ATTENTION: You are viewing a page formatted for mobile devices; to view the full web page, click HERE.

Main Area and Open Discussion > General Software Discussion

BitTorrent Inc. earns $15-20 mn; makes uTorrent Ad-supported to "keep lights on"

<< < (2/4) > >>

40hz:
^@Wraith - I stand corrected. Apparently they already have such cozy arrangements in place with several TV and motion picture studios. Hard to believe something that makes that much sense business-wise actually came to pass.

Live and learn huh? :-[

nosh:
Update:
uTorrent Makes Ads Optional Following User ‘Revolt’
http://torrentfreak.com/utorrent-makes-ads-optional-following-user-revolt-120815/

Stoic Joker:
You just gotta like those revolting users.  :D

f0dder:
--ability to add/remove files and folders from an existing torrent that is being shared live already.-superboyac (August 13, 2012, 10:02 AM)
--- End quote ---
Not possible without major protocol changes - and I'm not sure it could be done in a nice way, tbh. There's both performance and security concerns here. One of the nice things about torrents is that the torrents are identified by a cryptographic hash of all the included files (well, technically, blocks) which makes it hard to fake data for a given torrent. And IMHO, the protocol doesn't really need to deal with this anyway, it could be used to deliver changesets instead (but yes, you'd need to grab a new .torrent for the changesets).

--ability to add/block users on a per torrent basis-superboyac (August 13, 2012, 10:02 AM)
--- End quote ---
You can ban IPs with the existing system, isn't that good enough? The torrent protocol doesn't have a concept of 'users', and that's a good thing IMHO.

superboyac:
--ability to add/remove files and folders from an existing torrent that is being shared live already.-superboyac (August 13, 2012, 10:02 AM)
--- End quote ---
Not possible without major protocol changes - and I'm not sure it could be done in a nice way, tbh. There's both performance and security concerns here. One of the nice things about torrents is that the torrents are identified by a cryptographic hash of all the included files (well, technically, blocks) which makes it hard to fake data for a given torrent. And IMHO, the protocol doesn't really need to deal with this anyway, it could be used to deliver changesets instead (but yes, you'd need to grab a new .torrent for the changesets).

--ability to add/block users on a per torrent basis-superboyac (August 13, 2012, 10:02 AM)
--- End quote ---
You can ban IPs with the existing system, isn't that good enough? The torrent protocol doesn't have a concept of 'users', and that's a good thing IMHO.

-f0dder (August 15, 2012, 02:59 PM)
--- End quote ---
I wasn't really trying to discuss it on a technical level, just a conceptual one.  I'm saying the torrent system is really good regarding a couple of characteristics and adding a couple of features to it would push it to the next level.  Whether or not the protocol can do it is not an issue.  Then a new protocol can be created.  Nothing is really that impossible, and nothing I described is really out of this world.  People are already trying to do it, look at Retroshare:
http://retroshare.sourceforge.net/
It doesn't work that smoothly, but it totally has the right idea.

Torrent isn't the only option or foundation to build on here.  That's just an example.  There's also DC++ which has its own advantages/disadvantages.  First, it's really great for sharing files with a private group...think irc crossed with emule.  it could use more flexible options for user management and per-user sharing permissions.  Retroshare is trying to do all this, it's just klunky right now.

Here, I'll just post a recent thing I sent to a friend of mine:
Napster:
Napster was great because a centralized server indexed the files of all the users.  The central server didn't store any of the actual files, only the index.  The files were still on other people's computers.  I'm not sure why it was illegal because only the index was under Napster's control, but whatever.  The important feature to take away from Napster is it's indexing.  The index of shared files is stored on the central server...keep this in mind.  Also, the speed at which you can download is restricted by the upload speed of the user you are getting the file from.  Most residential upload speeds are very slow (usually around 20-30 kBps).  The good part of Napster is that you can easily search for ALL the files and the interface was clean and easy to use.  The bad part is that the upload speeds can vary wildly and is usually going to be far slower than what your max download speed is.  Also, things can appear to be available but they really are not, which is frustrating.  This is what I call "reliability".  For file sharing, you want to reliably know if something is available or not, and actually, you also want to know how fast you can get it.  if something is very slow, for all practical purposes it might as well not be available.  If I am downloading a 700MB movie to watch tonight, I don't want it to take two days to download, that is not practical.

