The question.
I was wondering whether or not the following assumption is true.
Let $E$ be a complete metric set containing a point verifying the property $\mathcal P$. Then $E$ is uncountable.
A point $x\in E$ verifies the property $\mathcal P(x)$ if for every neighbourhood $V$ of $x$, there exists $y\in V\setminus \{x\}$ and $y$ verifies $\mathcal P(y)$.
In other words, $\mathcal P(x)$ is true if, and only if, $x$ is not isolated, and for every neighbourhood $V$ of $x$, $V\setminus \{x\}$ contain a non isolated point verifying the same property.
What I did.
I tried to prove the veracity of the result since I wasn't able to find a counterexample.
Let $x\in E$ be a point such that $\mathcal P(x)$ is true.
Then we can construct recursively a sequence $(x_n)$:
$$\begin{cases} x_0\ne x \\ \mathcal P(x_0)\text{ is true.}\end{cases}$$
Then assuming $x_0,\ldots,x_n$ are constructed, let's define $r_n$ such that
$$\forall i\in\{1,\ldots,n-1\},\quad x_i\notin B(x_n,r_n).$$
We have a set $X=\{x_i\}$ of points of $E$. Since when can always double the size of $X$ because every $x_i$ verifies $\mathcal P(x_i)$, I would tend to think that we can construct a set $Y$ in bijection with $2^X$, which would have the cardinality $2^{\aleph_0}$, so $Y$ would be uncountable. So $E$ would be.