Just for the record.
Contrasting several simpler and more complex models of mtDNA molecular clock evaluation seem to show coincidence at least in this aspect our shared oldest grandma (by purely matrilineal ancestry) lived some 200,000 years ago.
Krzysztof A. Cyrana and Marek Kimmel, Alternatives to the Wright–Fisher model: The robustness of mitochondrial Eve dating. Theoretical Population Biology, 2010. Pay per view.
According to the news article at Science Daily:
Each model has its own assumptions, and each assumption has mathematical implications. To further complicate matters, some of the assumptions are not valid for human populations. For example, some models assume that population size never changes. That is not true for humans, whose population has grown exponentially for at least several thousand generations. Other models assume perfect mixing of genes, meaning that any two humans anywhere in the world have an equal chance of producing offspring.
Cyran said human genetic models have become more complex over the past couple of decades as theorists have tried to correct for invalid assumptions. But some of the corrections -- like adding branching processes that attempt to capture the dynamics of population growth in early human migrations -- are extremely complex. Which raises the question of whether less complex models might do equally well in capturing what's occurring.
"We wanted to see how sensitive the estimates were to the assumptions of the models," Kimmel said. "We found that all of the models that accounted for random population size -- such as different branching processes -- gave similar estimates. This is reassuring, because it shows that refining the assumptions of the model, beyond a certain point, may not be that important in the big picture."