Hacker News Reader

Comment by mnzaki | Hacker News Reader

We can easily be running our own internet. I recenlty discovered CJDNS[1] (protocol for encrypted p2p address allocation and routing for mesh networks, so essentially OSI layer 3) and project Hyperboria[2] (a community of local WiFi initiatives) while exploring scuttlebutt[3]. Confirmed my suspicions that we could probably be getting waaay better internet connectivity at way lower prices.

Decentralized tech is going to be the future. Now if we can only figure out how to not get squashed down by the powers that be.

[1] https://github.com/cjdelisle/cjdns

[2] https://hyperboria.net/

[3] p2p social network https://scuttlebutt.nz

View on HN

"trying not to get squashed by the powers that be" is the future. Only going to get harder each year.

Which is why we should be scrambling to get it done now. Free and non-centrally-controllable access to connectivity is going to be a powerful aide in all the other battles.

Priorities priorities priorities.

Andre Staltz gives[1] an approximate deadline of 5-10 years, or we're doomed.

[1] https://staltz.com/a-plan-to-rescue-the-web-from-the-interne...

I'd give you the same deadline of 5-10 years for figuring out a way to hold off the automation of military and security operations, which will be just as bad as losing the open web.

Non-centrally controlled connectivity can be solved by low-altitude communication satellites[1], and other sky based solutions[2].

We should be focusing on tech that makes aerospace less costly. The benefits will extend beyond a free & open network. There is good science and theory on ways to reduce costs in aerospace, it's an engineering/business problem at this point.

[1]https://www.cnbc.com/2017/05/04/spacex-internet-satellites-e... [2]https://medium.com/iot-5g-extreme-ideas-lab/googles-balloon-...

Unless you plan on conjuring spectrum out of thin air that's still got some serious barriers. Most of the frequency bands are pretty crowded already.

How are LEO satellites not going to be centrally controlled? It's not like the satellite owners will let you use them for free or be controlled by commercial interests and subject to laws and regulations.

You are also not going to be deploying your own balloon Internet.

1. There can be multiple sets of satellite services an area. How many entities currently own satellites? Maybe 100. Reducing the cost will increase diversity in satellite ownership.

2. Satellites can, in theory, run a mesh network among themselves.

3. Decentralized doesn't mean you own one. Decentralized means there are multiple agents and none have a commanding control of the system. I don't need to be able to deploy my own internet balloon to belong to a decentralized system. Cheap internet balloons will allow smaller companies to work on the "last mile" problem, further decentralizing the network.

4. Decentralized doesn't mean free as in beer, free from commercial interests, or free from laws and regulations.

Satellites are expensive and frequencies are limited. Even if you solve the first, the latter isn't about to change.

Demand is also limited. Not a great combination, so your vision on multiple low cost satellite service providers isn't likely to come true.

There are only a few satellite owners you can buy broadband service from, not a 100. It's capital intensive and past bankruptcies don't bode well for new entrants.

Satellites can and do run satellite to satellite comms, a "mesh" if you like, but it increases latency.

Neither are balloons likely to be a viable long term solution.

> CJDNS[1] (protocol for encrypted p2p address allocation and routing for mesh networks, so essentially OSI layer 3) and project Hyperboria[2] (a community of local WiFi initiatives)

None of this addresses the big point, which is the backhaul, the transit networks, the big networks which knit together all of the smaller community networks and make them more useful.

In short: If you make a nice little mesh network in your small town (New York City, say) you still haven't solved the problem of getting access to websites in Los Angeles, let alone Doha.

Nobody talks much about that. Probably because it isn't possible for that to be decentralized or open in any meaningful sense, and it's certainly impossible for hobbyists to build it.

The mesh networks target the last mile problem. That is the big point to be addressed as it causes the affordability issues and is the main connectivity bottleneck. To improve the existing system full decentralisation is not necessary.

Exactly. It would allow companies like Google to get involved by leasing out their fiber backbone at cost for long-distance comms.

Right now they're restricted by last mile costs, but otherwise they're incentivized to simply provide the fastest bandwidth to the most people, as their revenue simply scales with internet usage as a whole.

