Computer graphics is essentially the simulation of light and its interaction with matter in a 3D world, as viewed from a discretized 2D viewport. All aspects of the simulation are important to invoke suspension of disbelief in the viewer. Lower-fidelity simulation of one aspect will be apparent when juxtaposed against higher-fidelity simulation of the others.
Textures are a crude way to approximate the interaction of light with matter. The past decade has seen major improvements in the fidelity of this simulation via physically-based material pipelines. Personally I think there's a stunning difference between the end of the PS360 era vs. the end of the PS4 era, and I would attribute a majority of this improvement to improved material simulation. The *resolution* of material properties (textures) reaches diminishing returns at some point, and becomes far less important in terms of overall fidelity. Essentially what I'm saying is we've made major strides in "textures" over the past decade, they are not the fidelity bottleneck at the moment.
In contrast, light transport is still a bunch of horrific hacks. It stands out. We still see floating objects without proper contact shadows, vanishing screen space reflections, and nauseating cube maps. Artists spend effort faking GI with fake lights. Ray tracing is the most straightforward way to correctly simulate light transport. And no, we don't need full-on path tracing to realize its benefits. We've seen a range of effective hybrid solutions between Insomniac, Remedy's and 4A's stuff, all of which can reach 60fps on existing commodity hardware. But it still needs careful dev effort to balance effectiveness with performance. Think about that -- it's already viable and showing massive visual benefits, but there's also a huge runway in front for further improvement. That makes RT *the* prime candidate for graphics evolution in the years ahead.
I don't think texture *resolution* of all things is in need of comparable improvement.