Where is the mistake in this problem? Let $$S = \sum^{\infty}_{n=1} a(n)$$
be an infinite series such that the nth partial sum is given by:
$$S(n) = \frac{n - 1}{n + 1}$$
since $$
a(n)=S(n)-S(n-1)=\frac{2}{n(n+1)} 
$$
Now, $S(1)=a(1)\Leftrightarrow 0=1.$  Where is the mistake?
 A: Depending on your definition there are two possible problems:
If S(0) exists:
Then $S(0)$ has to definitions as $S(0)=\sum_{n=1}^0a(n)=0$ and $S(0)=\frac{0-1}{0+1}=-1$
If $S(0)$ doesn't exists: Then $S(1)$ is your first term and $a(1)=S(1)-S(0)$ isn't a valid formula for $a(1)$
A: Good example of why you have to be careful with bounds :) Let me fix your post:

Let $$S = \sum^{\infty}_{n=1} a(n)$$
  be an infinite series such that the nth partial sum is given by:
  $$S(n) = \frac{n - 1}{n + 1}$$

for all $\boldsymbol{n \ge 1}$, and $\boldsymbol{S(0) = 0}$.
It follows that

$$
a(n)=S(n)-S(n-1)=\frac{2}{n(n+1)} 
$$

for all $\boldsymbol{n \ge 2}$.

Now, $S(1)=a(1)\Leftrightarrow 0=1.$

But now this falls apart, because the previous statement only held for $n \ge 2$.
A: The mistake is in saying that the $n$'th partial sum is $(n-1)/(n+1)$, with the implication that this applies to $n=0$ as well as $n=1$.
A: When 
$S_n-S_{n-1}=f(n)-f(n-1)$,
you also have
$S_n-S_{n-1}=(f(n)+C)-(f(n-1)+C)$
for any constant $C$.
So you can only guarantee $S_n=f(n)+K$ for some constant $K$.  If $K$ happens not to be zero for your choice of $f(n)$ and you are caught unawares, you get the discrepancy reported in the question.  You have to check the first term specifically to nail down the correct value if $K$.
