Hacker News Reader

Comment by m12k | Hacker News Reader

Does anyone know of a successful server-based peer-to-peer project other than TOR and cryptocurrencies? I'm all for breaking free of the stranglehold of centralized proprietary platforms, but it also seems to me that most of the open source projects that ever really succeed at competing against entrenched proprietary incombents are the ones where a centralized service is being provided by a non-profit (e.g. Wikimedia Foundation, Mozilla, Canonical). Maybe expecting people to set up a server is just a bridge too far? Maybe there's just too many conveniences in centralization for todays users to put up with the hassle? Maybe there needs to be a centralized server that is as easy to use as the proprietary ones, and then the hardcore users can set up their own and federate with the central one?

View on HN

BitTorrent? For a while it was consuming a large fraction of the public internet bandwidth worldwide. This might still be the case.

Although paid streaming has certainly taken a chunk out of its market share, there remain to be many things that can't be had on the big streaming providers. For that, there is always the p2p system with the deepest content selection :)

"successful server-based peer-to-peer" - it's certainly P2P, but I thought the whole point of Bittorrent was to avoid needing servers?

At least in private tracker land a lot of us use seedboxes. Install some autodownloaders and plex on it and you've got your own private netflix with any content you want.

Avoiding servers might be the whole point of bittorrent per its creators but us end users just want a good media experience. Server, no server, legal, illegal, pay server company, pay media company, whatever. Best experience > all else.

Doesn't that also disqualify most cryptocurrencies?

As examples of successful server-based p2p projects go, I would have picked Matrix and Mastodon. Not as successful as e-mail, mind you, but steadily growing.

Matrix is envisioning a transition to a complete p2p system in the future, if that ever proves possible.

Some stats are available on https://the-federation.info/info , such services have been recently coined as the "fediverse", and ActivityPub is a standard that emerged out of the will to unify those services and make them interoperable: mastodon, peertube and diaspora are interoperable, AFAIK.

E-Mail

Right, and web servers too - neither had much competition when it launched though, so the point about no being able to beat incumbent proprietary solutions still stands. Not to mention the majority of people get their email hosted by a proprietary service these days.

>neither had much competition when it launched though

I'm sceptikal as well, about a federated protocol being able to replace an existing entrenched player. Buy I still hope.

>Not to mention the majority of people get their email hosted by a proprietary service these days.

But all the proprietary services talk to eachother, and, mostly, talk to private instances as well. There are absolutely allowed to be proprietary PeerTube instances.

This is interesting as when the subject of setting up your own server comes up, people always complain how much of a hassle it is to maintain sender reputation with Gmail and Outlook.

This leads me to two thoughts/questions:

1. Maybe successful decentralization schemes should be made providing for some peers that will be private for-profit? I know that (for example) IPFS encourages building services.

2. What could we do to ensure that small private peers will continue to be viable, even if some large for-profit peers will dominate? Of course such measures should be reasonably resistant to being used by bad spammy/illegal actors etc. ("Bad actors being bad" is always a handy rhetoric for overcentralization, but the problem itself should be obviously addressed.)

>Maybe there needs to be a centralized server that is as easy to use as the proprietary ones, and then the hardcore users can set up their own and federate with the central one?

I think if you take care of the basic concepts, the rest is easy. Technical details about server organization are not really that important. The root problem seems to be that people have trouble looking past the app/program that they interact with.

It is still quite common for people to think of the web as IE, Chrome etc. Because it is for them. It is a legitimate thing to think. There are undoubtedly people out there that think of Gmail as the thing that they use to communicate with people. That is unless they again consider the browser as the thing that does that.

I have a friend who refers to the XMPP network as "Xabber" simply because that is the app she uses to chat with me. She is entirely uninterested in any details of the network. Suggestions that she switch to a better XMPP client are met with confusion.

Perhaps we need to teach the concept of a "protocol" in school. There are a lot of important concepts we fail to teach people these days. That is probably one of them.

> Perhaps we need to teach

Perhaps we need to start accepting that not everyone wants to know the particulars of the things that we care about. The need to teach new things is one of the big forces standing in the way of wide adoption of a lot of open solutions.

Are you familiar with the proverb "Give a man a fish and you feed him for a day. Teach a man to fish and you feed him for a lifetime" ?

It's not about teaching everyone how to setup a PT instance, but how things works for the best outside a totaly centralised and company-owned system.

When someone is not interested in learning you're not going to be able to force them. In my experience you either make peace with that or live a frustrating life.

> Maybe expecting people to set up a server is just a bridge too far? Maybe there's just too many conveniences in centralization for todays users to put up with the hassle?

The users expected to set up servers should be tech-savvy people (not necessarily with any programming experience) wanting to do something for their online and offline communities. This should be like hosting forums, which is still a thing here and there. Or WordPress.

This infrastructure should be in place when some major event strikes that would drive people from centralized platforms (compare https://news.ycombinator.com/item?id=21480623 ). Of course who knows what will happen. But at least, we should have a durable infrastructure as an alternative to big corporate players, even if only some people will use it in practice.

It would be interesting if Alphabet would support some PeerTube-like initiative at some point, as an anti-antitrust measure.

Napster was so successful that the RIAA filed law suits to shut it down.

Wasn't Napster based on a central server and then a bunch clients who found each other via that server and then connected p2p? (And since the central server was the weak link that law enforcement could go after, Bittorrent evolved to get rid of the need for it). If so it actually further underscores the point about a central server providing ease-of-use and not being able to expect 'normal' people to set up servers themselves.

Yes, and PeerTube also uses a central server as matchmaker model, but in this case the match could be made to another Federated server on a different host.

>a successful server-based peer-to-peer project other

BitChute certainly, if only in part. The website proper is centralized, but the video delivery is ran via BitTorrent. Currently the video quality is kept somewhat low (720p IIRC), but this means initial wait time is pleasantly short, on the order of 2...5 seconds.

Nextcloud. Or depending on how you define a server, Bittorrent.

I didn't know NextCloud was P2P (I thought it was just based on a central server with a bunch of clients) - what's the P2P part?

It's not peer-to-peer, but you can share files between Nextcloud instances.

You can share content with users of other instances, or import content that is being shared with you into your own account. This is federation, but still quite limited.

Example: I can share a folder from I@myserver.org with you@yourserver.net

Syncthing is equivalent to the file-sync portion of Nextcloud and really is p2p FWIW. Been happy enough to replace my Onedrive subscription and once set up, just works.

Check Awesome P2P: https://github.com/kgryte/awesome-peer-to-peer

I like https://datproject.org and the idea of https://beakerbrowser.com (with Beaker you can setup a website without needing a server)

Edit: There is also an Awesome Dat at https://github.com/dat-land/awesome-dat/blob/master/readme.m...

Dat is cool, but it's hardly "successful". Who is even using it other than Beaker and hobby projects.

Invisible Internet Project, i2p: https://geti2p.net.

Not sure if you would consider it successful though.

>Does anyone know of a successful server-based peer-to-peer project other than TOR and cryptocurrencies?

Gun is being used successfully by a number of companies and projects AFAIK. Internet Archive, Hackernoon, NotABug.io, DTube (maybe they aren't anymore, not sure).

It performs better than webrtc in some use cases too

It's merely federated, but IRC doesn't get mentioned enough.

e-hentai with H@H[1]

[1] https://ehwiki.org/wiki/Hentai@Home

> Does anyone know of a successful server-based peer-to-peer project other than TOR and cryptocurrencies?

Git

Reply on news.ycombinator.com