On Wed, Feb 10, 2016 at 9:09 AM Ted Smith <tedks@riseup.net> wrote:
I'm a little skeptical of wireless mesh networks as a general solution to this sort of problem, because they're inherently chatty, and have very limited reach.
I think a better solution is local wired networks with something like Freenet running over them providing distributed censorship-proof storage. The next challenge is to synchronize contents between local Freenet darknets over sneakernet, which I don't think has been done.
Interesting. I would have thought wired for longer distance communication, with local mesh networks for the "last mile." Do you have some thoughts on how to set up a local wired freenet? Back in the days of modems over content-oblivious phone lines, freenets were easy because phone lines in bulk were pretty cheap per line, and pretty much everyone had a land line. In this day and age of video and high-resolution images, that no longer seems like an option. Whenever I think of "mesh" I think "mobile mesh," which is just part of the picture. The mesh networks I've actually seen in action are static "meshes", which are only really called that because the ISP puts omnidirectional antennas on some of their customers' roofs to daisy-chain other customers off of, not because they use any kind of dynamic mesh routing protocol. This approach might actually work for setting up a freenet: negotiate directly with local businesses (particularly public places like coffee shops) and residents to set up local hotspots, then put up high-gain directional and omnidirectional antennas on a different band or at least channel to handle the longer distance links. Of course, the businesses/residents involved are probably going to be primarily interested in access to the "regular" Internet, but that doesn't stop you from using strong encryption on the links themselves, tunnelling onion/mixnet networks, offering local services in private IP space, etc. Over time, you can replace tunnels over the government-controlled Internet with wireless links or leased lines with strong crypto. It won't stop the network from getting shut down if the links are specifically cataloged and targeted, but it seems like current "internet kill switch" proposals primarily focus on the layer 3 infrastructure, not on the link or physical layers. And certainly not on RF. But even if you do that, if you are offering a "public" service, and you actually get a lot of users, it seems likely you'd end up being a victim of your own success, noticed by regulators and then forced to comply with the same regulations that cover the Internet, including any retention and "kill switch" requirements. You might be able to mitigate some of that by encouraging mac address spoofing and end-to-end encryption, but that doesn't stop a government from taking over one of the local services and putting a gag order on the operator, then tracking someone to a specific coffee shop and sneaking up behind them and pinning their arms to their sides so they can't wipe their machine. Having a dead man switch/"panic button" on any local servers, routers, etc might help with that, particularly if there's plausible deniability there. "Sorry, can't comply with your retention order because our hard disk just died and it's going to take us months to get the server back online." It works for the Chicago PD with their constantly "malfunctioning" dashcams, so why not for us? One side benefit of bringing back freenets in a big way might be the ability to push back against regulations that have compliance costs. If a bunch of people voters care about are dependent for their Internet access on freenets that would have to shut down if they are required to retain logs for years and set up systems for easy law enforcement access, politicians might think twice about pushing for such things. Provided they can't get away with characterizing freenets as dens of "hackers", drug dealers, terrorists, and child pornographers of course. Which is why it's important to remain connected with society and not become hermits hiding out in Internet backwaters. There's always more social/community-oriented solutions ala Max Hernandez's Thieves Emporium where you have secret access points and people need to be vouched in, but that is incredibly hard and would likely remain quite small, because the probability of infiltration approaches 100% once you get into the thousands of participants. And of course, if it's small and secret, nobody will notice when it and its operators/users quietly get disappeared. Obviously these approaches are orthogonal to one another, and you can use any combination of them.