Gel assemblies of functional nanoparticles, reversibly associated into percolating networks using bifunctional linking molecules, offer promise as versatile materials platforms. Molecular linkers can be customized to template interparticle spacing and modify colloidal network attributes, enabling design for structure-dependent properties. Mechanical properties of gels are commonly studied by molecular simulation, but simulating the optical response of large-scale, disordered assemblies has been computationally intractable, limiting our understanding of light-matter interactions in structurally complex plasmonic networks. Here, we use a recently developed mutual polarization method, capable of predicting optical properties for large disordered configurations of spherical particles, together with an experimentally-informed coarse-grained model to study the behavior of plasmonic linker gels. The simulation results demonstrate how blends of short and long linkers with the same average molecular weight can be chosen to deliberately modulate structure-dependent near- and far-field spectral features of the colloidal gel, while preserving gel mechanical properties. Linker selection can also be used to prepare gel networks with qualitatively different mechano-optical responses. The structural changes occurring under strain shed light on possible origins of experimentally observed red- and blue-shifting of optical extinction of plasmonic nanocomposites under uniaxial extension.