This is a fascinating use-case which I find myself experiencing as well. As someone with a kinesthetic learning style, I can often struggle to absorb texts without audio, and vice versa.

The fractiousness of existing systems is a complex issue owing to business and technical considerations briefly discussed here (meta comment: Why is reference not a first class semantic construct within hypothes.is?)

Nonetheless, annotation of a fused audio/text stream would seem to be a really interesting product offering.

The centralized nature of today's systems is a problem for several reasons - Some are technical, but this applies socially and economically as well. This mindset of centralization is arguably a cultural afterimage from the advent of computing, with little incentive on the part of business to change.

In each of these senses (technical, social, economic) many of the computational abstractions we use today to simplify individual progress actually hold us back in aggregate.

In my view, there exist eery parallels between the technical foibles of centralization (CAP, etc), and the subtle manner of economic and behavioral coercion which come along for the ride. I do not believe there is a magic bullet to be found per se – but rather that it is possible to decentralize well, both from a technical perspective, and in terms of business as a function of value-addition rather than coercion.

What exactly do they mean by annotating? Is it note taking, or a conversation?