kazaa/emule/edonkey/gnutella:
These are all like Napster.  I don't have too much more to say about it.

torrent:
torrent is brilliant because everything is decentralized.  The whole downloading/uploading mechanism is randomly spread out amongst all the users for the files.  So the great benefit here is the reliability.  It is super reliable because you know how many seeds/peers there are.  So psychologically, the end user knows when to expect the file and how easy it is to get.  The more users, the faster you get it.  hardly any issues with broken connections or anything like that.  this is also a limitation, however.  Popularity dictates how fast you get your files.  This is good for public networks like torrent, but it's not good for the type of private (small group) sharing that I'm talking about.  This is also NOT like private torrents, which also have thousands of users.  We're talking just a handful of users, maybe one or two even.  So we don't want speed and availability limited by popularity.
The other difficulty with torrent is that it's hard to dynamically share, say, this entire folder on your hard drive.  Let's say I have a folder where I keep all my movies ("C:\movies\").  You can't just share that in torrent.  You have to make a tracker for each individual movie, etc.  So that's not convenient for something like this.  So there are some great things about torrent to consider, but also some impractical things.

newsgroups (usenet):
Ah!  Newsgroups!  My absolute favorite.  newsgroups are the best at reliability.  If you see something in a newsgroup, you will ALWAYS get it and it will ALWAYS download at the maximum speed your internet connection can handle.  It doesn't matter how popular it is, or how many users are currently downloading it.  None of that is an issue, and it's beautiful.  As far as speed and reliability goes, none of these protocols will top newsgroups.  So what's the problem?  It doesn't really work for private file sharing.  It's very public, and it's sort of quirky.  Again, like torrent, you can't just share a folder easily.  Uploading stuff to newsgroups is a bit of a process (you have to split all the files up, make the proper subject headings, etc.).  So I'm not sure what to learn from newsgroups other than we'd like our speed and reliability to mimic it.

dc++:
This is another favorite of mine.  This one is pretty ideal for what I'm talking about.  You can share entire folders very easily.  It's very similar to napster except for one thing: the indexing.  unlike Napster, the files are not indexed on a central server.  The indexing takes place at the end-user's computer.  For example, if user 1 want to download a file from user 2, user 1 must first download user 2's INDEX.  Then, you can download the file.  So eventually, user 1's computer will have a bunch of index files from all the users he's gotten files from.  It's a cool model and it works really well.  There is one issue in that it's not a secure or encrypted connection.  Fortunately, the newer cousin ADC offers DC++ with a secure connection.  Unfortunately, ADC is difficult to setup and is not widely used or developed right now.  I had and continue to have very high hopes for ADC.

ftp:
ftp and all it's flavor (sftp, ftps, etc.) is the easiest way to access your own private files.  It sucks though.  It's slow, it's totally unreliable.  Connections always break off, you always feel like something might go wrong while downloading a huge multi-gigabyte file.  There's no indexing.  It's ok for managing your html files, but for large scale file sharing it really sucks.  There's nothing to really learn from ftp, it's old hat.  Actually there is one thing which I'll get to later, multi-user support.  With ftp, using the permissions, you can control what each individual has access to.  You can't do that with these other p2p software, but that's a great feature to have.

VPN:
vpn is bullshit.  Too hard to set up easily.  Everything that currently exists is very "corporate".  Usually, you need an IT person managing it.  it's complicated to understand.  If it works well, it's totally bad ass.  It's also much more than just file sharing, so there are complexities that I don't even understand about it.  To me, using vpn for this sort of thing is like making a mountain out of a mole hill.  However, maybe there are some elements that we can learn from.

irc:
This is an interesting one.  Overall, a little too geeky to be considered easy.  It sort of like a mixture between newsgroups and dc++.  It's organized like newsgroups: there are channels and hubs.  DC++ doesn't have channels, but it has hubs.  It's more of a chatroom than DC++, which also has chat but is more focused on file-sharing.  No really good ways to easily set up folders to share.  Everything is done using commands, so that's inconvenient.
--- End quote ---

Navigation

[0] Message Index

[#] Next page

[*] Previous page

Go to full version