift. Sexual and natural selection also plays a large role.

Basically, if someone is descended from four europeans, three africans, and a native, it's not just going to look like some amorphous blob stuck somewhere between europeans, africans, and natives. It's going to be very clear for the most part which segments are from european ancestors and which are from africans or natives, usually 98-99% of segments can be classified in this manner. Also, people from certain populations that clearly have very relatively recent ancestors will be very autosomally similar. Any European, for example, will be incredibly similar to other europeans, and even the most distinct groups like sicilians or east russians are going to be within 90 - 95% similarity compared to other populations.

This is a simplification so don't just take it face value and think you understood it, but I've tried to be as honest and as objective as possible. Personally, I wouldn't use the word "race" because I don't see it as being the most up to date concept, but I also think that the ideas being described by race are pretty damn accurate and pretty close to the truth.

I won't get into major basal ancestries but it is simply known now that all human ancestry can be divided up into a handful of major archaic ancestries which correspond incredibly heavily to the variation discovered in humans by the early, "racist' human anthropologists. I mean, a race by any other name is still a race, right?

Here's the 5 main clusters you will identify in a 3D population clustering PCA, mapped out onto the locations of the countries they were taken from. Each box is a series of lines, each line showing the ancestry of one person, and the different colors in that line representing the proportions of the different clusters it assigned their genetics to. So you can see how some lines spike up a bit differently and that shows that individual have more or less of a specific ancestry than is common. It's a common format used for showing sampled and K-clustered genetic data, so that it can be easily visually sorted and distinguished in one graph.

>>27833
Also, you may have seen this fine reconstruction. Well, the guy that made the study that they based it one essentially walked back his study a little later to say that actually they fucked up and it could have been one of a number of different skin tones. But you never heard about that.
Here's the retraction, https://www.newscientist.com/article/2161867-ancient-dark-skinned-briton-cheddar-man-find-may-not-be-true/
here's it un-pay-walled https://www.informationliberation.com/?id=57998
Don't mind the shitty polticial site.

And a funny response article. I guess telling the truth is worse than alt-righters finding out about the truth? I dunno.
https://www.newscientist.com/article/mg23731673-200-does-cheddar-man-show-there-is-such-a-thing-as-bad-publicity/

>>27835
I don't know anything about genetics, but PCA assumes your data can be described by linearly uncorrelated features. Is linearity a valid assumption here? Wouldn't some other manifold learning approach like umap be more appropriate here, since I'd imagine a lot of traits are e.g. gaussian distributed.

>>27837
PCA is useful because a lot of populations have shared ancestry, it does flatten the clusters a lot in terms of fine scale detail but it's more informative when considering the relation between a population or individual and the rest of the populations in that species. UMAP is used occasionally but a 3 dimensional PCA simply works best due to the nature of how members of a population species with minor admixture from another population in the same species are pulled towards the secondary cluster. Some people do actually prefer UMAP but I think it doesn't do a good job showing the bigger picture details, and since a persons ancestry can be pretty damn well defined by simply plotting them on a 3D PCA of global populations, it's the most useful for overall understanding of an individuals similarity on a basal level to other individuals.

>>27836
Also apologies for the 3DPD I've been pretty drunk tonight.

>>27838
Thanks for the info! Are you a biostatician by trade or something? You should post on /navi/ more often, I'd love to read more about your thoughts.

As to
>it doesn't do a good job showing the bigger picture details,
Long ago I read something about how t-sne preserves local relationships at the expense of global ones (and vice-versa for PCA), not sure if the same holds for umap (I only have a vague understanding of anything more complicated than svd/pca).

Speaking of which, while you're here, in practice when working with multivariate real-world data how do you decide which dimensionality reduction approach to use? I ask because long ago I skimmed all these various methods (pca, ica, t-sne, umap, etc.) but I'm but a humble programmer who doesn't use such things day-to-day so the details are all hazy. Do you just start with the simplest (e.g. pca) and if it doesn't give good separations, move to the more complex ones?

And another question if I may – if you happen to be familiar with kernel pca (or SVMs for classification), I've long wondered how people actually choose an appropriate kernel. The explanations always seem to assume a good kernel is known a priori but that's clearly not the case in the real-world.

>>27837
>>27838
3D PCA of global populations using G25 coordinates.

>>27840
>Thanks for the info! Are you a biostatician by trade or something?
It is what I went to school for, but unfortunately not what I get to do as a job right now.
> Do you just start with the simplest (e.g. pca) and if it doesn't give good separations, move to the more complex ones?
Pretty much. Things like recent mixed population structures can be hard to unravel without doing it that way.

I haven't done a lot of work with kernel PCA, so I'd have most of the same questions you do.

Post edited on 20th Nov 2022, 12:08am

>>27842
> unfortunately not what I get to do as a job right now.
What do you do now, if you don't mind sharing?

>>27837
>>27838
Ah I also found this article that explains why PCA is usually prefered over tSNE et al. for population genetics [1].

>PCA can provide more accurate relation between groups of samples while it is perhaps not that handy for discovering heterogeneity across samples, i.e. sample populations. On the other hand, populations are often known in population genetics, prior to computing PCA, from anthropological and archeological studies, therefore the goal here is not to discover them but understand the mutual relatedness between populations.

Seems the article is part of an entire series [2], I'll have to set aside some time to skim that since it seems pretty wel done.

---

In trying to find that I also came across this recent paper [3, 4] that seem to criticize the use of PCA in population genomics. Have you seen that paper before? What are your thoughts on it?

>We demonstrate that PCA results can be artifacts of the data and can be easily manipulated to generate desired outcomes. PCA adjustment also yielded unfavorable outcomes in association studies. PCA results may not be reliable, robust, or replicable as the field assumes. Our findings raise concerns about the validity of results reported in the population genetics literature and related fields that place a disproportionate reliance upon PCA outcomes and the insights derived from them. We conclude that PCA may have a biasing role in genetic investigations and that 32,000-216,000 genetic studies should be reevaluated. An alternative mixed-admixture population genetic model is discussed.


I have no idea what a mixed-admixture population genetic model is, nor whether

[1] https://towardsdatascience.com/why-pca-looks-triangular-a642daac721a
[2] https://towardsdatascience.com/tsne-vs-umap-global-structure-4d8045acba17

[3] https://www.nature.com/articles/s41598-022-14395-4
[3] https://homolog.us/blogs/bioinfo/2021/08/30/pushback-against-PCA-tSNE-UMAP-biology/

>>27844
(Crashed before I could edit), cont:
But in your experience with the field is PCA cherrypicking actually an issue (like p-hacking supposedly is in clinical trials)?

>>27845
>But in your experience with the field is PCA cherrypicking actually an issue
It can be, and there has been an uptick in flawed PCA studies, but fortunately it's usually pretty easy to spot in my experience. Lack of sample size, biased sampling, etc. For example a 3D global PCA can appear to flatten oceania into southeast asia, if you only use ~3 samples, but if you use more than a dozen it very quickly displays a much greater drift from other global populations. This is something that you know to correct going in to it, I can only imagine a total amateur or someone with an agenda actually trying to do that. You'll know when you're seeing cherry picking most of the time.

It has become more common recently though. There's been an increase in younger people making flawed assumptions in their papers due to not refining their PCAs and not double checking their work.

Here you can see the example I mentioned. But it doesn't even need 24 samples for the change to appear, just another 5 from Papauns will cause the same change in structure.

It's an ugly world. Those few beautiful/pretty things make it a little less ugly, and a little more bearable.