[HN Gopher] Show HN: Decentralized, Mutable, Serverless Torrent ... ___________________________________________________________________ Show HN: Decentralized, Mutable, Serverless Torrent Swarm Websites Author : publiush Score : 108 points Date : 2021-12-10 18:26 UTC (4 hours ago) (HTM) web link (github.com) (TXT) w3m dump (github.com) | nynx wrote: | This is what decentralization is about, not crypto. | xwvvvvwx wrote: | this literally uses a cryptocurrency (handshake) for domains | iskander wrote: | I think the evolution of "decentralized" infrastructure will | start to bring out a lot more overlap between "traditional" | decentralization communities (building stuff like Beaker | browser) and some of the useful bits of crypto. | dannyobrien wrote: | Yes, I think this is both the biggest opportunity, and the | biggest challenge, especially as I think there's been a | growing separation between those two communities in the | last few months and years. There's so many good ideas and | implementations (and investment in harder problems of | distributed systems) in the crypto/blockchain/web3 space, | and a lot of hard-won experience and genuine applications | in, as you say, "traditional" communities. It's just a | matter of finding some sort of common ground. | | I do think that the https://getdweb.net/ community is a | model of how that crossover can work. It's also something I | think a lot about at FFDW, which because of IPFS and | Filecoin, has its feet in both camps. | publiush wrote: | There's a dilemma as the magnet links/hashes aren't easily | shareable. One option is to create a DMT directory, but this | would be centralized. Handshake is the most mature | decentralized domain name project, and I opted for it. It | uses coins to limit abuse, since anyone can flood a | decentralized system. You don't need coins to browse and use | federalist. That said, if there are any other immutable DNS | systems that aren't centrally controlled that I could review, | I'll definitely take a look! | panick21_ wrote: | Have you checkout out the Gnu Name System from GNUNet? | publiush wrote: | The website for GNUnet seems to be down/404, but it looks | like ownership of names is controlled by a central | authority | (https://manpages.debian.org/unstable/gnunet/gnunet- | namestore...). | anchpop wrote: | There's ENS, which seems on sturdier footing than Handshake | to me, but the gas on ethereum is ridiculous | throwaway94294 wrote: | Isn't this just trusting whoever runs https://query.hdns.io ? | | https://github.com/publiusfederalist/federalist/blob/3670867... | publiush wrote: | Good catch! | | It's temporarily using hdns.io as many people still do not have | an hsd node installed. A later version will be shipped with a | light weight SPV resolver, at which point, the last piece of | the puzzle will be complete. | mrtksn wrote: | What are some cool websites that I can visit right now? | publiush wrote: | It's new, so I don't think there are many out there yet. I do | hope that this changes the landscape of the ecosystem for | 'decentralization.' It's not supposed to be about tokens or | organizational marketing hype. First, it's about free speech | and freedom of information. "De-platforming" is now extinct. | Only then, can you even begin to discuss anything else. | | I'm very thankful to WebTorrent, DMT, Handshake and Electron | for making things possible. | mrtksn wrote: | Thanks, the work done is definitely cool but, as with any | other decentralization project, I'm yet to find the content | use cases. | | The only two kinds of successful content types that I'm aware | of are pirated movies on Torrent and cryptocurrencies on the | blockchain. They all depend on centralized discovery(torrent | websites and exchanges). | | Which makes me wonder, are these decentralized websites or | social media platforms attacking the right problems? | rektide wrote: | > _I'm yet to find the content use cases._ | | It's an alternative transport, not application layer. If | you can imagine doing anything with the world wide web , or | ftp, you can imagine what this is for. It can transport | hypertext files around, or whatever other file type you | want. | | In the web case, it'd just be some local http files you | could open in a local file origin. There'd be no server. | But that's still a way to exchange whatever art or media | you could ever imagine. | | We are bounded only by imagination. The internet is built | around the Internet Protocol (IP), a way of streaming data | arbitrarily from one computer to another. It has been up to | us to imagine uses, to chase new possibilities. What do you | think the use cases for IP are? Can you see what that made | possible? Your question is in effect that, and trying to | grasp at how broad, how possible, how potentiated this | great work is is dauntingly hard, for we could share any | type of content we want with either. | publiush wrote: | This is a project I've been working on to create decentralized, | peer to peer, "serverless" websites using several technologies | including webtorrent, dmt (mutable torrent BEP-46), and handshake | decentralized domains. | | You can create either an immutable (uneditable) torrent site, or | you can create a ed25519 keypair and create a mutable (updatable) | torrent site. | | This is great for blogging, whistle