I believe it is already being done: https://en.wikipedia.org/wiki/Long-range_Wi-Fi#Large-scale_d...

We'll still need big orgs to take care of those undersea cables, for now at least; but even with that still out of reach for hobbyists or community projects, the situation can already be much better than what is being forced on us.

On another note, if you are using p2p tech like scuttlebutt and DAT[1] and Beaker[2] browser then just having good neighbourhood and city wide meshes will already be a YUUUGE boon :D since they share data locally, and for a lot of uses you don't really need access to servers across the oceans

[1] https://datproject.org/

[2] https://beakerbrowser.com/

As a side note, it always bothered me how terribly inefficient it is to be sending data in circles around the world just to message someone two streets down or, worse, my house mate in the other room......

The decentralized future is going to be a lot more efficient, a lot more reliable, and even faster.

There's no monopoly in backhaul, in fact various market shenanigans have reduced long-distance data profits to pennies. The internet is designed entirely with the idea of making it easy to connect the first mile to the rest; that is a solved problem. In USA the remaining problem is local service provided by monopolists. This problem will persist in general until FCC is forced, kicking and screaming, into allowing more equitable use of the radio spectrum. Until then, as TFA notes, this problem will be solved piecemeal in those communities who are allowed to run their own services.

Yes, wholesale broadband rates are dirt cheap and falling like a rock forever because of improvements in technology.

What you pay for for broadband in the home is: 1) last mile connection (expensive to install) 2) bribes (sorry, "franchise fees") 3) advertising to maintain the illusion of competition 4) customer support to help you deal with your anger and cancel/activate you when you switch between various shitty providers to maintain the illusion of competition

What kills me about the advertising are three things:

1) It creates a huge barrier to entry because new ISPs have a hard time getting any name recognition.

2) Newpapers, radio stations, and television stations probably wouldn't want to say anything bad about telephone or cable providers. Imagine the phone call an NBC TV station would get if it attacked Comcast. That advertising is a good chunk of money.

3) Advertising is rolled into the PUC-approved "cost of doing business" and the cable or telephone companies get to add their profits on top of that.

We’re working on www.altheamesh.com to create networks where participants within the network can compete with one another to provide better service without the end consumers having to switch providers.

This is accomplished by adding a price metric to the routing protocol so that packets are routed along the best and cheapest paths.

So if you notice that a certain neighborhood has bad service, you can set up some kind of connection to it and sell into that network to make a profit. No advertising. The end users will only notice their internet access getting cheaper and or better.

1. build local net

2. feed it with a few links to the central net, but mostly transit the data locality

3. wait until ISPs goes out of business

4. plug isolated nets into new dark fiber lying around.

done.

How do you figure that when 70% of traffic is streaming?

if you are just consuming content owned by netflix or such, a central or decentralised net means nothing to the end user.

Which was just about my point. As this covers the vast majority of all end users, how do you figure your list logically applies?

Regarding a more decentralized web, I recommend this talk: https://media.ccc.de/v/34c3-8740-the_internet_in_cuba_a_stor...

This was great; very inspiring!

Also a shame they are not using much better tech (proper mesh protocols like CJDNS), which is already available and successfully deployed.

In Germany we have Freifunk [0], a very active community providing free WiFi and meshing in many places.

[0] https://freifunk.net/en/

This was my first ever look at mesh networks, years ago!! Met a bunch of people hacking on it in c-base. Those guys are awesome! And my go-to example for large successful installations of CJDNS over wireless mesh (40k+ nodes AFAIK!). Also I'm under the impression that Freifunk consider themselves part of Hyperboria, but I'm not so sure actually...

People are usually impressed just looking at the network map https://www.freifunk-karte.de/

Hey, add http://altheamesh.com to your list. We are creating a system where routers pay each other per byte for bandwidth, and packets are routed along the best and cheapest routes.

To address CJDNS, it’s a project that inspired a lot of people, but it’s not actually used for internet access. It’s mostly used by people who are connecting to it over their existing ISPs. This is because it is not good at finding efficient routes. People trying to set up an efficient network will use distance vector protocols like Babel or Batman. We have some patches to Babel that enable our price based routing. To my knowledge CJDNS is not used by any mesh networks actually providing people with internet access.

