I was reading a story on Slashdot recently concerning the initial results of their move towards a subscription-based model when it suddenly dawned on me just how revolutionary concepts like P2P and distributed-computing models like Gnutella or Freenet actually are in light of this economic problem. The reason for moving towards a subscription model for any website, especially a high volume site such as Slashdot, is to pay the upfront and ever increasing cost of hosting with finite amount of resources. Having a website costs money in terms of physical plant (the hardware involved with serving the site) and the delivery mechanism (the bandwidth used to serve the site). Indeed, with the exception of labor, these categories are a fair assessment of the majority of costs associated with having a site.
Imagine, however, if those non-human costs, both computational and bandwidth-oriented, could be shouldered in a fair and equitable way by the very users who consume those services? In a P2P / Distributed Computing Environment, that is exactly what happens. The distinctions between client and server no longer apply as each are expressed as individual endpoints in a vast matrix of other equally powerful endpoints on the public Internet. Every node on the network shoulders the responsibility for storage and bandwidth by allowing a portion of its processing time and physical storage to be used equally by other nodes whose responsibility is the same. When a request for service is made to the network, it doesn’t trickle to a single entity as it traditionally does in a centralized model. The request will literally cascade to other peers who may or may not have parts of that service on their local machines. Peers that don’t have the requested resource may in turn request the same service of other nodes and so on. The concept of the world wide web deservedly assumes its moniker in this environment. By transforming the already present population of underutilized peer nodes(any system that connects to the network), the possibility to utterly transform the landscape of what we know about what the Internet *is* looms closer.
Just as distributed computing has helped the SETI at home project, the vast amount of computer resources readily available in the world goes largely untapped. In fact, the raw computing power of today�s low-end consumer processors dwarfs the power of central processors that exist in most mission-critical mainframes. Where mainframes truly obliterate consumer PCs is in their ability to perform massively parallel computations. Similarly, a distributed network model could have the same performance impact we see in the comparisons between PCs and mainframes.
Distributed filesystem solutions like Freenet, which focus primarily on physical file distribution, are already in their beta stages. The next steps would entail the ability to create dynamic, data-driven applications that are the majority of today�s Internet, and deploy them using the Distributed Computing model. This is no small undertaking. It may involve a massive paradigm shift in our thinking about how web applications are programmed and architected. And then again, maybe not given human innovation.
Imagine now, framing our original problem and the possible solutions that P2P offers, that Slashdot exists now not as a physical location of servers or even a central point of information, but rather as this ethereal namespace, an true Internet Service Layer, where a piece of the physical site may reside on your machine and on many other desktops, a portion of the bandwidth is shouldered by your machine and everyone else connected to the network.
There is no more need to move to a subscription model because everyone who uses the service pays for it already with their participation in the network. The basic economic impetus that has taken away so many good websites has been removed. The kaleidoscope has been shaken. It is the equivalent of winning the Internet lottery. After that, what are the next steps? It is a very powerful question.
