Define unreliable. Suppose it was 400,000 years ago, or 150,000 years ago, does that change the story?
First, there are other methods of dating things a lot of the time, so the scale is not entirely arbitrary, but does have bounds. Second, does the inaccuracy affect your understanding of the relative sequencing of populating various parts of the world?
More to the point, however, refining these temporal scales does nothing to refute the underlying premise of evolutionary theory, namely random genetic variation and selection by fitness for the environment. I think almost any biologist is open to dispensing with almost anything else if compelling data that leads to a contradictory conclusion is presented AND it is shown that other contradictory data is invalid. [mere clashes of data are simply an invitation for further research to settle the issue]
I picked the story of the electron because it is real, and it took a significant amount of experimental work by a significant number of researchers to refine techniques to the point where measurements were reasonably repeatable and reasonably in agreement. At no point did a discovery of measurements errors (which started off being fairly substantial) "disprove" the hypothesis of a universal quantized charge for the electron.
Un reliable means useless in this case. As this link leads to the “unreliable” Our of Africa theory:
http://www.detectingdesign.com/dnamutationrates.html
“The original 1987 study involved mtDNA from 136 women from many parts of the world having various racial backgrounds. The analysis seemed to support the idea of a single ancestral mtDNA molecule from a woman living in sub-Saharan Africa about 200,000 years ago. Later, more detailed studies seemed to confirm this conclusion. Unfortunately though, there was a undetected bias in the computer program as well as with the researchers themselves. The researchers used a computer program designed to reveal a “maximum parsimony” phylogeny or the family tree with the least number of mutational changes. This was based on the assumption that evolution would have taken the most direct and efficient path (which is not necessarily true, or even likely). Also, the computer program was biased by the order of data entry to favor the information entered first. This problem was recognized when the computer gave different results depending on the order that the data was entered. Now, after thousands of computer runs with the data entered randomly, it appears that the “African origin” for modern humans does not hold a statistical significance over other possibilities.26
The problems with these studies were so bad that Henry Gee, a member of the editorial staff for the journal, Nature, harshly described the studies as “garbage.” After considering the number of sequences involved (136 mtDNA sequences), Gee calculated that the total number of potentially correct parsimonious trees is somewhere in excess of one billion.25 Geneticist Alan Templeton (Washington University) suggests that low-level mixing among early human populations may have scrambled the DNA sequences sufficiently so that the question of the origin of modern humans and a date for “Eve” can never be settled by mtDNA.22 In a letter to Science, Mark Stoneking (one of the original researchers) acknowledged that the theory of an “African Eve” has been invalidated.23”
As the rest of the link makes clear; GI-GO.