1) But that isn't true. There are a number of statistical factors which can be used to determine whether or not species are considered to be subdivided into distinct clades. 

2) I don't know about the genetic similarities between Finns and other non-European populations. That said, in some other populations, this is the case. Certain populations in Africa share more with Europeans than some other African populations.

3) If you do not argue for distinction between "races", the term race becomes arbitrary, fitting into convenience instead of natural biological divisions. Basically, that means that race is socially determined, not biologically determined. 

4) There are a few problems with this:

a) Even in Rosenberg's models, clustering is largely arbitrary. The way Structure works is by allowing an individual to plug in values for the amount of clusters you would like to receive. Rosenberg did this with 2-6 clusters. In doing so, he showed that while 2 clusters does not produce any means to clearly distinguish between clusters, 3-6 all produce similar results with the conclusion that this is an acceptable way of dividing the species. He also does not provide results for potential clustering above 6. Basically, what this means is that a division into three clusters is equally valid to a division into six clusters, and depending on the results for beyond 6 clusters, maybe 20 clusters is equally valid. In fact, it is likely that if you put in 52 clusters, it would come back as highly relevant, as that is the number of populations used. Additionally, Tiskoff (2009) performed similar methodological calculations reporting on up to 14 clusters, including six clusters within Africa. To prioritize any one of these clustering models without further information would be to overstep the bounds of this research. 

b) This brings us to the issue of using Structure in the first place. While it is a highly regarded program, the means in which it is being used exists outside its intended bounds. As such, robust evaluation of the program must be performed to ensure that this utility is valid. Rosenberg did perform some tests, however, they did not account for the complexity of human genotypic relationships. As such, a robust test was performed using simulation models to see how Structure handles this type of analysis, and many issues arose: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3183908/

This article shows that when number of clusters is lower than the number of populations (in Rosenberg's case, 52 populations were utilized), the program can sometimes incorrectly cluster data due to the analysis existing outside of the program's bounds. The author hypothesizes that the reason for this is that the program places as many results as possible into homogenous clusters, and fills the remaining clusters with heterogeneous populations. The highly homogenous clusters statistically overwhelm the statistical issues created by the heterogeneous clusters.

This pattern is shown when viewing the African populations I described earlier. Despite these populations being more similar to European populations (based on allele sharing calculations) they are not paired in the same group, indicating that the results within Structure may not be an accurate representation of the similarity between genotypes. 

c) Rosenberg's models are not models of the statistical significance of races. His models describe the significance of patterning. Basically, what this means is that his models aren't looking into whether or not the degree of variation between groups is significant to call them biologically distinct, he is only looking to see if he can break the groups up. While the ability to divide a population into smaller groups is relevant to the question at hand, it is not indicative of the biological significance of that division.

d) And finally, we discuss biological significance. First of all, I think it is important to note that Rosenberg does not suggest that the possibility of clustering defines race and in fact says that his work should not be taken as evidence of such. Second, I do not believe that Rosenberg (2005) adequately responds to potential issues with sampling. They seem to expand their loci analysis, but do not sample additional populations. These populations are theorized to potentially fill the "gaps" claimed to exist by Rosenberg. Further, it is worth noting that the second article I posted was released after Rosenberg (2005), so it could be seen as a response to Rosenberg, not the other way around (and in fact, Rosenberg (2005) is specifically referenced in this article). This is supported by additional articles which were released after Rosenberg (2005). So, lets look into this a bit.

There's a lot here, and I'm not really sure how to condense it. I'm going to try to be brief because I don't want to be here all night.

-Fst values for Rosenberg's distinct populations do not support biological differentiation. Utilizing Fst values between continents gives a value of 0.043 which falls in the range of "little to no genetic differentiation" and this value gets lower with different clustering. 

-"in a rational classification of biological organisms, the computational possibility to determine group membership does not imply that these groups are meaningful according to biological systematic and evolutionary classification criteria"

-The insufficiency of the modern view of race is highlighted through even Rosenberg, which shows certain groupings which do not match typical social "race" grouping, such as a grouping of many asian populations with European populations.

