BitTorrent's Maelstrom browser enters open beta | Hacker News Reader

BitTorrent's Maelstrom browser enters open beta

techcrunch.com

66 pointssimas49 comments

A totally neutral decentralize network run on closed software run by a corporation? Forgive me if I don't buy into it.

Amen, I don't trust BitTorrent (the company) at all. Sync was a MASSIVE let down and it feels like they've backtracked on decentralization and are just trying to slap their name on a bunch of products that leverage bittorrent (the technology) in some way even when the products don't align with what, I feel like, the majority of BT users want (open/decentralized).

Silently bundling cryptomining software with uTorrent is what crossed the line for me (not that I had very high regard for the company in the first place).

I've since then switched to open-source alternatives of their products and will not touch anything else they make from now on, and advise everyone I know to do the same (unless they eventually open-source their products, which I highly doubt would ever happen).

Well, they weren't doing it "silently", users had to agree to install the miner.

Syncthing is great though, which is an OSS clone of Sync.

After I saw the awful security review for BT Sync, which is a massive closed-source C++ codebase, I switched to Syncthing which that is written in Go and posted on Github [0]. I find it has a better UI and it's more reliable than Sync in terms of connectivity/syncing speed.

Maybe some people will be inspired by Maelsto, create a nice clone, and release it open-source.

[0] https://github.com/syncthing/syncthing

[1] http://2014.hackitoergosum.org/bittorrentsync-security-priva...

Syncthing is great though, which is an OSS clone of Sync.

It doesn't offer anything near the features of Sync, requires that you have control over the firewall or the firewall is set up to use uPnP (which is not true for many networks I use on the go), is be difficult to install for friends or family I would want to share files with, and doesn't offer sharing invitation links.

Syncthing is great for a subset of use cases of BTSync, it's mostly pre-1.4 BTSync (minus the trackers and relay servers to make it usable when you do not have control over the firewall.).

After I saw the awful security review for BT Sync,

Well, it was mostly the security review that was awful (did you read it, of course BTSync accesses the update server). iSec Partners also did a review of BT Sync and found it to be secure.

Disclaimer: relatively happy user of Bittorrent Sync. Just not too happy with their monetization model, would have preferred purchasing the software, rather than a subscription model.

See, the advantage of BT Sync over Dropbox/Drive/OneDrive was that it was secure and I didn't have to worry about my private files being spied on.

With that main use-case in mind, why would you lock down the client to closed source? Why should I trust it at all?

Instead of all these audits that don't have much to go by, we'd have real audits with real results.

For me, I'm on ownCloud atm. I used BT Sync exclusively before I used ownCloud with VPS I have. It's serving me well, and the full-time availability is a plus.

feels like they've backtracked on decentralization

You can still use BTSync completely decentralized. 1.4-type shares are still available (shift-click '+') and you can disable tracker/relay server use.

But it would've been better if they continued on the open protocol/source path and earn money by offering encrypted read-only peers.

> But it would've been better if they continued on the open protocol/source path and earn money by offering encrypted read-only peers.

Yes this would have been ideal. I had issues when I tried BTSync before with too many shares bogging everything down so I stopped using it.

And BitTorrent isn't the company it once was, developing software and protocols with a vision and for a purpose. Just look at what a pile of infested crap uTorrent has become.

Yes, I do not trust any BitTorrent software. I wasn't being careful at installation and was shocked to find uTorrent changed my default search engine and added a bunch of bookmarks. I wasn't even aware it was developed by BitTorrent.

On the positive side: Transmission works great!

It wasn't. They bought it out several years ago. You can still find the last original version on the net somewhere - it's surprisingly good and has aged really well. In fact, I get better performance out of it than out of Deluge or Transmission.

Edit: here you are. you can still run it without instaing even: http://www.filehorse.com/download-utorrent/7858/

I'd just be a little bit afraid of security issues no? There have been a couple of uTorrent vulnerabilities in the past.

http://www.cvedetails.com/vulnerability-list.php?vendor_id=6...

Of course Deluge, written in python can not compare to uTorrent's tightly optimized C++ code. Transmission is better, but its back-end (libTransmission) is not as well optimized as libtorrent-rasterbar. Any C++ client based on rasterbar such as qBittorrent should offer similar performance to uTorrent.

