Is Higman's group a free product?
As pointed out by @BenjaminSteinberg in the comments, a simple group cannot be a non-trivial free product.
It's also true that the Higman group cannot be a non-trivial free product. It has a presentation
$$G=\langle a,b,c,d | a^{-1}ba=b^2, b^{-1}cb=c^2, c^{-1}dc=d^2, d^{-1}ad=a^2\rangle.$$
Suppose that $G \cong A\ast B$, a free product of non-trivial groups $A, B$. Consider the subgroup $H=\langle a,b\rangle$. Clearly the relation $a^{-1}ba=b^2$ holds in the subgroup $H$, and in fact one can show that this is the only relation needed, so $H\cong BS(1,2)$, a solvable Baumslag-Solitar group. $H$ is non-trivial and not a non-trivial free product or free group since it is solvable and non-cyclic. Then by the Kurosh subgroup theorem, $H$ is conjugate into $A$ or $B$, let's say into $A$. Now we may take a non-trivial quotient of $G\cong A\ast B \twoheadrightarrow B$ by killing $A$. In turn, this kills the subgroup $H$ which is conjugate into $A$, so kills $a, b$. But one sees directly from the presentation that killing $a$ and $b$, one kills $c$ and $d$ as well, giving the trivial group, a contradiction.
Here are some details about constructing normal subgroups and some properties of negative curvature of groups (such as splitting as a free product).
The connection between these two subjects comes from small cancellation theory. Initially, the goal was to solve the word problem in some groups admitting specific presentations $\langle X \mid R \rangle$, which amounts to determining when an element of the free group $F(X)$ belongs to the normal subgroup $\langle \langle R \rangle \rangle$. Such methods have been generalised to many groups, including free products. And, as an application, it allows us to construct "many" non-trivial normal subgroups. One of the main ideas is that, if you take a group $G$ and an element $g \in G$ that satisfies a good small cancellation property, then $\langle \langle g^k \rangle \rangle$ is a normal subgroup of $G$ that does not contain $g$ if $k$ is large enough. The most recent and general theorem is this direction is:
Theorem: (Dahmani-Guirardel-Osin) A group admitting a non-elementary acylindrical action on a Gromov-hyperbolic space is SQ-universal, i.e. every countable group embeds in some quotient of $G$.
Notice that being SQ-universal implies having uncountably many normal subgroups, so it is a very strong negation of being simple. The theorem applies to (non-elementary) free products as they act acylindrically on their Bass-Serre trees (but some classical small cancellation can be also used instead, allowing more explicit constructions).
So simple groups are very far away from free products in the world of groups. For Higman's group, the situation is less extreme. In fact, the group is acylindrically hyperbolic, and in particular SQ-univeral like (non-elementary) free products. However, there is a classical hierarchy among negatively curved groups: $$\text{hyperbolic groups $\subset$ relatively hyperbolic groups $\subset$ acylindrically hyperbolic groups}.$$ (Here, I consider only non-virtually cyclic groups for simplicity.) Free products turn out to hyperbolic relative to their factors. Again, this can be seen from the actions on their Bass-Serre trees. However:
Fact: Higman's group is not relatively hyperbolic.
The justification is close to Ian Agol's argument:
Higman's group $H$ contains four subgroups $H_1,H_2,H_3,H_4$ such that:
- they are all isomorphic to the Baumslag-Solitar group $BS(1,2)$;
- they generate $H$;
- the intersections $H_1 \cap H_2$, $H_2 \cap H_3$, $H_3 \cap H_4$, $H_4 \cap H_1$ are all infinite.
If $H$ is relatively hyperbolic, then we know by (1) that each $H_i$ must lie in a peripheral subgroup $P_i$. But the collection of peripheral subgroups is almost malnormal in a relatively hyperbolic group, so (3) implies that the $P_i$'s must be all the same. We conclude by (2) that our peripheral subgroup must be the whole group.
So, loosely speaking, free products are "more negatively curved" than Higman's group.