-One of the issues with the modern system of race (that is, the common application, not the Rosenberg divisions), is that it does not necessarily reflect evolutionary history, but instead geography. Often, physical traits are adaptive to an environment, not an indicator of shared ancestry. ( https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3737365/ )

Overall, the problem with the concept of biological race, is that it is largely abstract. We could divide up the species in hundreds of different ways, but this wouldn't be evidence of "race" on its own. So the question becomes, "if biological race exists in the abstract, how do we apply it?". As I stated in my conversations with other users, concepts which run parallel to race do have some utility. You pointed this out yourself. Now, there is a question of whether the utility of "race" is the best means of wielding this utility, but there is some utility there. 

So, what have we said? Variation exists across the species in some form. Race in this context exists based on arbitrary criteria. Application of race is potentially relevant, but also potentially harmful and often not ideal. 

As such, where do we go from here? Well, lets talk about the different contexts. In academia, race is a dangerous concept because it is so nebulous. Different schools of thought can break it up differently, and what may be applicable in one application may not be applicable in another. Further, its use runs the risk of either misutilizing these concepts or misrepresenting them. In an academic setting, generally more concrete terms should be used when discussing topics other than the term itself. As such, the use of "race" should be immediately contextualized, or avoided if possible. In the context of medicine, race brings up a number of different challenges. Notably, how the concept of race fits into genetics. No matter what model you utilize, there is a wide variation within "races". As such, it is dangerous to apply racial concepts broadly within medicine without properly testing the validity and the robustness of the "race" hypothesis. Similar to academia, if more specific terms can be used, that is generally ideal.  

I suppose I'll reiterate the original point I was making several pages ago, because this conversation has completely gone off the rails and I have no idea what we are talking about anymore:

Race lacks broad relevance as a classification system. As such, it should not be used as a classifier alongside broadly relevant variables such as age. That does not mean that physical characteristics cannot be used in describing an individual. What it does mean is generally two things: One, that grouping outside of those directly observable characteristics should not be performed in a broad setting (meaning that when such characteristics are not directly relevant, they should not be used by grouping by systems and institutions);  two, it should not be used as a classification outside of the specific contexts where it holds relevance (note: classification systems are distinct from observations. This means that we should not assume things about an individual based on these characteristics as we would with characteristics such as age (other than those characteristics which are self evident)).

I have to assume that this will find a way to be misunderstood, but I've explained my point as clearly as I can.

Lo and behold through genetic analysis does indeed confirm many statistically significant genetic factors for human races ...  

This map even shows the 3 great races all shared a common ancestor nearly ~100000 years ago during the split at Western Asia ... 

Vastly more often than not, humans did NOT go around having offsprings with random people around the world ... (otherwise a geneticist wouldn't be able track you down at a specific location if humans really had interbreed so much)

I'd be interested in how some of the certain populations in Africa were closer to Europeans and how the researchers also did their sampling too with colonization having affected Africa for much of the last century ... 

It most certainly does fit into natural biological divisions like I showed in the table in my previous post ...  

I do not argue traits as being exclusive to specific human races. Just for the fact that I can find statistically significant genetic factors alone means that I can group these populations ... (there's nothing that precludes them from existing according to population genetics findings) 

That's all I need for the minimalist concept of races to exist ... (biological categorization is nearly entirely abstract but that doesn't take away it's utility)

That doesn't change the fact that his findings matched our suspicions about how human races are genetically clustered along continental boundaries and it further confirms how the vast majority of humans did not interbreed across the world and that these geographical races only shared a common ancestor ~100000 years ago ... 

The paper you linked is a non-issue since Rosenberg's sampling nearly matched the evolutionary history of human along the continental boundaries ... 

Sampling "heterogeneous" populations more than necessary is also flawed since humans did not randomly go around the world having offsprings ... (they were isolated for the most part and attempting so is grave error of over representation and I do not like having data fudged around) 

This sort academic dishonesty fundamentally goes against a scientists principles ... 

It's not of "biological significance" up until to the point where you need a bone marrow transplant ... (race is most certainly of biological significance in medical science to this very day) 

These so called "patterns" also seem to match our idea of common ancestors doing the great split ... 

Are you really going to retread what MDMAlliance linked again ? No matter how statistically small that variation in genetic cluster along these groups are if it's reproducible then you don't need to meet an X% of criteria to start classifying populations by whatever number the article arbitrarily sets ... 

