News for hackers. Stuff that matters.
Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by jsaxton86 on Sunday February 16, 2014 @12:04PM (#7248266)
A large scale mesh network is interesting to me for a couple of reasons:

* Consumers would no longer be dependent on monopolies like Comcast for internet service (OSI layer 1)

* It is much more resistant to eavesdropping and censorship (OSI layers 2-3)

These seem like two totally different problems to me, and I'm still struggling to understand how the physical layer part can work on a large scale.

I get how they could work on a small scale. For example, you could easily set up an ad-hoc wifi network in a college dorm for, say, file sharing purposes. But in most other settings, I don't get how it could work. Most people don't even talk to their neighbors, so even if you lived in a neighborhood that had a high enough density of nodes for this to work, why would you want to share a little ad-hoc network with your neighbors?

Of course, wireless networks alone aren't going to be sufficient to build a large-scale mesh network that could disrupt existing ISPs. The latency alone makes this infeasible. So you need someone to invest in physical links, especially physical links that cover long distances. That's why the BTC part of this proposal is so great -- it provides a way to compensate these people. Of course, this is just a slightly different problem than the one we have today, but instead of being stuck with Comcast we're stuck with whoever runs these physical links.

Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by JoshTriplett on Sunday February 16, 2014 @01:14PM (#7248516)
> It is much more resistant to eavesdropping and censorship (OSI layers 2-3)

Not inherently. Arguably, a large-scale mesh network would be more susceptible to interference, due to the larger number of distinct parties involved in each communication. The difference is that because a mesh network assumes the lack of trust in the intermediate nodes, a sensible design for one has to include end-to-end security.

Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by jewel on Sunday February 16, 2014 @06:37PM (#7249854)
> Most people don't even talk to their neighbors, so even if you lived in a neighborhood that had a high enough density of nodes for this to work, why would you want to share a little ad-hoc network with your neighbors?

I've been thinking about this recently. The idea is that it's transparent. Imagine, for example, if Apple started shipping this on all iOS devices, or it became a common android feature. Here is a blog post I wrote on the topic last month: http://stevenjewel.com/2014/01/android-mesh/

Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by lotsofmangos on Sunday February 16, 2014 @12:14PM (#7248319)
wireless networks alone aren't going to be sufficient to build a large-scale mesh network that could disrupt existing ISPs. The latency alone makes this infeasible.

I'm not so sure, your links are almost c, whereas optic fiber is substantially slower, so it depends entirely on how little processing you can get away with in the message passing code.

edit - was just thinking about this, an arm cortex runs at ~ 2,000 MIPS, which gives you just under 15 cm per instruction for radio in air. Fiber is about a third slower, so your links do not actually have to be massively long to give you quite a few cycles to play with.

Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by ris on Sunday February 16, 2014 @12:33PM (#7248376)
It's got nothing to do with the propagation speed. Wireless has high latency because it requires comparatively high amounts of processing, has to operate over a shared channel (requiring collision avoidance), is subject to unexpected interference...
Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by adestefan on Sunday February 16, 2014 @01:26PM (#7248552)
You also need to send a lot of additional information for error correction.
Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by jsaxton86 on Sunday February 16, 2014 @12:31PM (#7248373)
I think it mostly depends upon the range of each of the nodes. If your nodes have an average range of 820 feet (the range of 802.11n (in good weather, I assume)), and you have 1 node exactly every 820 feet between NY and SF, that's over 16,000 links. If each link has 10ms of latency, that's 16s of latency.
Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by ZenoArrow on Sunday February 16, 2014 @04:34PM (#7249419)
16s of latency is interesting to think about, do think we'll get around this issue with local web page caching (obviously won't work for non turn-based games). SyncNet or something similar could prove very useful... http://jack.minardi.org/software/syncnet-a-decentralized-web...
Re:Open Libernet: a Bitcoin-based fully encrypted mesh netwo... by rmrfrmrf on Sunday February 16, 2014 @01:42PM (#7248644)
I think you have your OSI layers mixed up.