Smithsonian Article by Richard Conniff

[K. Campbell]

I was all set to sit down and read Conniff's article in the Smithsonian and work up a good lather from what I had heard about the article. However, when I read it, I realized that it's pretty light and broad-brush, but it's also fundamentally sound.

One criticism that I do have is that he mixes the concepts of genealogy and population genetics. He incorrectly uses the term genealogy when he refers to what we do with DNA. (I prefer population genetics but will grudgingly accept the term genetic genealogy.)

Also, when he talks about the likelihood of a common ancestor, he means for all possible grandparents (in the recumbent sense) and not just the non-recombining paternal and maternal lines that many of us focus on.

But his basic argument is valid. His basic point is that our backwardly exponentially growing family tree must be bounded by the overall envelope of all people on the earth. When we do this, we have many more common ancestors than many people think.

For instance, when we discuss TMRCAs we consider 5,000 years ago a relatively recent SNP event. But some estimates have the entire population of the world at about only 14 million during this time period.

See:

http://www.census.gov/ipc/www/worldhis.html

I love the quote, "Thus Edward III, the King of England from 1327 to 1377, appears more than 2,000 times in the family tree of the modern-day Prince Charles".

When we think about this type of analysis, we must also remember that Steve Olson made the same argument a few years ago in the Atlantic Monthly article.

[Maria_W]

Thanks for article.

Thank you for posting this article. Will try to read it later!
Maria

[lgmayka]

When we think about this type of analysis, we must also remember that Steve Olson made the same argument a few years ago in the Atlantic Monthly article.

Yes, and his "analysis" was just as incompetent then.

Very briefly:

1) Beyond 15 generations, any ancestors other than the patrilineal and matrilineal ones are utterly irrelevant, because on the average each such ancestor contributed less than a single gene to the descendant. Even if the ancestor did contribute a gene or two, we have no way of knowing which.

2) Any attempt to use a video game ("simulation") to "prove" that the entire human race shares a recent (e.g. within the last 5000 years) ancestor is utter idiocy. Video games prove nothing. Anyone who wants to prove that the entire human race shares recent (within the last 5000 years) DNA knows very well the scientific standard: He must collect DNA from every remote tribe on earth, including the African bush, the Australian outback, and the Papuan jungle, and point out the specific DNA in common and show why it is recent. Y chromosome and mitochondrial DNA studies have done exactly this. The standard is just as high for autosomal DNA.

Once again, an author who relies on video games instead of conducting a scientific investigation is simply an idiot.

[lgmayka]

A couple more points of foolishness:

1) The claim of 10% misattributed paternity is a very popular lie, used often by promiscuous people to justify their own sins. In reality, both scientific and genealogical studies have shown that the historical rate of misattributed paternity is in the range of 1-3%.

2) The last paragraph of Conniff's article is perhaps the most ridiculous of all:
---
But I also want my daughter to understand that nothing in her genealogy defines her...What makes us who we are, what makes us people worth knowing, comes from within ourselves, and from the often annoying, somewhat laughable, occasionally lovable families we live with now.
---

Conniff apparently has not studied enough biology to understand that if his parents had been cats, he would be a cat too. In other words, genealogy is exactly what makes us who we are--human rather than feline or canine or simian. It is exactly genealogy that "defines" us and makes us "worth knowing," precisely because humans are inherently worth knowing whereas cats and dogs and apes are not (in the same sense).

Put another way, it is terrible to teach a child that her inherent worth is dependent on what she does or who she is with. On the contrary, the most mentally disabled human with no family at all is nevertheless "worth knowing" simply by being human--i.e., because of her genealogy.

[vineviz]

2) Any attempt to use a video game ("simulation") to "prove" that the entire human race shares a recent (e.g. within the last 5000 years) ancestor is utter idiocy. Video games prove nothing.

The problem with Olson's argument was not that it was based on a simulation, but rather that he constructed his simulation with ludicrously improbable assumptions (e.g. near total exogamy).

