Hacker News Reader

Comment by dbingham | Hacker News Reader

Decrentralizing the web's software isn't good enough. We need to decentralize the hardware. Right now, connections to the web look like a tree, where a whole bunch of connections get funneled through an ISP. That ISP has the power. The power to throttle, the power to block, the power to record. And that ISP can be pressured by other powers. We need to decentralize so that instead of looking like a hierarchical tree, the internet looks like a graph. With each building forming a node that connects to its neighbors.

Of course, the amount of work it would take to build such a web and move to it likely rules out the possibility of it ever happening. I mean, how do we go about forming a movement to build this? It only works, really, if everyone's on board.

View on HN

>Right now, connections to the web look like a tree, where a whole bunch of connections get funneled through an ISP

This is very true. More so if you look at countries like the UK where ISPs have recently been consolidated into MAFIAA-complying monopolies that censor just about anything that Hollywood and the government ask them to. It's scary how close ISPs are to the 'governements' of the UK and USA. Obamas' recent NSA 'reform' proposal even suggested that ISPs be in charge of surveillance & storage of Metadata instead of the NSA! WTF? Fascism.

Mesh networks have a big role to play in decentralizing the the post-NSA internet. Micropayments through bitcoin and cryptocurrencies will encourage individual nodes to provide bandwidth to users [1]. Smartphone apps like OpenGarden [2] and systems like the Athens Wireless Metropolitan Network (AWMN) [3] are the frontrunners of the new internet.

[1] https://crypto.cs.virginia.edu/courses/14s-pet/2014/01/30/aw... [2] http://en.wikipedia.org/wiki/Open_Garden [3] http://www.motherjones.com/politics/2013/08/mesh-internet-pr...

I'd think content providers in general would be willing to provide bandwidth under a mesh network. They'd also be the ones most likely to pipe the mesh into the existing infrastructure. Who else besides the likes of Google / Netflix / Facebook would front heavier duty APs city-wide. Every ounce of traffic would have to be encrypted though. The main barriers for mesh networking when I look at it are:

First and foremost: Antennas and hardware in general. It's one thing to maintain a multi-channel 500mbps connection within 10-20 feet, let alone doing it with multiple devices simultaneously connected to other multiple devices. Even if you scale the speed down, the consumer grade hardware out there isn't targeting this use case.

Second: Protocols. If we had the hardware, the whole mesh has to be trustless while being affected as little as possible by connectivity and pathing overhead. Plus backwards compatible with existing infrastructure (which is less hard I guess).

Third: Gatekeepers. If you get them on board, creating the mesh as a natural extension of the internet becomes easy. Since they invariably won't be on board, it'll be that much harder.

You have some valuable points.

Naturally the internet is not a mesh network because nobody will take this role of gatekeeper without benefit. If somebody does, it becomes ISP again.

However, although the current web is messy, we should be able to organize it using distributed network solutions in a peer-to-peer manner, as proposed in the blog here: http://bit.ly/1e8vrLL. It serves an additional layer on top of the nature internet. It has the protocols built-in and the "Gatekeepers" sitting in between the distributed nodes and a feasible business model to support its execution, rather than the ideal Semantic Web solution requiring DRM.

I sometimes wonder about this. Pretty much everyone already has a connected computer in their pocket. Wouldn't it be nice if we could use the phone without a cell provider? The web without an ISP? Connect to our friends without a social network? Exchange money without a bank?

Thinking further, what if all these services could be plugged into a well-abstracted peer-to-peer network, consisting of every connected device in the world? Services similar to Twitter or Facebook would no longer require a central host. Redundancy would be built in. Uptime would be pretty much guaranteed. Ads would go away. Freedom would be an implicit part of the system; no longer would profit motives sully (or censor!) services that people use and enjoy. And it would be more natural, too: pumping all our data through a few central pipes makes a lot less sense than simply connecting to our neighbors. Our devices would talk to each other just like we talk in the real world, only with the advantage of being able to traverse the global network in a matter of milliseconds.

New technologies like Bitcoin and BitTorrent would crop up naturally as part of this arrangement. It would put power back into the hands of the people.

Sadly, it seems a future of locked-down app stores controlled by a small number of large corporations is much more likely.

(Sorry, I realized this started sounding like a Marxist rant halfway through. And I'm not even a Marxist! It just sucks that this fascinating future is one that we're not likely to experience, even though we're almost at the point where we could actually implement it. I think.)

Your idea is similar to the existing distributed search engines which are up and running on many individual pcs of a P2P network.

Unfortunately, they are not quite practical either due to the security (open local firewall) and duplicates from every node, etc. It's hard to control without a centralized facilitator. However, agent technology provides a solution for that. Check out here: http://bit.ly/1bvvIJ3

Where I live at least, nearly everyone has a wireless router, which hardware-wise should have no trouble talking to other wireless routers in nearby buildings. Often, there are even multiple wireless routers in the same building, which can't talk to each other except through the ISPs, which I find quite ridiculous.

So, in urban areas of the US at least, it really is just a matter of software.

Of course, a large scale wireless mesh network would probably be somewhat lacking in terms of performance, but we also still have wired connections to ISPs and could start setting up wired connections to each other. Peering with your neighbor (wirelessly or not) to share idle bandwidth shouldn't make things any slower than they already are...

The problem of course is that everyone would have to use a compatible mesh protocol. There are several options for a mesh network protocol now, but I don't know of any that scales well while remaining decentralized. Even if someone did come up with a good protocol now, considering how long the switch to IPv6 has been taking, I wouldn't expect any switch away from IPv4 to happen very soon.

> So, in urban areas of the US at least, it really is just a matter of software.

and, if you're really zealous about it, the firmware. there was a high-priority FSF project a while back for free wireless router firmware. that listing mentioned something called "orangemesh," which now appears to be on hold. what are some of the other options out there on the software side?

Unfortunately I've noticed a major obstacle to getting nearby hardware talking to each other: WPA.

Seriously, why prevent people from using a network? If you're relying on WPA for security then you shouldn't be connecting to the Internet in the first place. The most common thing I hear when guests see that my WiFi is open is that people will 'steal the bandwidth', as if the local exchange will magically run faster if there are enough residences connected to it! Also, given the monopoly position of phone provider, getting everyone to throw more money at them for the privilege of having multiple overlapping, mutually-exclusive networks is not a good plan.

Yes, probably too difficult to move entirely, but you can build it off of the current web, and just migrate it over, or: You don't move a mountain, you move stones.

Just throwing an idea a the wall here: Maybe it could be combined with the decentralization of the power grid, i.e. smartgrids? This would (a) give additional incentives to home owners (subsidizing their investment) and power companies (going after the broadband market) and (b) it would solve the question on where to get the bandwidth from.

One of solution is a different protocol: http://named-data.net/

This looks extremely interesting, basically the entire network works like a giant CDN.

Yes indeed, NDN is a serious contender for the future Internet architecture. There are still some privacy concerns (due to the ubiquitous caching), and some performance/throughput issues due to signature verification on content packets (which are all signed), but lot of research has been done on ICN and I really would love for it to be the next Internet.

Reply on news.ycombinator.com