Dreamwidth

[juliet]

I've been doing a little (very little...) coding over the last couple of weeks for Dreamwidth, which went into closed beta at the weekend. (Which was/is very exciting!).

It's a code fork of Livejournal, but they're taking the opportunity afforded by starting up anew to rip out large chunks of code by the roots & rewrite things *better*. The biggest example (for me, anyway) of this being the replacement of the friends system by a system which explicitly separates "people I want to be able to read my stuff" from "people I want to read". For me, that's pretty much the Killer App right there. The increase on the comment limits from 4,300 char to 16,000 (there's a similar post size limit increase, but I've never hit the post limit so that bothers me, personally, less :) ) is also shiny. Plus: no ads. And one of the owners (synecdochic) has discussed at length why ads on social content sites Do Not Work, financially speaking, so I believe that they actually mean this.

The meta-things that I'm impressed with:

- a commitment to openness (of business practices and site-running) and consultation that I believe in. Possibly in part because I've been involved with and/or witnessed discussions about proposals or ideas that have resulted in decisions being changed :)

- a diversity statement that doesn't read like corporate bullshit.

- the business plan is along the lines of "small, stable, functioning, ongoing business" rather than "let's flog it in due course & bathe in web2.0 cash". An interesting post on the subject of business aims & so forth.

So, yeah. I shan't be moving off LJ altogether unless/until various things change[0], but I'm certainly going to be doing a certain amount of moving across. I gather there is a crossposting tool, for starters. I'm very fond of LJ, but only in the sense of being fond of the things it enables me to do. DW feels like a project I can be fond of in a more specific sense, and I like that.


[0] The real Killer App would of course be full interoperability - i.e. being able to read one's LJ friends-list, even locked posts, on DW via OpenID. This is a Difficult Technical Problem for which there is at present no decent fix, but it is being contemplated. Full import of your own LJ will be in place before open beta; OpenID already allows cross-site reading & commenting on non-FL posts.

Comments

[ciphergoth.livejournal.com]

What you call "full interoperability" here I call "federation", and it is as you say the feature that makes all the difference. Where is it being discussed?

[juliet]

Some discussion here:
http://wiki.dwscoalition.org/notes/Cross-site_authenticated_RSS

(And occasionally some on IRC (#dw on irc.dreamwidth.org) at intervals (& doubtless some privately between the main coders/owners).)

[steer.livejournal.com]

Whatever you call it, it's certainly the right concept. At the moment there's a quite ridiculous situation with people maintaining multiple online IDs for social networking. when I was half way down reading julietk's entry I was thinking "wonder if it will do OpenID". Of course I'm not familiar with the technical ins and outs so I don't know WHY it will be a difficult problem.

Why is "federation" your preferred word for the idea?

[juliet]

that link I posted above has a fairly succinct summary of why it's difficult: the main problem is one of sheer volume. (doing it the obvious way, i.e. prodding the remote site every time the local user hits refresh, will lead very rapidly to being banned by the remote site due to overload).

(In case it wasn't clear: yes, it will do openID - you can already log into the closed beta using openID to make comments. But that's only half the problem - it's the FL posts thing that is the real issue.)

[steer.livejournal.com]

Of course -- polling would be a huge issue. I can see there's a big problem in getting the different identities to interoperate properly -- especially when it's not (or at least not clearly) in the interest of (say) Facebook or LJ to allow such access.

[babysimon]

I think getting federation popular once the technical problems are solved might not be that hard. SUP is unpopular, LJ is unpopular, a lot of the fan people maintain lots of IDs on LJ clone sites, with attendant aggravation. Getting buy-in amongst the LJ diaspora should be easy - sites that permit it will become more popular.

Of course Facebook will never do anything to co-operate, but the challenges involved in just working out a common data model between FB and LJ would be hard enough...

[steer.livejournal.com]

Hmm... Getting buy-in amongst the LJ diaspora should be easy - sites that permit it will become more popular.

I think this requires a certain critical mass. At the moment few people use Open ID and for a profit focused entity like LJ there's no pressure to allow it -- indeed there's probably the counter pressure "if we allow this people will leave LJ faster" because it enables people to host their journals and read other journals from different sites. Beyond a certain point the pressure would be the other way "if we do not allow this people will leave LJ faster"... that's my thought anyway.

Facebook's incentives to co-operate or not are surely the same as LJs it's just that they are more in the ascendant right now. Such things change though.

[babysimon]

I guess I see LJ and FB as different because I perceive that the LJ user base as rather more pissed-off-and-ready-to-leave.

I could very easily be wrong on that.

[steer.livejournal.com]

Ah... perhaps that is a factor of your social circle? I don't think I know anyone leaving specifically because they are pissed off with LJ but there is a gradual drift to Facebook. It seems to be a "pull" to Facebook not a "push" from LJ in my friends group though.

In my experience internet technologies have a very short half life! There's a transition from "have you seen this cool new thing, usenet news" to "how come nobody posts any more". I'd guess it's an average of around ten years for a very successful social networking tool to move from "cool new thing" to "legacy system". I'm thinking here of BBS (various local/campus based and international), usenet news, LJ.

[juliet]

See, I tend to think that it's the other way around - LJ (at least *this* LJ - the clone sites it may work differently) has an incentive *not* to solve the cross-site-auth-RSS problem, because it gives them some level of lockin.

I very much doubt that I'm the only person who would up sticks & depart from LJ altogether (once there's an alternative I can live with), *if* that wouldn't mean that I wouldn't be able to read other people's locked posts any more. So if LJ cooperates with other sites getting at FL content, they lose users.

