<p>Standard accounts of logical consequence take it to be determined by valid inferences. Recently, it has been argued that valid inferences underdetermine consequence. Instead, consequence should be characterised as determined by metainferences, i.e., inferences between inferences. For all its benefits, this view faces various problems, including the fact that it appears to be incompatible with the phenomenology of the inferential practice, and the widely endorsed idea that simplicity and naturalness should guide logical theorising. This paper rejects this idea, arguing that, in its extreme form, it amounts to an untenable form of foundationalism about logical consequence. Instead, I will argue that metainferentialism is best embraced and further justified from the standpoint of a broadly constructionist (in the sense of Floridi (Metaphilosophy 42:282–304, 2011)) approach to logic.</p>