There is a binary number of length $N$ which consists of a consecutive series of 1s. For example, if $N=5$ the number is $11111$. How many ways are there to intervene on this number (i.e., replacing $1$s with $0$s) such that the series of $1$s is interrupted?
For clarity, the numbers $11101$, $11010$ and $10101$ are valid interventions but $11110$ and $01110$ are not, since the series of $1$s are not interrupted in the latter cases.
It is quite easy for the cases where the endpoint numbers are 1 (i.e. the first and the last bit of the binary number is $1$, such as $1XXX1$). These cases, I believe, sum up to $2^{N-2} - 1$, where the subtraction accounts for the case when all the numbers are $1$s.
However, I am unsure how to solve this for the cases where the endpoint numbers need not both be $1$s, such as $10110$.
All of the analysis below assumes that such degenerate sequences have been disallowed. Given the stipulation above, the adjustment is simply to add $(n)$ to the computation below.
This is because there are clearly $n$ such degenerate sequences that I originally disallowed.
Given any set $E$ with a finite number of elements, let $|E|$ denote the number of elements in the set $E$.
Let $S$ denote the set of all $n$ binary digit strings that are not all 0's.
Under the assumption that the string of $n$ 1's, when converted, can not be all 0's, after the conversion, then the set $S$ represents the possible results of this conversion.
Let $T$ denote the subset of $S$ where all of the 1's are in a single clump.
Then, the desired computation is
$$|S| - |T|,$$
where $~\displaystyle |S| = \left(2^n\right) - 1.$
This computation accommodates that you are not allowed to have the converted number be all 0's.
Any element in the subset $T$ will have the following characteristics:
It will contain a single clump of 1's that will have a length of between $1$ and $n$ digits. Here, I am assuming that if the converted number is all 1's, that that number is considered unsatisfactory (i.e. that the number belongs in the set $T$).
If the clump of numbers has length $k$,
where $k \in \{1,2,\cdots,n\}$,
then the starting position for this clump will be
some element in $\{1,2,\cdots (n+1-k)\}.$
For the element in $T$, besides the clump of 1's,
every other digit is a 0.
So, each element in $T$ is uniquely identified by two characteristics:
For $A \in \{1,2,\cdots,n\}$,
$B$ will range from $\{1,2,\cdots,(n+1-A)\}.$
Therefore
$$|T| = \sum_{A = 1}^n (n+1 - A) = \left[(n+1)\sum_{A=1}^n (1) \right] - \left[\sum_{A=1}^n A\right]. \tag1 $$
In computing the two terms in (1) above,
the first term is $(n+1)\times n$,
while the second term is $~\displaystyle \frac{n(n+1)}{2}.$
So,
$$|T| = n(n+1) - \left[\frac{n(n+1)}{2}\right] = \frac{n(n+1)}{2}.$$
Therefore,
$$|S| - |T| = \left(2^n\right) - 1 - \frac{n(n+1)}{2}.$$