Why people connect to CJDNS over their ISPs, I’m not sure, but who am I to judge what people do in their spare time. My guess is that it’s easier than actually doing the work of setting up a real network but they can still feel like they are part of a “mesh network”.

Cool project (altheamesh.com) but please don't spread FUD about CJDNS...

What do you mean by "internet access"? If you mean access to the "commercial internet" then I assure you a lot of CJDNS peers provide gateways to the "clearnet", as they like to call it, and traffic gets routed to it when needed.

A lot of the nodes are connected over existing infrastructure because infrastructure is hard; but yet there are already several dozens of thousands of wireless mesh nodes running CJDNS. For one example of a network (that is not over existing ISPs): https://www.freifunk-karte.de/ and I'm sure you can google for more. The CJDNS project clearly promotes setting up wireless mesh over going with your existing infra.

I believe the flexibility to be part overlay part independent at the same time is a powerful boon to the project actually. Perhaps that's why there are 100s of thousands of CJDNS nodes around the globe?

Regarding routing, CJDNS uses "source routing" which has various advantages. All the satisfied nodes running CJDNS don't seem to be complaining about the routing efficiency.... BATMAN is superior when nodes are actually changing location (i.e. mobile mesh networks), which is not the intended use-case here (CJDNS nodes are at fixed locations and peering connectiongs are set up manually by exchanging keys out-of-band). Also BATMAN is only a routing protocol whereas CJDNS does address allocation and cryptographic peer auth as well, and the routing protocol can actually be changed anyway.

One final note, I personally prefer networks where the incentive to support the network is simply because the members want to see it succeed (mutual benefit all around), not because each node wants to accumulate cryptocurrency.

And it seems to be working just fine without cryptocurrency being involved. Call me naive but I think it's possible to motivate people without money.

>What do you mean by "internet access"? If you mean access to the "commercial internet" then I assure you a lot of CJDNS peers provide gateways to the "clearnet", as they like to call it, and traffic gets routed to it when needed.

In this case "internet access" refers to access to the big backbones of ISPs that allow CJDNS peers to exchange data. If you don't have an ISP to connect you up to the larger networks a CJDNS peer will not be able to send any data anywhere.

Or how does CJDNS send data across the transatlantic fibre without an ISP?

It was sorta rhetorical to make a point about 'internet' meaning just interconnected networks :D And any wireless mesh based network will face the same issue (my reply was mainly about FUD against CJDNS, and some implication that http://altheamesh.com/ is superior because ????)

But anyway, access to the commercial internet and sending data over the transatlantic obviously need "normal" ISPs, but as others have noted below (replies to https://news.ycombinator.com/item?id=16135626) that's not the "big issue".

Why would I imply that Althea is superior to CJDNS? They are not the same kind of thing. I was relating my experiences as an engineer that assessed CJDNS for my project. You’ve somehow taken it as an attack on a sacred cow.

Are you implying that Freifunk uses CJDNS? This is not true. I am not “spreading FUD” about CJDNS. Althea is not superior to CJDNS, because it’s not a competition. We looked at CJDNS when building Althea and found that Babel or Batman (which is what Freifunk actually uses) were much more efficient and effective. They also don’t roll routing, address allocation, and crypto into one huge ball of code.

People tried a long time ago. What was needed was an access point with two radios (so you can bridge without interference) and with enough range to hit a few neighbors (in case your nearest aren’t participating).

I think we are there now but the people who were interested in mesh have lost interest.

Those are some pretty noble ventures, but how can something with a name like scuttlebutt ever hope to become mainstream? There's a p2p torrent video service called bitchute, which is equally noble as the services you mentioned, but it too is just so not-sexy - it'll never make it to the mainstream. I think these services could eventually take off, but the marketing and branding are going to be parts that need the most work.

Anyone have experience with hyperboria? Looks like there is a node in SF, I'd be interested in setting up a node in the south bay.

Reply on news.ycombinator.com