Having said which: I'm not actually sure how LJ's business model works any more (free users & ads vs paid users). I gather that it works *totally* differently in Russia from, um, everywhere else, and that they're increasingly more interested in the Russian side of things. So that might increase the likelihood of co-operation.

*Providing* an OpenID login from your own site, and allowing other people to use their OpenID from elsewhere to do stuff on your site, that's certainly a popularity issue. But basically everyone does that bit already.

[babysimon]

Oh, sure. I guess what I'm trying to say is that I would have a lot more motivation to move from LJ to federating-LJ-successor than just to LJ-successor - since I don't have to wait for all my friends to come to the same place as me, it's enough that they leave.

I suspect that if DJ/IJ/GJ etc. had all been federating with each other for the last few years without LJ, LJ would already be dead outside .ru.

But this definitely comes under the heading of Great Imponderables...

[juliet]

Oh, right, am with you now. Yeah, that's a more plausible thing to happen, and might well be an incentive for other clone sites to cooperate with an openID extension as ciphergoth suggests.

[ciphergoth.livejournal.com]

I think the Right Thing is to propose and implement a new protocol for efficient blog federation a la the "Make it DW-only and customizable" proposal on that page.

OpenID is pretty specific to users sitting at browsers - Step One would be designing an OpenID-like protocol for sites to authenticate to each other, which wouldn't be that hard.

Then you create a user list on the remote site, authenticating that you represent all the users in that list, and then your query is "what's new since this date for that list?". To make it more efficient, the query hangs if nothing is new, so you don't have to poll, you can get updates as they happen. The information you're given must tell you which of the users on the list should see each item.

[mr-tom.livejournal.com]

Sounds vg and all that. Do you have the ability to add people to the beta?

[juliet]

Closed beta is very very small - I don't have an account myself at this point (I think there's maybe 20 or 30 accounts currently? Because there are, as always at early beta, lots of bugs emerging :) ).

Open beta (which is currently expected for mid-April) - see here: http://wiki.dwscoalition.org/notes/Dreamwidth.org:_Launch_timeline
If you want a free account, you'll need an invite code - am happy to take requests for when I do have an account & codes to go with it :) Paid accounts will be available from then.

[mr-tom.livejournal.com]

Awesome. Thank you!

[(Anonymous)]

Most likely you have tons of friends but if you find yourself with an extra invite down the road, and don't mind giving it to a stranger, could you possibly send me one? Thank you in advance! :)

sculptor.irl @ xoxy.net

[ludy]

i like the diverstiy statment

[purpletigron.livejournal.com]

Does sound interesting.

[d-floorlandmine.livejournal.com]

That sounds pretty cool/interesting!

[babysimon]

I am in general quite excited about Dreamwidth, but:

It's a code fork of Livejournal, but they're taking the opportunity afforded by starting up anew to rip out large chunks of code by the roots & rewrite things *better*.

this really scares me! I'm sure we can both think of plenty of OSS projects where "let's have a big rewrite!" has been the precursor to never delivering and gradually fading from sight...

[synecdochic]

Well, as one of the DW owners, I can pretty comfortably say that both xb95 and I are totally aware that You Never Rewrite From Scratch, Ever. *g*

For the most part, we're doing all of our rewrites in a small, contained fashion -- mostly with iterative changes -- if they're pulling out old crap functionality and replacing it with new. (The exception to this: we just nuked to-do support in the codebase entirely. Actually, julietk nuked to-do support, in fact. And it was just so old, and so broken, that it was easier to pull it entirely and replace it with Something Else down the road.)

For instance, take memories -- we want to completely redo them, but until we do, we'll keep the Old Way around. Etc. We have a roadmap that goes out about 18 months, and our bug tracker has a clear list of everything we're going to change and how (and most of them even already have a spec!) You can also see a list of what we've added/improved -- and what bits of LJ we won't have.

Our reason for forking absolutely isn't the "this code sucks! let's have a big rewrite!" that you get so much in the OSS world -- our reason for forking is that the people who've had the commit bit on LiveJournal-the-code-project for years and years never caught on to what the users of the site were actually looking for, and so many of their decisions steered the codebase down (in our opinion) the entirely wrong direction. So we're forking so we can Do It Right. And since both of us worked for LJ for a really long time (me from 2003-2007, Mark from 2003-2006, with both of us having put considerable volunteer time in beforehand), we actually remember all the forks in the road where LJ turned right and we'd like to go back and turn left instead ;)

[babysimon]

Wow, thanks for a really interesting comment. It seems like your goals are reasonably sane ;)

Out of curiosity, is the roadmap public? I searched but couldn't find anything.

[synecdochic]

The roadmap currently exists as a bunch of scribbles in pencil in the notebook next to my computer, so no. ;) It's on the "list of stuff we will dump on the site before launch", though.

You can check our open-beta-blocking items, though, which is the roadmap for the next two months. There's also the planned enhancements and planned bugfixes. (The Priority flag on those is roughly akin to how much we care about them at this particular moment, at least for the enhancements. Anything that's keyworded why-bugfix is pretty much functionally P1.)

[juliet]

And nudge. You forgot to mention that you killed nudge ;) Although, obviously not to *replace* it with anything, just to be able to dance around on the code-corpse.

[synecdochic]

Sometimes I consider rolling back to the changeset before just so I can kill it again.

[juliet]

synecdochic has obviously said all of the useful stuff. But also: just from the small amount of work I've been doing, this does not have the feeling of that sort of OSS project. There are plans, and people doing lots of active work, and a very sensible attitude to what needs to be done/fixed *now*, and what has to go on the "put this in down the road" list.