Power to the People
The “new” p2p technology is for many as much a “political” statement as a technological one. Perhaps this explains some of the buzz—it’s a revolution of sorts.
Bit 1.16 P2P is about distributed ownership of resources.Characteristic for current p2p networks is that the computing resources (PCs) are owned and managed by the users. |
The issue has several aspects. At its most trivial level, any one entity bears few centralized investments and administrative costs. Even server-mediated solutions can be deployed with modest means, as numerous startups have shown. As nodes connect to the network, offering their resources, a computing entity on a larger scale emerges. This building of distributed supercomputers on a shoestring has been used to great effect with various distributed computing (DP) projects, although I hesitate to accept some of these shared resource and distributed processing implementations as valid examples of p2p technology (I’ll explain why in Chapter 2).
On the other hand, corporate entities have made use of distributed PCs for various kinds of p2p solutions to allow local employee groups to actively retain and maintain resources, yet provide global access without the need for new central servers and storage. The bottom line here is better resource quality at a lower cost.
In the corporate world, control ultimately rests with the boss. For the public at large, control rests with the individual—at least that’s the view of many.
Bit 1.17 P2P is about distributed control of resources.The fact that the data and resources are controlled by the users is characteristic of and is the motivation for many current p2p networks. |
A strong element of anarchy and “total freedom” characterizes p2p usage, not infrequently with considerable disregard for the commercial values of “ownership” and content control. This view largely fueled the Napster phenomenon: the free exchange of music tracks that otherwise everyone would have had to buy bundled on CD. The distributed control in p2p meant that it was exceedingly difficult for the commercial interests to stop the, as they saw it, illegal copying. The legal and moral issue is more complex than it might appear, and we deal with this in Chapter 4.
Related to this issue is the stance taken by, for example, Free Haven and Freenet, two p2p models that primarily intend to ensure that resources such as files or Web content are protected against censorship and forced site closure. Encrypted and distributed across a p2p network of independently managed nodes, the material becomes as persistent as the network itself.
Virtual P2P
Using existing Internet connectivity as a base for deploying a virtual p2p network leverages the reach and capabilities of the implementation enormously. To join such a network only requires the proper software and Internet access from anywhere.
It’s interesting to note that e-mail was the “killer application” for the early Internet, which in its extended form includes newsgroups and mailing lists. The popularity of e-mail confirms the assumption that what people always want to do, whatever the technology, is communicate with each other. A major reason it became so popular is because the e-mail transport mechanism and domain addressing means that a sender needs to know only a person’s permanent identity in order to contact that person—location doesn’t matter.
Whether e-mail should still be considered p2p technology is an open question. Like a number of activities, it has a person-to-person appearance, whatever the technical infrastructure, but the actual implementations can vary greatly. For most people these days, their mailbox is not the endpoint of the connection. In the same way, much so-called p2p functionality can be implemented in different ways, with a varying amount of actual p2p technology and using different architecture models.
Few people today maintain 24/7 connectivity and run their own e-mail servers (although this is a p2p architectural possibility) but instead rely on the prevailing ISP and server-centric implementations. Most users can therefore retrieve or send mail only through their ISP mail servers, and they are often constrained to using ISP Internet access lines to do this—perhaps even “hard-wired” from home due to cable or other location-specific connectivity. For them, e-mail fails to be p2p technology, because of the indirectness of the transport and the reliance on remote server-hosted accounts to receive, process, and store content.
But this perspective is subject to change as the technology and architecture changes. Recall the telephony analogy for a moment; phone booth to phone booth between town squares, compared to individual roaming cellular handsets. The same individuals might be talking to one another in either case, but current cellular technology is more akin to what we’d care to call “p2p”.
In any case, it’s the experienced totality of the virtual connectivity model that determines the characteristic signature of the technology. The possible virtual models can be vastly different, yet use the exact same underlying physical transport infrastructure. Changing connectivity model is as easy as swapping out the software at the endpoints and can at least in theory be transparently moved across to different physical network technologies: standard phone lines, integrated digital service networks (ISDN) or digital subscriber lines (DSL), cable, wireless, satellite, optical fiber, power lines, and so on. The perceived network and its protocols don’t change, only the packet transport medium.
Original Chat and IM
The first widespread virtual p2p technology to hit the Internet was chat, which is a method of establishing a typed conversation between two or more users in real-time. A natural evolution of the operator-to-operator chat relay mentioned earlier, this chat includes any Internet-connected PC in a client-server connectivity model.
Internet Relay Chat ( IRC) evolved into server-mediated private or public virtual chatrooms, moderated or unmoderated, where many can converse at once. Thus it lost much of the original person-to-person aspect. Chatrooms are implemented by creating multiway connections from a hosting server and relaying any input to all clients participating on a named “channel” (a convention that allows IRC servers to selectively forward or not forward messages for different groups).
The original p2p functionality moved on into another technology, what is now called IM, although chat and IM modes are often combined in the same client software to the confusion of some due to the imprecise usage of the term “chat”. From the multiperson discussions, people can pair off for private chats or shift to an alternate technology IM connection.
Server mediation in either technology provides the transport protocol details that at least in theory allow two clients to establish a direct connection and participate in a private conversation. As shown in Chapter 6, however, some IM implementations retain the IRC characteristic of relaying messages through the mediating server, even for private conversations that are perceived as being p2p-connected. Despite this confusion, I have arbitrarily drawn the defining p2p line between IRC and IM—only the IM implementations are considered in any detail in this book.
Today’s chat and IM are perceived as p2p in the same way conversations on the telephone are, people talking to people—distinguishable in the strict p2p sense much as a rural-line, marketplace phone might be to personal cellular. The convenience of chat is significant to people who are nearby their computers, have an active (24/7) Internet connection and wish to exchange some quick words with someone. In addition, the social dimension that has grown up around chat is significant.
Chat technology resolves the individual addressing issue by maintaining a centralized directory to correlate a registered user identity with current online Internet address. This directory is automatically updated by the client software whenever the user connects to the Internet. Client update additionally resolves the issue of tracking user availability (actual and user-selectable) for messaging, notification, and other features particular to a given implementation.
IRC identity is essentially based on current IP and an arbitrary user “nick” (nickname or handle) that can be changed at whim as long as it is unique to a particular chatroom context. The main focus is on informal multiway conversations, not any lasting permanent network identity spanning different sessions. By contrast, the IM model relies on each user having a server-registered identity that persists across sessions, possibly supplemented with an arbitrary and visible handle chosen by the user. Chapter 6 examines a few IM implementations in detail, showing how their version of p2p works.
A probable convergence of text and voice messaging is on the horizon, as chat and IM merge with Small Message Service (SMS) in the next generation of 3G mobile services. Such a development is bound to have significant consequences for how people hold conversations, and we might well see a large-scale shift in some contexts from telephony to chat. Such a shift might critically depend on what kind of mobile user interfaces are designed for the devices—phone convenience aside, the typical cellular user interface is not pleasant to use for typing much of anything.