As in the title, but to make more clear:
If, with probability 0.1, an algorithm A can distinguish between two ensembles D_n, E_n (indexed by sec. paramater n), then are D_n, E_n not computationally indistinguishable?
More generally, is it true that, $$\Pr_A\Big[\big|\Pr_{x\sim D_n}[A(x) = 1] - \Pr_{x\sim E_n}[A(x) = 1]\big|>\epsilon\Big] = 0.1 \implies \big|\Pr_{x\sim D_n}[A(x) = 1] - \Pr_{x\sim E_n}[A(x) = 1]\big|>\epsilon/10$$
