> Decentralized Interweb Ah yes, the technological solution. Though it should be remembered that technology rarely *solves* problems, so much as it *refactors* them. The main benefit of decentralisation as far as the internet is concerned, is that it allows us to replace a decidedly non-trivial political/managerial issue (namely, "how to get good stewards for online communities, and keep them that way") with slightly more feasible engineering issues (such as "how to make this network function" and "how to resolve routing problems in a reasonable timeframe"). Though we should distinguish between network-end decentralisation (we all connect to a forum called poat, but we do so via a mesh-net or similar) and service-end decentralisation (we all run our own poat instances, which exchange posts and comments via some p2p-ish mechanism), since the challenges faced by each would be different. It's also worth keeping in mind that many implementations of distributed networks will be suceptible to their own particular weaknesses - for example, in mesh-type networks, each node is dependant upon topologically adjacent nodes for connectivity, and is thus vulnerable to "cold shoulder" type attacks that could significantly limit or completely deny network access to the target. Retaliatory actions from there could quickly put overall network cohesion in jeopardy. The problem comes back to the people in the end. Pushing it back into a lower OSI layer just makes it less visible. > how in the fuck someone develops this perspective First, you've got to predate the world wide web. You've gotta spend several years on newsgroups, in the company of hackers, crackers, phreakers, and all the crowds of phonies, and through inertia and cocaine come to hold the position of a senior network administrator for a small to mid-sized IRC network, and then, with no particular attachment to any of this, you've got to watch that community run head-first into the problem and burn itself down as a result. Though credit to my fellow operators - we kept the fucking lights on until the bitter end. The people broke before the network did. > the numbers you use are way too specific They just look that way because, having ballparked them where I did, I stated them clearly and left any ambiguity either to the reader's interpretation, or - as you have done - for someone to just up and ask me about them. First-hand experience gives me a number of ~7000 +- 2000 for when the situation becomes critical, though this has a sample size of one and other people might naturally have other observations or data that dispute it - if they do, then I'd welcome the sharing thereof, as it may shine further light on the true nature of the problem. From there, consider the neurologically-determined maximum size of social networks - 250 +- 50 as a rough approximation of what I remember reading; also consider the typical size of most small villages and other likewise close-knit communities, which often stabilise at the low to mid thousands. There is a seperation of one order of magnitude between these two numbers, and I would conclude that, barring extreme cultural unity, there will not very easily be two. We can, for some practical data, look at reddit, where a number of its subforums have grown to multiple tens of thousands of users, and these are always either very tightly focused on a single subject, thus obviating some of the requirements vis-a-vis cultural cohesion; or they've culturally degenerated to meaningless posturing and mass-signal-posting. > Why specifically ten thousand? *Wànsuì, wànsuì, wànwànsuì!*