This topic contains 1 reply, has 1 voice, and was last updated by 
josh February 28, 2022 at 9:02 am.
-
Posts
-
joshThe technical flavor of the approach that favor is contained in earlier proposals – i.e. a thread about a Virtual MultiSystem DNS/Hardware ID Scheme. I wrote about the utility of allowing the possibility of universal message passing between unafilliated networks. Where the light “servers” for this sort of thing can be *held* at known locations & used for security debugging purposes, it can help support the robustness of other goals. One can consider how to make such an arrangement itself maximally robust to unwanted interference.
This concept is different than traditional network models because:
a) Messages & Virtual IDs and Virtual DNS/routing are network primitives at a lower level than general applications. Constant on is not assumed. Homogeneity is not assumed. Same Party & free gateways are not assumed.
b) Sessions or behavior that depends on keeping open sockets is more of a luxury than basic messages, not a prerequisite or foundation.
At a needs/strategic level, the idea is to be maximally flexible in the ability to use whatever is available at any given time, allowing other strategic needs & economics to shift traffic to the forms that are most advantageous. So GT can compete to be the end-to-end supplier in all the routes that matter.
Considering video conferencing as an example – suppose a format that allowed for micro-pauses or chunks of video plus text summary were available at a much lower cost with greater robustness. Is the loss of the constant stream a big deal? For most cases, it is not. It can be added where it is.
You must be logged in to reply to this topic.