I have talked with pembo a little bit about possibly making Phuks into a decentralised p2p type site. Similar things already exist. Aether is one that has been around for a while now, I remember trying to connect to it during the Reddit blackout and I couldn't connect, probably because they were having the same mass influx problems as Voat. I recently came across another called Zeronet, which is a little different due to it being a kind of network of sites within the same client, so it isn't just an aggregator, it has other parts too. It doesn't have that much content at the moment though.
I have also been reading about other kinds of decentralised networks like pirate boxes, and something called the Briar Project, which is a mobile meshnet kind of thing that is (apparently) used by journalists and activists during internet blackouts/for secure communication. I don't fully understand how it works, but it essentially creates a huge LAN between phones, and it also uses bluetooth in some way.
My question is this: Will some form of this technology eventually replace ISPs?
If you were to take a densely populated developed area like London, for example, would it be viable to create a LAN for the entire city? It seems to me like it could work with current technology. There are mobile devices metres away from each other, without even taking computers into account. If it was possible to do this, what would it take to get people to make it happen? It seems like not having to pay an ISP would be a major incentive, but there is also the privacy aspect too. Over the next few years, laws relating to the internet are probably going to become more and more invasive. This has basically already happened in the UK.
Could the internet of things actually serve a purpose in a project like this?
There are cameras, fridges, toasters and all sorts of ridiculous things that are connected to the internet now. If a lot of appliances had the capacity to join/support a network like this and the range was improved, would it be possible to have a decentralised world wide LAN? Is it even a good idea, considering the security risks that it could produce?
It seems to me like it could be part of a natural progression that is happening. We ditched wired connections for wifi, and now there is potential for something else.
Decrentralization is good but distributed computing is the future, in my opinion. Email, the Web are arguably decentralized, for example, while Aether, Bitcoin, Ethereum are distributed. Many decentralized projects like Owncloud, gnusocial, diaspora, are really federated- they still require a server, but anyone can set up a server and the reachable network extends beyond any one server, exactly like email and the web.
Distributed computing is what drives bittorrent's magnet links. No one can take them down because the data is distributed amongst the peers that make up the swarm, as long as the swarm exists, the data will be there. The data is copied and reallocated to different peers as the swarm grows and shrinks.
A big problem with current distributed computing (DHT based projects, like magnet) is that a malicious entity could subvert the system if it can control a significant number (majority) of peers in the network. The core problem is that an identity has power (rating discussions, voting, popularity ranking) within the rules of the system and that identity creation is anonymous and trivial. Sort of like how mass unrestricted account creation thwarts Voat's account restrictions by upvote farms. Most attempts to make identity creation harder end up with centralization on an authorizing entity, like a certificate authority. Some, like the blockchain, require proof of work, largely a waste of computing power.
More and more algorithms are being developed which enable complex behaviour by a swarm of simple nodes. The hashgraph algorithm is exciting in that it claims to solve the byzantine generals problem (ensuring success in an unteliable network with potentially malicious peers). It is similar to the blockchain, the peers create a signed, encrypted message to add, and gossip about the addition with other peers and, crucially, they gossip about the gossip from other peers. Eventually, each peer is able to verify that all peers know about all messages, and they all reach consensus on the order of messages.
Meshnets:
I can't see meshnets replacing ISPs anytime soon. ISPs exist to connect to high bandwidth low latency backbones and meshnets are exactly the opposite: low bandwidth, high latency.
The physics of a meshnet are a not insignificant obstacle to low latency operation, too, because a message must be routed amongst more peers than in a backbone network. There are fancy distributed algorithms for constructing spanning trees that could create optimal low hop paths through the network, but still the range of a wireless peer is less than a wired ISP network; low range = more hops = high latency. Wireless connections also suffer from interference when in high densities, already problem in apartment blocks with wifi.
Bandwidth for meshnet peers is also a problem, but perhaps less so since many paths can be chunked together when transferring information. The main problem is that the bandwidth of a path is capped by the slowest peer in the path. Again, distributed algorithms could come to a consensus on optimal routes involving the high bandwidth peers to achieve better throughput. But then these algorithms basically recreate backbones through the peer network.........
Where meshnets might work is in communications which are primarily local. If 99% of the information you access is within the city, then the number of hops to get to it could be low.