A decentralized web would give power back to the people online

+1 for bringing up mesh networks. I have been reading about mesh network projects in Germany and other places. What I think is needed to increase adoption is something like the raspberry Pi, a cheap and open mesh router that would allow setting up community networks, with hooks for volunteers to provide low bandwidth gateways to the web, tools for sharing locally hosted copies of Wikipedia, projects for local social media, market places for local commerce, etc.

When there is a low cost system available, then I can visualize myself starting a class at the local library, etc.

In an increasingly centralized world (central banks, the awful one worked government philosophies, winner take all hubs on the web, etc.) it is too easy to lose track of the fact that local issues, economy, and life are more important than global issues.

> Most conservatives would say the papers and media outlets are _all_ liberal. Would you expand on that spectrum?

Really? Isn't Fox ranked at like number 2? If anyone think that's liberal they need to realign their gauges

Successfully attacking this problem means progress on numerous fronts: technology, connectivity, bog-simple configuration and operation, recoverability from user or system/hardware errors. Whittling these down over time may work. Most significantly, hardware costs are not presently constraining, limitations lie elsewhere.

Google and Facebook's advertising support strongly tend to centralisation.

Kicking the legs of advertising out from under the stool of publishing might be a means of attacking that particular tendency, which would make a more distributed technical model much more viable.

There's the question of just how distributed you'd want a system to be. There are problems at both low and high levels of centralisation. A fully decentralised model might have a tendency to go rogue, or be subject to petty dictators. It's been interesting to note that even a well-capitalised entity such as the NSA is reported to prioritise specific and generally small values -- single-digit, often single-hand -- of systems for specific attack and interest. There's an argument to be made that strategically weak targets, such as the perennially troubled Yahoo, headed by the morally compromised Ms. Meyer, can be made to crumble where a more commercially robust firm, say, Google, Apple, or Facebook, would be willing to resist. I'm mindful of this.

There's some reason to believe that perhaps a suite of Free Software alternatives would offer more robust installations, and compromises would impose high workfactors, for any given would-be surveillor. The question of creating systems by which data exfiltrations might be more readily detected is another area for exploration.

Going back to the publishing world, there's a long-standing practice of including intentionally fictitious entries within compilations, maps, etc., whose observation in the wild would indicate copying. This might be worth pursuing on a formal basis, particularly through such indicators as financial accounts, email or communications addresses, URLs, etc., extant as canaries which would reveal a data breach.

A huge problem for any widely distributed infrastructure is maintenance and administration. I'm reluctantly concluding that a world in which individuals and households host and administer their own personal data servers isn't viable. I'm not abandoning all hope, but it seems a difficult problem, and one which the Internet of Shit seems determined to prove intractable.

Much of this has to do with the financial underpinnings of the infotech world. As several recent authors have noted (Paul Romer, Jeremy Rifkin, Paul Mason), information-dense goods function poorly in a market-based economic system. They have numerous characteristics which make for poor price discovery and dynamics, with informational asymmetries, heavy up-front (and hence average & fixed) costs, low-to-zero marginal costs, susceptibility to low-cost copying by others (notably China, though this is a practice with ancient traditions -- see Ha-Joon Chang and Frederick List), diminishing marginal returns, long-tail support obligations, and unintended and non-evident consequences (to vendors, users, and bystanders).

A few possible models for a largely-distributed personal informations ervice suggest themselves.

One is the residential-server-as-utility model. Comcast has had such an offering for nearly a decade, that I'm aware of. Essentially, it's a set-top box which can take on additional responsibilities, including home automation and security functions. There are several other utility-type providers who might offer similar capabilities.

There might well be ad-hoc collections of friends or neighbours. A tech-oriented person could easily provide services to hundreds or thousands of others on commodity equipment. The main limitations here are trust and discoverability.

There are arguments for making email into a government-provided service, via the post office. This introduces the risks of government surveillance (already an issue with postal systems), though the protections of legislative restrictions, and public-sector union whistleblower protections, against gross abuse. The fact that physical mail delivery already ensures a brick-and-mortar point of presence in virtually any habitation means that one of the perennial problems of information technology -- establishing, asserting, and recovering identity -- can be achieved through a local visit.

The early Internet spread through a set of social institutions, largely universities. These provided points of access, administration, and accountability to populations of users ranging from a few hundreds to a few tens of thousands per site at major public universities. Whilst this was not an all-encompassing level of provisioning, it is a model of access and social organisation which might be useful to draw on for a more modern implementation.

Public libraries, as an alternative to universities, might offer another option. They already serve as an internet access option for a significant population.

Banks, schools, major retail establishments, and religious centres might be other options.

I'm not sure what exactly will work, but the dynamics I'm pointing at here involve:

Technical capabilities and equipment. A very low bar, and getting lower all the time.

Connectivity. Slightly more difficult, given distribution, land-use, and reliability concerns, but still generally tractable.

Trust. A major factor, especially in a world of eroding social institutions and values. This plays into the dynamics of various systems providers / maintainers quite heavily.

Workforce technical capabilities. Information management skills are remunerative, and could prove difficult to retain. This might change, possibly rapidly. The present market is, however, exceptionally geographically centralised. A distributed node-and-service model might allow for greater flexibility in specific node administrators and technical staff to operate in a much broader choice of areas, including those with lower housing costs than New York City or San Francisco, closer to family or hometowns, etc. I expect some de-skilling of routine obligations, but technology tends to require at least a modicum of fire-fighting capabilities, and perhaps some engineering and planning capacity as well.

Identity and pseudonymity. Both matter. Balancing the risks of data disclosure with those of data loss gets complex. Principle control over data, plus remote backups, ensures against physical loss. Encryption with some model of key escrow, in which a set of parties, several with strong and vested interests in personal privacy can balance privacy against the problem of loss of specific tokens (physical or otherwise).

Ease of use / access. This stuff has to be bog-fucking-simple to provision, deploy, and maintain.

Copyright and copyright compliance. Ultimately, my view is that we're going to have to recognise that attempts to re-bottle the genie, and provide centralised or authorial control over reproductions is doomed. Information-as-public-good strikes me as increasingly inevitable. (Which isn't the same as "inevitable", though it's getting warmer.)

Misbehaviour. People can be asshats. Another reason for forms of key escrow and the like is to allow specific access to specific individuals' information in specific instances, subject to very strong controls.

Search. Traditionally this has relied heavily on data-center based operations. There are distributed alternatives, but none I've seen are particularly adept.

So: yes, challenges, but the fundamental capability (hardware) is well in hand. Spitballing at approaches should eventually see something stick.

No worries. I never repeat repeat myself.

May be yes, may be no. It really depends from who owns agenda of decentralizing the web.