If you don't require streaming, older uTorrect clients are still your best option. If you do, it IS possible to install the recent version of uTorrect without any external adware and to disable all of the banner ads.[1]

[1] http://forum.utorrent.com/topic/81421-321-how-to-turn-off-ad...

The performance of bittorrent has almost nothing to do with the performance of the language the client is written in. Its not an io-bound task. Performance in bittorrent comes from selecting the right chunks at the right time from the right peers. I think utorrent uses its own algorithm whereas both deluge and transmission use libtorrent's built in algorithm (which is configurable, but only to an extent).

Transmission and deluge often max out my bandwidth though, so its not like they're poor performers.

Deluge and Transmission do NOT use the same libtorrent back-end. Also, you may be right about Deluge not being CPU bound on your i7 based desktop, but I can tell you from first hand experience that both Deluge and the Java based Vuze were maxing out my Atom based home server. Switching to the C++ based qBittorrent, resolved the issue.

aria2c on command line. Fastest download application ever for everything including bittorrent and including metalinks.

Has BitTorrent the company ever not been shady? It seems like Bram Cohen made one cool open source project and then sold out as soon as he realized he could get rich off of it. Not that I wouldn't probably do the same but I'd probably draw the line at installing malware and lying to my users.

I have been using Bit Torrent Sync and I feel the same way. In action it works but in terms of philosophy it doesn't for me. Though the decentralized network could make all our concerns could totally be unfounded since if there was something not right we have the tools to find it through the bits it sends.

what is with the wining that bittorrent isn't open source? most of the software you use isn't FOSS.

they came up with great tech.

Why is Sync such a let down? It works very well for me. Not knowing what a software might be doing is problematic but isn't it the standard?

> most of the software you use isn't FOSS.

> Not knowing what a software might be doing is problematic but isn't it the standard?

This sounds so bizarre after using Linux for so long. Most of the software I use is indeed FOSS.

As far as BitTorrent Sync alternatives go check out Syncthing[https://syncthing.net/].

Does Syncthing work well?

It has some serious limitations compared to Bittorrent Sync. E.g. BTSync can relies on a tracker to find peers and uses relay servers in case port forwarding cannot be set up. Syncthing does not have anything like that yet, meaning that it cannot traverse many firewalls that you don't have under control. E.g. at work, all incoming connections are blocked unless they are initiated from within. So, Syncthing wouldn't work. It's pretty much the same with many hotels, etc. So, if you are on the go, it may not be any good.

Another problem that I have with Syncthing is usability. With BTSync I can be pretty sure that if I send a sharing link to family, that they will be able to install the software and use the link. Syncthing would be too difficult to set up for most family members or friends.

So, if you are looking for syncing some machines on a network under your own control, Syncthing is great. If you are looking for a Dropbox replacement that works everywhere with good sharing capabilities, BTSync is clearly the winner of the two.

I've been using it for about 6 months and I've been very satisfied.

As a matter of fact, how can we even know it does peer discovery in a distributed manner when we can't review the code?

Proprietary software is adequate for a lot of things, but when the main competitive advantages of your software is privacy and security, not being able to review the implementation makes it a non-starter for anyone who cares enough about privacy and security to want to use it over the competition.

I don't use mostly closed source software. I am more a Linus than a RMS but I use very few closed source programs.

1) Lightworks video editor on Linux and a open source project will happen soon with it.

2) I play closed source games

3) Outlook at work :(

4) Bittorrent Sync

5) Yeah that's all I got seriously

I guess mobile apps and websites aren't often open source, but just about everything else I use is too.

Only because it's standard are you not supposed to complain with something problematic?

Let's release the FOSS kraken and see if they turn the inspiration into something we all would like to try.

That's what I like to hear :) This sort of thing reminds me of MaidSafe, but I haven't really seen any traction in the years it's been around...

What's your hardware setup & web browser setup?

Cool idea but coming from the BitTorrent company? I'll pass. After what happened with the original BitTorrent Client and BitTorrent Sync I'd have to be an idiot to give them another chance.

Care to elaborate?

After they bought µTorrent they started bundling tons of adware, including, most recently, a bitcoin miner. Sync started out as a pretty cool decentralized self-hosted dropbox replacement that was incredibly easy to use. After promising the free version would never be limited they did a 180 and limited it to 10 shares when they launched a commercial version and ruined usability with a bunch of access controls and user accounts.

