October 4, 2007 5:02 PM   Subscribe

Decentralized social network: " your social network in one place and have other NoseRub clients connect to it instead of you having to build multiple networks on multiple social networks. The best part is NoseRub is released under the open source MIT license." via eHub. Previously related.
posted by yoga (23 comments total) 4 users marked this as a favorite
I was thinking about making something like that, but "NoseRub"? Really?

Actually Google is supposedly going to announce it's own "social graph" system in November, which promises to be very open. We'll see if it allows people to run their own server. It should really be possible.

When you think about it, social networking systems are nothing beyond Web 1.0 home pages. The systems just automate building your page, and finding and linking too your friends and otherwise updating it.

Really, it doesn't make any sense that these pages should be locked into proprietary systems, but then it doesn't make much sense that IM would be locked into proprietary systems either.
posted by delmoi at 5:19 PM on October 4, 2007

heh. beats ButtSniff, I s'pose.
posted by yoga at 5:24 PM on October 4, 2007 [3 favorites]

It's a fascinating technical challenge. I've been thinking of ways to do this since friendster came out and I've been stumped, but then I'm not a programmer.
posted by empath at 5:27 PM on October 4, 2007

I think it's a spoof on backrub.
posted by lupus_yonderboy at 5:50 PM on October 4, 2007

I think it's a reference to junkies, not unlike the point being made in Crackbook.
posted by UbuRoivas at 6:13 PM on October 4, 2007

It seems pretty obvious to me that NoseRub is referring to the way Eskimos 'kiss'.
posted by SenshiNeko at 6:38 PM on October 4, 2007

posted by Mid at 7:13 PM on October 4, 2007

nooooooooooooooooooooooooooooooooooooooooooooo. make it stop.
posted by unknowncommand at 7:15 PM on October 4, 2007

Wasn't FOAF supposed to solve this?
posted by adamrice at 7:34 PM on October 4, 2007

Whoah, who edited that site. . . or didn't as the case may be
posted by BrodieShadeTree at 7:49 PM on October 4, 2007

Really, it doesn't make any sense that these pages should be locked into proprietary systems, but then it doesn't make much sense that IM would be locked into proprietary systems either.

It doesn't technically make sense, but I bet whoever is counting Facebook's ad revenue is laughing all the way to the bank.
posted by arto at 8:02 PM on October 4, 2007

I just deleted my myspace account, I never use orkut or friendster, and I only use facebook because the interface is sorta clean and my academic friends are all on there. Second life is too slow even on my six month old macbook. Twitter is surprisingly fun, and irc is ok.
posted by craniac at 8:24 PM on October 4, 2007

craniac: GYOBFW
posted by UbuRoivas at 9:28 PM on October 4, 2007

Err, hi - I liked you better the first time when you were called Spokeo or, if you prefer to host things on your own server rather than aggregate, People Aggregator. Ah yes, so many computer problems to solve, and so many startups already doing the same thing, seemingly with no differentiating factor and a silly name to boot.
posted by rmm at 9:49 PM on October 4, 2007

Give me a break. I'm sure the major social networking sites would just love to let you automatically trawl their sites for your assigned user ID (note -- conspicuously absent from, e.g., FaceBook's ToS), let alone the IDs of your 'friends' (specific ToS violation, and assuming they're not munged or corrupted -- and I don't see a rollback function, so it seems to me that one bad update in this 'protocol' = toast).

On first glance, the error checking in their inheritance algos is wrong, but comments like this make me not care enough to really delve into this so-called 'protocol' and double-check:
# not quite sure, if this is still neccessary. was copied from
# a cake 1.1 project
# fixing bug with expects and belongsTo associations
# probably caused by joining belongsTo associations cake doesn't reset
# the bindings of this associations
Mmm, crappy pre-rolled PHP code. I'm sure the oppressed, unwashed social-networking masses will be firing up their servers any moment now. Yep, that's a killer app right there.
posted by spiderwire at 10:22 PM on October 4, 2007

I just don't want any new things when it comes to social networking. Please.
posted by strangeleftydoublethink at 10:58 PM on October 4, 2007


FOAF is incomplete for social networking for 2 reasons:
1) FOAF in its current implementation isn't a verb (actions, i.e. adding friends), most people use it as a document (noun). Future social networking protocols, such as Open Friend Format, is about defining relationships on top of FOAF AND adding/removing them. Note: Open Friend Format was created by Brad Fitzpatrick (creator of livejournal), who is now working at Google.

