How should I think of an open vs. closed set?
Solution 1:
A set is open if, from any point in the set, you can wiggle in any direction a little bit and stay inside the set. What "wiggle" means depends on the context at hand.
In metric spaces, "wiggle" means what you might expect: "move a small distance". That is, for each point in an open subset of a metric space, there is a ball around the point which is contained in the open set.
In one important extreme, the trivial topology $\{ \emptyset,X \}$, there is no wiggle room anywhere: everything is somehow collapsed together, and by wiggling at all you "bump into everything". Formally, any sequence in the trivial topological space converges to every point in the space.
In the other important extreme, the discrete topology, all sets are open, including singletons. In view of how the subspace topology works, a nice way of viewing this is by thinking of a discrete space as a set of isolated points in a larger space. For instance $\mathbb{Z}$ is a discrete subset of $\mathbb{R}$. (I am not sure how to give an intuitive explanation of the subspace topology, however.)
Other topological spaces are in between these two extremes.
The axioms of topology make some intuitive sense in this framework:
The empty set is vacuously open: my intuitive definition says "from any point in the set" and there are no such points.
The entire set is also certainly open, provided we assume that "wiggling" does not take you out of the set.
If $U$ is a union of open sets and $x \in U$, then there is a particular open set $O$ in the union with $x \in O$. This $O$ gives $x$ some wiggle room, and this wiggle room is also contained in $U$.
If $U$ is a finite intersection of open sets and $x \in U$, then we can take the "wiggle room" of $x$ from each of the sets in the intersection and intersect all of them. This will still leave some room to move around $x$. This is easier to interpret in metric spaces, where the wiggle room consists of balls: the finite intersection of balls of a fixed center is just a ball whose radius is the minimum of all the radii. From this perspective, infinite intersections of open sets needn't be open, because this "minimum" (really, infimum) radius might actually be zero, which would leave us with no room to move.
Some (maybe most) general topology books prove that the definition of a topology with open sets is equivalent to the definition of a topology with a neighborhood system. This latter definition is more closely tied to the intuitive picture I've been trying to draw here.
Note that this idea is not really universal. For example, I don't think there is a way to apply this idea to the Sierpinski space $(\{ 0,1 \},\{\emptyset,\{ 0,1 \},\{ 1 \} \}$).
Solution 2:
I (personally) think of closed sets the way I explained here: How would I explain an 'open set'
One way to think of closed sets is they are the complement of an open set. (In many instances this may even be the definition of a closed set.)
The way I think of closed sets is in terms of convergence of sequences... if you take any sequence $s_{n}\to s$ where each $s_{i}\in S$ then if $S$ is closed you are guaranteed that $s\in S$ too. The "intuitive" way to think of this is that if you are moving around inside a set but slowing down and coming to a stop (eventually a stop... after $\infty$-ly many steps). Once you've slowed down enough, you can predict where you will stop. If it must be the case that your prediction is always a stopping point inside the set, then it is closed. Of course in practice, a sequence could be very bouncy and not come to a stop evenly at all... but this is how I think about it.
Solution 3:
Topology is literally the study of open sets. A topology is a collection of open sets.
In $\mathbb{R}$, open sets are arbitrary unions of open intervals, $(a,b)$ and closed sets are arbitrary intersections of closed intervals, $[a,b]$. These are important because they define limits, continuity, etc.
In topology, you simply define what "open" means for your space. In that, you define limits, continuity, etc.