The role of audio wire/cable has been routinely underestimated by mainstream engineering. However, in aerospace and now in high-speed digital technology, wire has risen to the status of a system-limit - and therefore importance. There are some design engineers who state that audio cable is a large multiple more complex than digital cable. Audio must support radical changes of voltage and current with back-forces. The problems are real and the solutions difficult.
The speakers you cite are taking their best guesses and engineering solutions within their defined parameters. Those parameters, understandings and blind spots vary widely among designers. Welcome to the murky world of audio engineering.
My recent deep dive builds on Thiel's history. In 1977-8, we faced the daunting task of whether we would tackle the impossible dream of coherence. We found that a coherent source showed glaring problems which vanished when reverting to normal high-order, non-coherent topologies. A near deal-breaker was a persistent gritty, homogenized haze in the coherent iteration that defied our attempts to understand and mitigate. Our aerospace-physicist (non-audio enthusiast) cousin made a visit, heard the problem and suggested we all (everyone) were hearing the effects of inter-strand cross-talk as learned from deep space probe image retrieval. Lesson learned included how our audio neurology processes coherent audio input at qualitatively higher scrutiny than non-coherent input, which it categorizes as artificial and less important.
I suggest that you would hear significant, qualitative differences between your two, or any, cable scenarios if using a coherent source such as Dunlavy, Thiel or Vandersteen, whereas those differences would fade to near meaningless with any non-coherent speaker.
My present work is, of course, with Thiel speakers. I've used the CS2.2, 3.5, and now extensively the SCS4. But the lessons apply to all speakers, and become meaningful for all coherent speakers. I thought that Thiel's 1978 model 03 brought the use of solid, rather than stranded wire, but we later learned that Dahlquist (from aerospace) used 18 gauge solid in the DQ10 darling of the day. Our 18-2 solid twisted pair in teflon spanned Thiel's whole timeline with critical comparative re-evaluation in 1988 for the CS5 and again for the 2007 CS3.7.
Regarding variable gauges for frequencies - the differences are subtle, but again, become meaningful for coherent sources. The math that describes propagation in wire diverges in the lower audio frequencies. Below 1kHz the rules get squishy and below 100Hz the rules diverge. Bass frequencies are supported with greater propagation integrity with conductors with lower surface to core ratios. So larger gauges are 'better'. Happily, skin effect saturation frequency also decreases since woofers are attenuated in the upper audio octaves.
The soup of interactive ingredients is extremely complex. The cable that I have developed makes sonic and measured advances without harm in Thiel's coherent topology.