You could use some WebDAV blah blah blah for actions, but I haven't seen any implementation OR a prototype of that, so something like OFF is necessary for decentralized social nets.

2) FOAF at its present time doesn't have a standard for unique key-id references. I think they're adding OpenID as a tag, and I hope that will take off as the de-facto identifier. I've been using the foaf:weblog tag to play with the idea of decentralized social nets, but again, not a standard. In layman's terms, this means that there is no unique ID for people (e.g. you have a global namespace for your phone number or your social security number, no one else has one).

Of course the beautiful thing about using URLs as your global identifier (a la OpenID), is that you can create even more verbs (actions) behind them -- not just social networks. You can't run XML calendar lookups automatically/easily via email (your current global identifier on the internet), but it's dead easy to do so with URLs.

... I don't see how NoseRub is doing IDs and adding/removing people (or extending things, e.g. adding a file sharing tool or voice call behind my identity), does anyone see any specs around?
posted by amuseDetachment at 11:01 PM on October 4, 2007

What about using a belief-propagation algo for unique id verification? (Gallager codes seem to map well to this flavor of data set.) Seems like in practical terms that'd be more practical to prevent spoofing anyway. Even if you can keep your ID and your site secured, isn't OpenID potentially subject to DNS spoofing too?

I dunno, it's late. It's a fascinating problem domain -- but the linked site ain't the answer. Nifty Web 2.0ish design, though. Maybe they should focus on glassy buttons instead. We could use more of those.
posted by spiderwire at 11:17 PM on October 4, 2007

sipderwire: that sounds totally doable, especially if you can get enough 2-way assertions, and I believe OFF uses some of those ideas. The reason I'm a little bit squeamish about that (including certain aspects of OFF) is you're going to be building trust on top of social networks, and belief-propagation (i.e. really, really good guessing :-) scares me when it comes to something as socially complex as privacy level with friends.

Even if you can keep your ID and your site secured, isn't OpenID potentially subject to DNS spoofing too?
Sadly, most people use HTTP right now for OpenID, and it's a REALLY BIG HOLE (like drive a semi-truck through big). But it's difficult if you use HTTPS -- if they spoofed your DNS, they won't have the site certificate (without some major gymnastics). It's an issue of contention inside the OpenID community with no easy answers, as people really want to use HTTP OpenIDs, but you may see fall out of favor as there's no easy way to secure HTTP against DNS spoofing on the entire web.
posted by amuseDetachment at 11:48 PM on October 4, 2007

With OpenID surely even if you use HTTPS the server has to accept whatever cert it gets as being valid when you sign up? Obviously if the server has spoken to your OpenID server over HTTPS previously it can spot the cert change, but for the first time you connect there's not much HTTPS is going to be able to do right?
posted by public at 2:30 AM on October 5, 2007

I thought orkut was going to morph into one of these aggregators...
posted by chuckdarwin at 3:31 AM on October 5, 2007

You can make an extremely lightweight social network with XFN.
posted by PenDevil at 4:39 AM on October 5, 2007

XFN has the same problem FOAF does for defining relationships: no need for confirmation. I could put up a web page saying I'm best friends with Barack Obama and secretly married to Natalie Portman, and there's no solid mechanic for either one of those people disagreeing with me.

People aren't using social networks just for the lists; they're using them for social validation. To make FOAF and XFN work for that, users need some sort of user clients that map individuals' reported relationships against each other and "decide" which relationships are true ("true enough for the Internet" might be a better way to put it). Last time I looked, that was an idea that the FOAF and XFN developers were actively resisting.

The issue is not really about listing relationships, it's about managing them -- the websites are really "software" for handling the "data." In this case, the data is easy (everybody can make their own), but the software is hard. If the FOAF and XFN developers still think it's all about the data format, they've got a solution for the wrong problem -- people need the software.

I'm not going to bother with NoseRub, myself, but at least they get that's it about providing the user with an interface. That's a step FOAF and XFN haven't got to yet.
posted by faster than a speeding bulette at 9:47 AM on October 5, 2007

« Older So, it wasn't the Happiest Town on Earth?   |   Covering Photography: Photographers and Book... Newer »

This thread has been archived and is closed to new comments