As far as I'm concerned this paper more or less describes HTTP(S) fairly well. The "3.5" address is the domain name that resolves to a meaningful IP address for the L3 domain you are in.<p>Folks are already using disjoint L3 address spaces with elements like load balancers providing the bridge between the public IP space and their own private IP space. I don't know if you could ever sell me on having more than one "main" IP space though, as I don't entirely understand what purpose that would serve, other than massively complicating the process of understanding what it is you are purchasing connectivity to.<p>See also RFC 1925, particularly sections 2.6a, and perhaps 2.11 :p
dmytri kleiner gave a great talk about doing this at a little bit of higher level back at SIGINT10.<p><a href="https://media.ccc.de/v/sigint10_3821_en_peer_to_peer_communism" rel="nofollow">https://media.ccc.de/v/sigint10_3821_en_peer_to_peer_communi...</a>
Setting aside the cultural baggage of its authorship and nomenclature, from a purely objective standpoint I could not pass <i>Introduction / Motivation</i> without attempting to elucidate the objections I felt to its generalizations.<p><i>The most frequently cited architectural flaw is the lack of a coherent security design</i>: The success of IP is simplicity/general utility - the ability of the system to support different use cases as a packet-switched alternative to previously dominant circuit-switched telephony systems. This is precisely the capacity of the system to vary service types and levels based upon application requirements. Viewed in this lens, not having a 'coherent security design' is <i>the core feature</i>, not a bug.<p><i>many question whether the basic service model of the Internet (point-to-point packet delivery) is appropriate now that the current usage model is so heavily dominated by content-oriented activities</i>: CDNs, content-addressable P2P networks (torrents), and multi-mirror package management databases are all excellent, broadly deployed counter-examples. The fact is, by normalizing packet-switching, IP has made bandwidth so cheap that inefficient distribution becomes a trivialized cost. Again, this is a <i>core feature</i>.
Between the name, the acronym TP, the tagline, and the fact that it was a paper co-authored by <i>A Panda</i> whilst presented in Beijing, all led me to assume it was an elaborate setup for a joke for at least half the paper.
> In this paper, we try to reconcile these two perspectives by proposing a backwards-compatible architectural framework called Trotsky in which one can incrementally deploy radically new designs. We show how this can lead to a permanent revolution in Internet architecture by (i) easing the deployment of new architectures and (ii) allowing multiple coexisting architectures to be used simultaneously by applications<p>In Trotskyist theory, permanent revolution has a specific meaning. There are two main components:<p>1. A socialist, proletarian revolution has to occur in a pre-capitalist society (such as Czarist Russia), bypassing the step of a bourgeois, capitalist revolution (such as the French Revolution), which Orthodox Marxism expected to be necessary in such countries.<p>2. The revolution has to be global, in contrast with eg. Stalin's theory of Socialism in One Country.<p>Seeing as permanent revolution is 1. not incremental and 2. does not allow for co-existence with capitalism, it seems sort of strange to name this approach after it.