1: \begin{abstract}

2: We study the class of networks which can be created in polylogarithmic parallel time by

3: \emph{network constructors}: groups of anonymous agents that interact randomly under a uniform random scheduler with the ability to form connections between each other.

4: Starting from an empty network, the goal is to construct a stable network which belongs to a given family.

5: We prove that the class of trees where each node has any $k \geq 2$ children can be constructed in $O(\log{n})$ parallel time with high probability.

6: We show that constructing networks which are $k$-regular is $\Omega(n)$ time, but a minimal relaxation to $(l, k)$-regular networks, where $l = k - 1$ can be constructed in polylogarithmic parallel time for any fixed $k$, where $k > 2$.

7: We further demonstrate that when the finite-state assumption is relaxed and $k$ is allowed to grow with $n$, then $k = \log\log{n}$ acts as a threshold above which network construction is again polynomial time.

8: We use this to provide a partial characterisation of the class of polylogarithmic time network constructors.\newline\\

9: \textbf{Submitted to Track B: Distributed and Mobile}\newline\\

10: \noindent\textbf{All omitted details are included in a clearly marked Appendix, 

11: to be read at the discretion of the Program Committee.}

12: \end{abstract}

13: