Some good sense.
I essentially agree with and endorse this article, with some caveats, and it should be read together with this piece I wrote several years ago.
The AltRight.com article is reasonably sound, although one caveat is that if one approaches these tests with a sense of realism with respect to their limitations - limitations spelled out in my Counter-Currents piece - then getting tested may not be a bad idea. Having the raw data could be useful if you can find someone who can do a genetic kinship analysis with it. But taking the details of the data literally - thinking that there's a real difference between 100% A, 0% B vs. 99.3% A, 0.7% B, for example - is ludicrous. I would take even the 90% confidence readings with a large grain of salt, and the 50% confidence readings are so absurd that the salt grain needs to be the size of the iceberg that sunk the Titanic.
The other caveat to the article is that the comments section is mixed; some comments are useful, some are asinine, so caveat emptor.
There are two basic questions here.
1. Is 23andMe a good test?
2. Assuming an ancestry test is good, is it worthwhile?
To which I answer: 1) No and 2) Maybe, depending on context.
In an absolute sense, 23andMe is superior to DNAPrint's tests from ~15 years ago; in a relative sense - comparing each test to the "state of the art" available at the time - it really isn't better at all. With the level of understanding and methodology we have today, coupled with a prudent interpretation of the data, one could do much better.
What if a test was sound? Well, sure, it can be interesting, but I'll repeat something I've been hammering home here over the past few years - the only biopolitically relevant genetic metric is genetic kinship (at all levels of genetic integration). If one can measure that, then it is useful. All else can be interesting, but not directly important from an EGI standpoint.
And if people are going to hysterically obsess over sub-fractional admixture percentages then this is missing the forest for the trees.