It's either we make these biological categorizations or we don't and jeopardize the theory of evolution altogether like other religions do ... (making a biological categorization for human races is valid so far since we have biological data to base it off of)

We're already past the adoption of utilizing race in the medical world so why is it so hard to accept that they are different biological entities statistically speaking ? 

1) "Race lacks broad relevance as a classification system. As such, it should not be used as a classifier alongside broadly relevant variables such as age."

so why are you advocating for the system described in the op where students are encouraged to pick a race?

shouldn't you therefore be completely opposed to it?

"classification systems are distinct from observations"

this is the most stupid thing i've read in this thread and that is saying something

1) This is a move which de-emphasizes race, so I do not believe supporting it contradicts anything I've said.

2) I explained that statement the sentence after I made it: This means that we should not assume things about an individual based on these characteristics as we would with characteristics such as age. 

" This is a move which de-emphasizes race"

it de-emphasizes race by asking people to discriminate according to their perceived race? what??????

" I explained that statement the sentence after I made it: This means that we should not assume things about an individual based on these characteristics as we would with characteristics such as age"

so to assume that a "black" person has dark eyes, dark skin and curly hair would be flawed even though this is consistent with what we see in 99.9% of black people?

so to revisit the earlier example i made where lets say a black guy robbed me, i can't just say "dude! a bleeping black guy just robbed me!" 

instead you are proposing i not do the sensible thing and use language everyone is familiar with and instead dance around the issue by listing characteristics everyone knows describes a black person

you are joking right? i mean you don't really believe this do you

when you shut down your computer and you interact with people in the real world you aren't going to be acting in accordance with this nonsense i'm sure... right?

if that's the case and it has to be, what's the point?

1) That is not an argument against the point I am making.

3) The cluster is not a good distinction of meaningful genetic diversity as demonstrated in later points. 

4a) I don't believe that this point sufficiently rebuts this critique. The issue is the arbitrary nature of race determination. If clustering is your criteria, and clusters are accurately formed with 3 "races", 4 "races", 5 "races", 6 "races", 14 "races" and more, that makes the classification of "race" fairly meaningless. 

b) That isn't really a defense of the use of Structure, when quantifiable issues with the clustering has been pointed out. To say that the results matched expectations and therefore criticism should be thrown out is to perform science backwards, and thus break the entire process. If your methodolgy proves to be flawed, you cannot ignore those flaws because your results look good.

I am not sure what you mean when you say "Sampling heterogenous populations is also flawed". 

c) That is not a criteria for the establishment of race. Ancestry is relevant to medicine at a level far exceeding that of "race". 

d) That is simply not true. The biological relevance of clusters needs to be established, otherwise the term race becomes effectively meaningless within an evolutionary context. Rosenberg does not seek to establish the significance of these clusters, and as such, does not recommend that clusters be used as a stand-in for race. 

1. It absolutely is since we can make a distinction for human races through genetic analysis ... 

3. "Good" is subjective and clusters are a statistically significant measurement ... 

4. 

a) Well then this just means that just about any biological categorizations are "arbitrary" according to your issue ...  

b) From your article: 

"In STRUCTURE's defense, no single analysis is appropriate for all data, nor can a single analytic method be expected to reveal all patterns in data. Furthermore, the simulations presented here show that in many circumstances, STRUCTURE will produce evolutionarily appropriate clusters. If this wasn't true, STRUCTURE probably wouldn't be as widely used as it is."

From an evolutionary perspective, the classification of race is valid ... 

c) Ancestry is a criteria for the establishment of race ... 

If "race" means someone who has certain physical characteristics, the attribution of those characteristics is not an assumption, it is a prerequisite. 

1) But what differentiates between "race" and any of the other distinctions that can be made by genetic analysis?

3) If you don't like the word "good", replace it with "valid". Clusters are a statistically significant measurement of clustering, not races. The author of the study even specifies that the two are not synonymous within the study.

4a) How so? Species, for example, are determined primarily by the ability to produce viable offspring. This is a very real biological division. There is nothing arbitrary about it.

b) Humanity could theoretically be broken up at any point along our evolutionary history. What is the deciding factor for determining which one of these points is to be considered a "race"?

c) I am saying that there are many levels of ancestry which provide medically relevant information. This exists at levels of ancestry more recent than what is typically used to describe race.