Dave Winer: "Twitter is a ghost town" | Hacker News Reader

Dave Winer: "Twitter is a ghost town"

scripting.com

7 pointstimr15 comments

So, the interesting thing, to me, is that a few weeks ago everyone was discussing the axing of Blaine Cook from the Twitter team for his "failure to scale". Turns out, nobody else in the company can scale Twitter or keep it online reliably, either. Huh. Must be a hard problem. It's probably a good time for folks to apologize to Blaine for piling all the blame on one guy for a problem that, in reality, took the whole development team to create and will take the whole development team to solve. (And it's time for everyone to stop pretending scaling can be solved by one magic bullet or another...if you don't build, or rebuild, to scale, you won't scale.)

Deleted comment

OK. I trust you've built the scalable replacement for Twitter already and were just waiting for the opportune moment to launch? Now's the time!

Would you not agree that Twitter is a subset of the functionalities found in existing technology that does scale? email, IM networks, blogs...

Obviously you're entitled to your opinion about sending messages being a very hard problem, but I think it's been solved a few times now.

All of those systems apart from IM follow the same route to success - a large number of independent systems with a common interchange format that makes it possible to create the illusion of integration.

Every mail server, every news server, every blog is a little island that runs on it's own. I would bet that behind the scenes even the big IM networks are actually a mountain of small, effectively independent, service providing systems that understand how to talk to each other - that's certainly how the XMPP/gTalk network functions.

This is only possible because all of those designs are fundamentally highly parallel - if your communication method has data dependencies (real or the result of bad design) that cannot be managed in that way then you get something that won't scale.

your opinion about sending messages being a very hard problem

I said scaling Twitter must be a hard problem, and that it will involve correcting their implementation rather than any one magic bullet. I said nothing about IM, email, blogs, etc.

You've decided that I'm talking about something that I am not.

Here's an off-the-cuff prediction: Twitter itself, as well as the Twitter community, will be just fine. In a year or two nobody will remember these problems.

Twitter may have scaling problems, but it's not as if their potential competitors are magically immune to those. Meanwhile, Twitter holds a huge marketing advantage -- you can't say "Site X is just like Twitter" without saying "Twitter" -- and a big Metcalfe's Law advantage: Everyone who is remotely interested in Twitter-like sites already has a Twitter account, and those accounts won't magically disappear. Meanwhile, there will be more than one Twitter competitor, all desperately trying to distinguish themselves from each other as well as from the original.

Sooner or later Company X will figure out how to scale Twitter. At which point either Company X buys Twitter for the name and the userbase, Twitter buys Company X for the technology, or Google buys everybody. Or, Twitter eventually reverse-engineers the superior architecture, at which point they will beat Company X over the head with superior name recognition and larger userbase. This is especially true if Company X is actually the open-source community -- if some random Python programmer manages to create a superior, open-source Twitter architecture, Twitter will just install it on a server farm within a month and be right back in the game. The open-source movement has a lot of brilliant techies, but I doubt that its marketing team can outrace Twitter's from a standing start.

The Twitter concept has enough mindshare that I doubt it can just evaporate. They're not being sued out of existence like Napster. I'm not convinced that a scalable Twitter is physically impossible. So I would suggest that rumors of its death are greatly exaggerated. It's just... resting.

Agreed. I signed up to Pwnce right after I signed up to Twitter, but quickly decided I couldn't be bothered with Pwnce since Twitter was clearly fine and had more users.

Twitter is very active from where I'm standing.

Agreed.

The strange thing is that yours is the only comment here that responds directly to Winer's statement about a ghost town.

Is anyone else wondering what the hell Winer is talking about?

I started using twitter a lot more right before blaine left. Now I barely use it because the IM integration is gone. Hopefully someone will move forward to implement the distributed twitter on top of XMPP and Pubsub. There is already a tentative little XEP: http://www.xmpp.org/extensions/inbox/microblogging.html

This is a wild guess, but from what I've seen Tumblr is picking up a lot of momentum. It's started to attract a pretty smart community and it's building steam. I don't know if it's a REPLACEMENT for Twitter, but it serves the same purpose but without the downtime.

I've jumped on Plurk, and I like it a lot.

I just heard about Plurk. Funny. Looks like it's a new thing. I don't think it's going to last. Its problem is that it adds on complex, unneeded features to a simple idea.

That's what made Twitter so huge: its astounding simplicity. It lost points for being TOO simple and for having downtime, but that's what made it huge. Tumblr has that feel to it: that feeling that it's as simple as it can get, with a few exceptions. It's an exhilarating feeling.

Plurk doesn't have that. It feels to me like Twitter with clutter.

The magic simply has vanished. Twitter just isn't as....fun anymore. It's kind of like the eerie silence you have when you've been informed someone you know has passed away.

Reply on news.ycombinator.com