3 ms·

Building a connection across a continent just by hopping from mesh node to mesh node is like trying to drive across a country by just using dirt paths. It is cl

by WJW 9y ago

Building a connection across a continent just by hopping from mesh node to mesh node is like trying to drive across a country by just using dirt paths. It is clearly not impossible, but there is a reason that most people choose to use highways for long distances. It is similar with a several-hundred-hop long path through a mesh network vs a single hop over a suboceanic fiber connection. In particular, step 5 is untrue over large distances such as across the Sahara. Step 6 is true, but does not necessarily lead to mass adoption. Step 7 is feasible, but costs a LOT of money at scale. Also, it seems that economies of scale would rapidly push out all inefficient means of communication in step 9 (assuming that sailing ships across the oceans and/or launching satellites will continue to cost money) and we would end up back with suboceanic fibres. Finally, while the "next billion" may not be attached to their unlimited netflix just yet, it seems highly unlikely that people in developed nations are willing to accept drastic slowdowns in their connections. Streaming video takes twice the bandwidth (both downloading and reuploading) as required for just watching on EVERY mesh node between the viewer and the producer. Caching might reduce this a bit, but is by definition not effective for "long tail" content. I'd be curious to see how long the meshes hold together when it turns out you can't see Game of Thrones because the next node over has no more bandwidth to spare.

gregknicholson 9y ago
> step 5 is untrue over large distances such as across the Sahara Yep. Focusing on local, social connections makes this less relevant. > Step 6 is true, but does not necessarily lead to mass adoption. Yeah, there's more scope for growth in places that don't already have abundant internet, where the pull is practical. But there are also a growing number of enthusiasts for whom the pull is principle, and they can help make mesh networking viable for poor people in rich cities. > Step 7 is feasible, but costs a LOT of money at scale. Step 7 is what you and I are doing right now :) — using the existing internet. This point was supposed to mean: the existing internet, can co-exist with and compliment mesh networks. Staltz's [Multi-modal connectivity](https://viewer.scuttlebot.io/%25OryfD6hGpjiqwFkKyFnhVPZMPsZFgdFutEhzxycyolQ%3D.sha256 https://viewer.scuttlebot.io/%25OryfD6hGpjiqwFkKyFnhVPZMPsZF...) puts it more eloquently. > it seems highly unlikely that people in developed nations are willing to accept drastic slowdowns in their connections. Never underestimate the value of cheapness! I know some people who don't have home broadband; they just use their phone's data connection. It's slower, but it's adequate, and it's cheaper. Mesh networking doesn't have to completely supplant ISPs. I want it to become a viable alternative, suitable for certain people depending on their priorities — like Linux! > Caching might reduce this a bit, but is by definition not effective for "long tail" content. If we're talking about data storage here (the flatbed truck — as opposed to data delivery, the series of tubes) then HTTPS can still be a highly-available option of last resort. If you want people to see your content, and no-one else is interest in replicating it, you can host it yourself or pay someone else to. Maybe use hashbase.io; they seem generous. If you mean data delivery, the packets have to traverse the internet somehow; it's just a question of whose box they go via.
staltz 9y ago
OP here. I find more than 4 hops impracticable. That's not what I was suggesting in the article. I suggested (probably not clearly enough) three ways meshes can be efficient: (1) local-first social networking where the use case is "download latest updates and read later" much more than it would be like instant messaging, (2) gossip, the eventual propagation of data over long distances, (3) satellite-based meshes to cover large distance hops.

by WJW 9y ago

gregknicholson 9y ago

> step 5 is untrue over large distances such as across the Sahara Yep. Focusing on local, social connections makes this less relevant. > Step 6 is true, but does not necessarily lead to mass adoption. Yeah, there's more scope for growth in places that don't already have abundant internet, where the pull is practical. But there are also a growing number of enthusiasts for whom the pull is principle, and they can help make mesh networking viable for poor people in rich cities. > Step 7 is feasible, but costs a LOT of money at scale. Step 7 is what you and I are doing right now :) — using the existing internet. This point was supposed to mean: the existing internet, can co-exist with and compliment mesh networks. Staltz's [Multi-modal connectivity](https://viewer.scuttlebot.io/%25OryfD6hGpjiqwFkKyFnhVPZMPsZFgdFutEhzxycyolQ%3D.sha256 https://viewer.scuttlebot.io/%25OryfD6hGpjiqwFkKyFnhVPZMPsZF...) puts it more eloquently. > it seems highly unlikely that people in developed nations are willing to accept drastic slowdowns in their connections. Never underestimate the value of cheapness! I know some people who don't have home broadband; they just use their phone's data connection. It's slower, but it's adequate, and it's cheaper. Mesh networking doesn't have to completely supplant ISPs. I want it to become a viable alternative, suitable for certain people depending on their priorities — like Linux! > Caching might reduce this a bit, but is by definition not effective for "long tail" content. If we're talking about data storage here (the flatbed truck — as opposed to data delivery, the series of tubes) then HTTPS can still be a highly-available option of last resort. If you want people to see your content, and no-one else is interest in replicating it, you can host it yourself or pay someone else to. Maybe use hashbase.io; they seem generous. If you mean data delivery, the packets have to traverse the internet somehow; it's just a question of whose box they go via.

staltz 9y ago

OP here. I find more than 4 hops impracticable. That's not what I was suggesting in the article. I suggested (probably not clearly enough) three ways meshes can be efficient: (1) local-first social networking where the use case is "download latest updates and read later" much more than it would be like instant messaging, (2) gossip, the eventual propagation of data over long distances, (3) satellite-based meshes to cover large distance hops.