How to verify new digits of $\pi$?

Bob makes a claim that he made a new record and computed $\pi$ to 10 trillion digits (or your favourite number here). How would Alice verify that the newly computed constant is actually a correct approximation of $\pi$?

Given a finite string $x,$ of $n+1$ decimal digits: $(3\ 1\ 4\ 1\ 5\ 9\ 2\ldots),$ is there an efficient algorithm to decide whether $x$ is an approximation of $\pi$ up to the $0.\underbrace{00 \ldots 01}_{n-1}$ decimal places?

Edit: Clarification.

1. Alice does not have access to Bob's method (so she can't prove that his method is correct).
2. Alice only receives $x$ from Bob (in any number bases), and wants to verify that $x$ is indeed a correct approximation. No further communications between them.
3. Alice could look at all digits of $x$ but should be able to verify in time $\ll$ than what it takes to compute $x$.
4. Motivation: Assume Alexander J. Yee did not publish his code nor his method. He only publish $x$ in many number bases. He said it took him 3 months to compute $x.$ How could we verify his claims that $x$ is correct in a day or week or two without access to his code and formulas? Is there such a verification formula or algorithm?
-
–  Arthur Fischer Mar 14 '12 at 22:12
Not sure if this link helps or not. But you may refer this as well numberworld.org/misc_runs/pi-5t/details.html –  Kirthi Raman Mar 14 '12 at 22:28
Can you ask Bob to give you the digits in base 16? –  Aryabhata Mar 15 '12 at 0:12
@Aryabhata yes. In any base. –  user2468 Mar 15 '12 at 0:27
My earlier link was just silly, but perhaps this one might be useful in the context you are looking for tinyurl.com/84396h3 –  Kirthi Raman Mar 15 '12 at 0:38

Since you allow any base, (16 in particular) and randomized algorithm, you can use the Bailey-Borwein-Plouffe formula which allows you to compute the $n^{th}$ digit of $\pi$, without having to compute the earlier $n-1$ digits! (Alas, such a algorithm seems to have been discovered only for base-16.)
@PyRulez: Not sure what you mean by efficient, but given an n digit number in base b, there will be a linear time algorithm ($\mathcal{O}(n)$) to convert between bases. People are still trying to find such formulae in other bases... –  Aryabhata Aug 19 '13 at 1:51