blowing, and other things. It | also scales well since torrent technology is great although it | had previous been pigeonholed to other use cases. | | Please give it a try and let me know your thoughts! I don't take | any credit for this since I just weaved the great technologies | others already made together! | RobLach wrote: | Very cool and actualized proof-of-concept. "Decentralization" | beyond a marketing term. | derbOac wrote: | How would you compare this in a use-case sense to something like | Beaker browser? | | Also -- and I'm unfamiliar with handshake -- but is the sort of | thing that could work over ad hoc networks? | publiush wrote: | The Beaker browser looks interesting -- it looks a lot like | DHT/DMT (https://hypercore-protocol.org/), but I speculate it's | likely more specialized for the use case. | | Handshake names are decentralized and on chain, so as long as | you have access to read the chain, it would work over ad hoc | networks as well (and offline/local). | oscargrouch wrote: | This is cool. | | I'm also working on a decentralized distribution mechanism based | on torrents, and while i've being working on a different | architecture, the network mechanism are basically the same with a | couple of differences on the network level. | | In my case i'm working on a very customized version of chrome | where the web api is actually available for native applications | beyond Javascript. | publiush wrote: | That sounds interesting! I would love to hear more! | timbit42 wrote: | Would this work over I2P? | publiush wrote: | BitTorrent works on I2P so I believe this should as well but I | haven't tried. | champagnois wrote: | These sorts of things are still blockable through DNS and IP | filters cutting off access to the root nodes and such I would | guess. | | As such, what is the use case? | publiush wrote: | Seeds for the torrents can change and, although not | recommended, in terms of bloating the Tor network, using Tor | will also help. With the recent attack on exposing Tor users, | it may not be a bad thing if everyone starts torrenting on the | Tor network actually. | | Everything, of course, can be blocked at some point, but the | thing to remember is that there are other kinds of contracts | that existed long before smart contracts - and these things are | already leveraged in society. | | The internet isn't going to disappear tomorrow, but I would | agree there are longer term risks, so let's build today. | | There's been a lot of talk about decentralization lately, and | Cypherpunks write code. So I wrote code instead of a blog post. | rektide wrote: | Wow that's a big huge dollop of nihilism, of why even bother. | It's very unclear what your slam even means, what you are | trying, technically, to express as the problem. Which makes | your rejection even harder to handle with faith & respect. | | I also don't think it's accurate. Handshake is a | cryptographically certified way of establishing identity . | Since torrents are now mutable via handshake, it seems like | webtorrents can be updated & moved as needed. Further, peer- | exchange processes mean that having the initial seeds up | probably isn't even a requirement. Even if one particular ip | address or site gets cut down, the swarm can use other | webtorrent trackers to re-spawn & carry-on. | | Aside from your criticism being either inaccureate or | misleading, I'd also say the use case doesn't need to, imo, be | 100% perfect in every way to be worthwhile. I'm glad someone | did seek better, & bothered. A decentralized, updateable, | browser-based torrent is incredible leap for a web of data, for | interconnection. Even if it's not 100% completely uncensorable, | it's many leaps in the right direction, towards | decentralization. Especially decentralization without | coordination/consensus, which I think is great & vast | improvement over the harsh & strict type of computing that | *coins have dominated the field with. This work is far more | interesting to me. | champagnois wrote: | You are assuming bad faith here. | | The project is marketed as unblockable, etc and yet we both | conclude it is blockable from the start with existing network | management tools that are deployed in places like Iran and | China. | | I was asking a genuine question of use case. | rektide wrote: | > _and yet we both conclude it is blockable_ | | Extremely polarizing language for something that is way | more a scale to me, and I don't agree that this technology | is blockable. I already elaborated some scenarios where | hosting can respawn easily, with no coupling to any | particular infrastructure. That to me is pretty | unblockable. There's other resilliencies offered here. | | The story keeps getting better, given all the resilience | features at Bittorrent's core: webtorrent clients could | connect to multiple trackers just fine. Since it's tracker | based, there's no need for any specific host to stay | online. We can all just use the existing tracker network. | | Even if the entire tracker network is taken down- something | that has never happened to bittorrent & which is beyond | imagining to me- bittorrent still has a peer exchange | network & allows for peers to manually be added. Whence | peer exchange can kick in & keep resilliency going. | | I'm sorry but I have no agreemenet whatsoever with