Simulations are not "video games" and can be used productively in all fields of science to help understand phenomena that are not directly observable.

[lgmayka]

Simulations are not "video games" and can be used productively in all fields of science to help understand phenomena that are not directly observable.

No, a simulation is indeed a video game. Since the human supplies both the algorithm and the inputs, the output cannot possibly 'show' or 'prove' anything more than what the human himself provided.

Simulations are used productively in scientific fields as a way of suggesting hypotheses. In other words, given some concrete data, a scientist may run a simulation to help in constructing a hypothesis, a proposed explanation of that data. But at best, the result of the simulation is still only a hypothesis. It proves absolutely nothing in itself--it must be tested (verified or refuted) through further real-world experiments.

Even if Olson were to improve (i.e., make more realistic) his algorithms and inputs, his simulation would still only produce a hypothesis--interesting, but certainly not worthy of publication in itself, except perhaps as a brief 'letter'. To have any scientific credibility, he would then have to do the real work: travel the world, collect DNA, and test it for recent commonality. This is exactly what the investigators of the Y chromosome and mitochondrial DNA have done, and we should demand of Olson nothing less.

[Nagelfar]

No, a simulation is indeed a video game. Since the human supplies both the algorithm and the inputs, the output cannot possibly 'show' or 'prove' anything more than what the human himself provided.

"computer game" in your context would be a better word. Where's the video?

[Bracari]

Conniff wrote that genealogy is meaningless, relatively.
Come on, give me a break !
People that throughout history has destroyed historical genealogical registers, old family pictures, old passports, old genealogical documents unfortunately have always thought just like Conniff but they had not culture and had not studied in any educational or scientific institution. Usually that's the same people that think that DNA analysis is relatively meaningless, too.
Next step is to declare things like history, arts, philosophy, literature and general culture as relatively meaningless too because you can't immediately eat them or you only mind for people or things right in front of your nose.
Too bad for the development of genealogy as a scientific discipline.

[lgmayka]

"computer game" in your context would be a better word. Where's the video?

You are correct, no video. I was thinking of the fact that Olson's simulation was actually quite reminiscent of The Sims.

[vineviz]

Simulations are used productively in scientific fields as a way of suggesting hypotheses.

Video games, on the other hand, are NOT used productively in scientific fields. Therefore, computer simulations are not video games. Which is exactly what I said to begin with:

Simulations are not "video games" and can be used productively in all fields of science to help understand phenomena that are not directly observable.

Don't confuse my motives: I have no inclination to defend Olson, who is a journalist not a geneticist or anthropologist. I could not care less what he does or doesn't do.

I do care, however, about folks who might be led by your pejorative dismissal of a technique you don't or won't understand.

[Ann Turner]

Yes, and his "analysis" was just as incompetent then.

Very briefly:

1) Beyond 15 generations, any ancestors other than the patrilineal and matrilineal ones are utterly irrelevant, because on the average each such ancestor contributed less than a single gene to the descendant. Even if the ancestor did contribute a gene or two, we have no way of knowing which.

2) Any attempt to use a video game ("simulation") to "prove" that the entire human race shares a recent (e.g. within the last 5000 years) ancestor is utter idiocy. Video games prove nothing. Anyone who wants to prove that the entire human race shares recent (within the last 5000 years) DNA knows very well the scientific standard: He must collect DNA from every remote tribe on earth, including the African bush, the Australian outback, and the Papuan jungle, and point out the specific DNA in common and show why it is recent. Y chromosome and mitochondrial DNA studies have done exactly this. The standard is just as high for autosomal DNA.

Once again, an author who relies on video games instead of conducting a scientific investigation is simply an idiot.