Obviously they have to make money somehow and it's perfectly within their rights to do so but they've shown that their business model has been to offer a product for free and then, once they suck people in, break promises and use underhanded ways to monetize.

Knowing maelstrom is likely to end up filled with ads or under an overpriced subscription model once it's done killed any interest I might have had in the product.

You can still create 1.4-style folders with 2.0. The 10 folder limit sucks and they broke their promise with that. I think lot of people would've been happier if 2.0 shares were limited to 10, but an unlimited number of 1.4 shares was still possible.

Thank you for the reply. Very useful to know. I never got to try out sync.

Copying my comment from the other thread today (which got lost):

I gave the beta a shot. It's a chromium fork with added support for browsing to magnet links. Basically, it's making BT a Web protocol. Not a bad idea.

Performance is inconsistent so far. Sometimes the downloads stall or fail for no apparent reason. Sometimes I can't find any seeders. Sometimes it works great. (Same old bit-torrent!)

Am I suppose to remember that long hash? I'd like if they provided us with some information how this works. Otherwise http://ipfs.io/ seems like a better approach that works in a way that I can actually understand.

It doesn't look like ipfs has human friendly named pointers yet either? At least that's what I gather from their documentation: http://ipfs.io/docs/commands/#ipfs-name

In answer to your question: bookmarks have been a long standing solution to remembering names. I absolutely understand that it's not the most ideal solution - but hopefully acceptable for a beta.

Unfortunately, the linked to TechCrunch article doesn't do much to explain how the browser works. One of the original-source blog posts does a (slightly) better job of explaining how things work - or at least points you in the direction: http://blog.bittorrent.com/2015/04/10/project-maelstrom-deve...

The hashes directly represent the infohash for the created torrent website.

Do you guys plan to release a video about it works? For example, how do you deal with content changing? If my index.html file changes, do I need to distribute a new torrent?

I mean, it's a cool idea and all, but why not release a thorough explanation perhaps in the form of a scientific publication?

Similar project, but OpenSource and works on any OS/Browser: https://github.com/HelloZeroNet/ZeroNet (Decentralized websites using Bitcoin crypto + the BitTorrent network)

Centralized decentralization

I would really rather read technical details about how this works. I looked into the mutable torrent specification (that lets you make changes to a torrent) as a way to create a bittorrent "hosted" blog, but mutable torrents ultimately need a feed hosted by an http server, which destroys the decentralization. I am not on Windows, I cannot try this out if it's currently Windows-only.

Regardless of which way you look at decentralization, there has to be an origin, and due to the nature of these networks, enough active peers that "cache" the content. By having a hosted http feed for at least a small percentage of the time you guarantee the network at least 1 active peer.

Personally I can see decentralization becoming a much bigger thing - I don't want to get into the politics of the likes of China, or domestic net neutrality. But these will drive the smartest minds to create a decentralized self maintaining network - think bitcoin.

Forgive my ignorance, but could you not use the same concept as the blockchain?

That probably actually would be possible by emitting a negligible transaction from an address under one's control that has the SHA1 sum of the torrent encoded inside the OP_RETURN. This would be sufficient to construct a bittorrent magnet link to download the post/data/whatever. The bitcoin address itself would act as the feed. This doesn't rely solely on bittorrent, but nevertheless that does seem like a workable, decentralized solution for publishing a bittorrent feed. Bitcoin solves another problem, thanks!

Another neat side-effect of using a bitcoin address like that is that people could make donations to you through it :-)

Not really. The blockchain is meant to work for transaction type of information, not documents. It also keeps all of the transactions ever made. This type of web would probably contain a lot more information than that. Therefore you wouldn't want every peer with a copy of the whole web.

The case where this is relevant is in decentralized source control. The only project I know of that set out to tackle this is dead in the water:

https://www.kickstarter.com/projects/612530753/gitchain/post...

I continue to be surprised that the Web standards folk and browser vendors haven't (AFAIK) pursued a standard for content addressable resources backed by a common DHT. There's Magnet, but it doesn't have much traction outside of the usual P2P clients.

Reply on news.ycombinator.com