you on | blockability. This has multiple vast layers of resillience | that are excellent, which have never seen active threat | against them. | | What would make you happy? Do we need a system able to | resist long term nuclear winter to be unblockable? What's | the goal here? You've still been extremely unclear what | your technical complaints actually are, extremely specious, | & aggressive. And now you are also putting words in my | mouth. | | > _You are assuming bad faith here._ | | I think I've done quite well working around doing that, & | worked hard to find something to discuss. In contrast, you | have yet to specify a specific technical case at all, to | explain what issue you actually have or what you think the | problem or scenario you are imagining is. And you use | aggressive position & hard words to double down on your | argumentless snub, which to me reads like a bad faith | follow-up, one I challenge you to do better on. | | I mostly think you don't know the tech. Which is fine. | Bittorrent is fairly complex & has a lot of layers to it. | There aren't great primers to get filled in on it. | Bittorrent is a great basis, one that countless millions | have been spent trying to block & take down & combat, but | the system has been extremely resillient. Magnet links | "just work", fantastically well, and are super easy to | share, to get started, over any medium one wants. The | underpinnning distributed P2P technologies are broadly | capable of taking over from there. I don't think your | short, undetailed, savage doubt is contributive or | accurate: you should better define your issues if you are | going to throw such a hard heavy opinion down against great | tech. | dang wrote: | Please don't argue in the flamewar style on HN. We want | curious conversation here. | | https://news.ycombinator.com/newsguidelines.html | vgb2k18 wrote: | Block-resistant and censorship-resistant are terminology | I've seen used in other projects. I agree that ublockable | and uncensorable are hard phrases to sell here, they kind | of immediately beg to be challenged. | michaelscrypt wrote: | How is this different from ZeroNet [https://zeronet.io/]? | generalizations wrote: | Also, it's a variation on the concept. It's really good when we | have multiple implementations of a general design. | sprash wrote: | Seems to be not much different from zeronet. It even makes the | exact same mistakes like not building in anonymization by | default (preferably using i2p). Zeronet hovever is much more | mature and has a very active community. | ccakes wrote: | At a quick glance, no Bitcoin | nephanth wrote: | Iirc, zeronet used to not have bitcoin before the craze, so | this would be like early zeronet I guess | zcw100 wrote: | No but Handshake does use coins which I assume to be some | sort of blockchain "Handshake uses a coin system for name | registration" | algo_trader wrote: | Well done, good luck. Always wanted to implement something like | this. | | > dmt (mutable torrent BEP-46), | | Is dmt supported (made scalable?!) by the existing torrent infra | structure (DHT/trackers/etc??). Sorry if this question doesnt | make sense. | algo_trader wrote: | Also, is this suitable for publishing an RSS-like feed which | out polluting the world with a new torrent for each atom? | publiush wrote: | Yes and no - DMT (https://github.com/lmatteis/dmt) is | implemented into this so you can use a single hash in the DHT | for the 'site', but that will be updated to point to a new | torrent infohash on every update. | easrng wrote: | It depends. You an either have your mutable torrent point to | the head of a liked list like this: head | | |- post.txt +- prev.torrent | prev | |- post.txt +- prev.torrent | | You get the idea. With the not yet widely-supported | BitTorrent v2, you can just add files to a new torrent and | seeders of the old torrent will seed the files that are also | in the new torrent too. | toomuchtodo wrote: | What are the challenges slowing the v2 deployment? Client | uptake? | publiush wrote: | Thank you! I believe it's somewhat scalable, but one thing to | note is that DHT itself is not as fast as using a tracker. | f0e4c2f7 wrote: | Seems cool at first glance! I've never heard of Handshake, will | be looking at that more in particular. | | So I see the address in the screenshot starts with federalist:// | | Is it possible to access these sites with a normal web browser | from the internet too? | oscargrouch wrote: | from the look of it, you will need to have the handshake | resolver installed and replaced as your local dns resolver. | With that it will return you a the public key address on the | bittorrent DHT which can be solved by some torrent client. | | There it should have a torrent info payload that your client | can turn into a ordinary torrent, where you can proceed to | download the files, and can open in your browser on your local | filesystem. | publiush wrote: | Thank you! | | I do think there is a way - since the beautiful WebTorrent | (https://github.com/webtorrent/webtorrent) can do so in | browser. I'm keen to see something like this in a normal web | browser (if possible as an extension even), hopefully developed | by someone with better skills than me haha! ___________________________________________________________________ (page generated 2021-12-10 23:00 UTC)