[tech, lj] Distributing LJ

[siderea]

Ever since the Strikethrough of '07 -- actually, ever since I realized that LJ was something of an attractive nuisance of basket in which to store eggs, way back when -- I've been thinking about how one would go about turning LJ, the software, from a client/server model to a peer-to-peer model. That is, how to make LJ distributed

Comments

[tangerinpenguin]

Wired published another riff on this theme recently. They're coming at it more from the Facebook/MySpace side of the feature set, but also quickly run up against distribution of the "friend" model as one of the thornier open questions.

[siderea]

Thanks for the link; will review.

Distributing the friends system doesn't really seem like the hard part, to me. I mean, I don't know anything about implementing OpenID, but I've used it and it works fine. On LJ, friendings are bidirectional, and unilateral. So I don't see what the hangup is.

[siderea]

OK, read the article. Wow, their understanding of the problem is wrongheaded I don't quite know where to start. Maybe we could just have danah boyd beat them over the head with her article on "autistic social software".

[siderea]

Actually there's something profound here, maybe deserving of its own post, about insiders vs. outsiders trying to solve problems. Examples abound in the social services.

[mellyjc]

I'm not entirely sure I understand the purpose. Of course, because this is where my friends/journal are, I haven't made friends on DJ...

And I'm certainly not in for paying for hosting. I love getting comments and all, but I'm a cheapskate and wouldn't pay for any sort of blog utility. I'd revert to my Microsoft Word journaling long before that.

I love the idea of it, but I can see it much more usefully on facebook. With all the addable/creatable applications, it's potentially a great hub for people like me who have no web programming knowledge. I'd love to have a feed reading in from yelp and sparkpeople.

[siderea]

I'm not entirely sure I understand the purpose.

I'm quite sure you don't. This isn't the place to discuss current events, especially in light of just how many other journals are and have been overflowing with that discussion. I recommend you use this as an entry to the issues, but carry on the conversation somewhere else.

I'd revert to my Microsoft Word journaling long before that.

It's good that you have that figured out, because you may yet get a chance to use that exit plan.

[conana]

Yeah, I've been meaning to write a post suggesting this since the day I joined LJ. Part of larger nostalgia (OK, I'm not that old) for the days when the internet was made of protocols instead of centralized web services. I also have this suspicion that OpenID recreates Kerberos. But anyway. I lack the relevant skills and the motivation. Another case of being technical enough to know how I want things to be, but not about to make it happen. Does anyone have a clearinghouse for ideas like that? There are a few I'd put cash towards.

[siderea]

I think that sometimes, too. But no, I know of no such clearinghouse.

[eichin]

re openid and kerberos: not really. OpenID only *carries* authentication from one site to another; Kerberos actually performs it. (Given that IE and Firefox (and maybe safari?) can do kerberized (GSSAPI) HTTP against mod_auth_krb, now, you could use kerberos to authenticate to your OpenID "provider", if you had your own...)

[m_danson]

Interesting. WAY out of my skill set but interesting. If it were easy to impliment I'd probably give it a try.

[londo]

Without addressing any of the other issues, it occurs to me that the Internet has already partially solved the unique ID problem in that we have DNS. That said, those names are going to be longer.

Finding people who are familiar with the technologies involved is easy, too. Whether or not they can be convinced to do anything about it...

[siderea]

Without addressing any of the other issues, it occurs to me that the Internet has already partially solved the unique ID problem in that we have DNS. That said, those names are going to be longer.

Absolutely. However, I was thinking about a slightly more complicated application which could be considered "gravy", except that it is necessary to the importation of LJs. There needs to be an interface for the LJdist owner to say "lj:user{londo} == myljdist:user{londo@centauri.gov}". Once you have that, it seems tempting to say, "Let's make that a 1-to-many!" so that you could also say "lj:user{londo} == myljdist:user{londo@ambassadors.b5.gov}". Then you could authenticate against either server -- which has the awesome property of giving you backup authentication. Because one of the interesting properties of this architecture I've been thinking of is that if your site goes down, you can't access anybody else's site, either, as a logged-in user, because your site isn't there to authenticate against