You're setting up a straw man. The argument is not DNA-based -- it's ancestor-based, and there's no requirement that you have any of the DNA of the MRCA. Have you actually read the technical paper on the topic? Here are links to a couple of posts in the GENEALOGY-DNA archives where I have attempted to explain the concepts.

http://archiver.rootsweb.com/th/read/genealogy-dna/2006-07/1151867920

http://archiver.rootsweb.com/th/read/genealogy-dna/2006-07/1151981818

But back to DNA, I think the paucity of highly informative AIMs (Ancestry Informative Markers, the autosomal SNPs used by DNAPrint and other researchers) actually provides support for the theory.

Ann Turner

[lgmayka]

The argument is not DNA-based -- it's ancestor-based, and there's no requirement that you have any of the DNA of the MRCA.

False. Did you actually read the article? He devotes an entire paragraph to slandering genetic ancestry, though his article makes clear that he doesn't understand it in the least.

Have you actually read the technical paper on the topic?

If you mean Chang's and Olson's discussion of their computer games, yes, I read them, and I have already given my opinion of playing The Sims and then publishing game results as if they were fact.

But back to DNA, I think the paucity of highly informative AIMs (Ancestry Informative Markers, the autosomal SNPs used by DNAPrint and other researchers) actually provides support for the theory.

What evidence do you have that AIMs are not available for the vast stable majority of the world--i.e., excluding immigrant-mixing nations such as the United States, and nations that have undergone large-scale genetic shuffling in the past 2000 years such as England? In other words, does anyone claim that no AIMs exist to distinguish African Khoisan from Australian Aborigine?

[Ann Turner]

MRCA of all mankind

False. Did you actually read the article? He devotes an entire paragraph to slandering genetic ancestry, though his article makes clear that he doesn't understand it in the least.?

You were criticizing Olson's article "The Royal We" in Atlantic Monthly 2002, right? I'm away from home right now, so I can't refer to it, but I don't recall that his arguments depended in any way on possessing DNA from the MRCA.

If you mean Chang's and Olson's discussion of their computer games, yes, I read them, and I have already given my opinion of playing The Sims and then publishing game results as if they were fact.

I'm referring to the article by Rohde, Olson, and Chang, "Modelling the recent common ancestry of all living humans", Nature. 2004 Sep 30;431(7008):562-6, which addressed some of the unrealistic parameters of the earlier Chang model (e.g. totally random mating). Again, I can't refer to the paper because I'm away from home, but here's a link to the preliminary version on Rohde's website.

http://tedlab.mit.edu/~dr/Papers/Rohde-MRCA-two.pdf

What evidence do you have that AIMs are not available for the vast stable majority of the world--i.e., excluding immigrant-mixing nations such as the United States, and nations that have undergone large-scale genetic shuffling in the past 2000 years such as England? In other words, does anyone claim that no AIMs exist to distinguish African Khoisan from Australian Aborigine.

A couple of points: firstly, this "large-scale genetic shuffling" is exactly what causes the surprising recency of the MRCA, along with smaller-scale movements that have gradually accumulated over time (via the "connectors" between the "hubs").

Secondly, I did not say there were NO AIM's -- I chose my words carefully when I said there was a "paucity" of "highly informative" AIMs. However, I am claiming that they are hard to find, and I think that could be a consequence of a relatively recent MRCA (within say the last 5 to 10 thousand years). A maximally informative SNP is found in 100% of one population and 0% of all other populations. For instance, this article screened 450,000 SNPs for the purpose of distinguishing between African and European ancestry. The BEST 3,011 SNPs (less than one percent) showed an average difference in frequency of just 56% between African and European populations, so any SNP can be readily found in both populations.

http://www.journals.uchicago.edu/AJHG/journal/issues/v74n5/40977/40977.web.pdf

This article has data for Native Americans vs Africans

https://www.ancestrybydna.com/%5Cfiles%5Cskin.pdf

I don't get hung up on the absolute nature of the word "all" -- as I said in my GENEALOGY-DNA post I linked to in an earlier message, I think it's interesting if the results only apply to 99.99999...% of humanity.