Will the spec define what is meant by "high speed, low latency networks"?

[leeprtc]

Section 3.1 Introduction (page 11 in v0.3 PDF) has the following sentence.

"Sunfish provides a method by which compute resources, that are attached by high speed, low latency
networks (e.g., CXL/PCIe, RDMA, Ethernet), can be assembled together and managed as if the
components were part of a traditional compute server."

Will the spec define what is meant by high speed, low latency networks or leave it ambiguous by design?

My concern is if we are looking at Ethernet, speed can vary. While one can logical group resources into a compute server, in the real world, things may not work well. If this is the case, do we care?