Original URL: http://www.theregister.co.uk/2008/09/22/dziuba_anti_revolution/

OpenSocial, OpenID, and Google Gears: Three technologies for history's dustbin

A poke is not a revolution

By Ted Dziuba

Posted in Developer, 22nd September 2008 01:02 GMT

Fail and You Hey, does anybody remember Google's OpenSocial? Come on, it hasn't even been a year since it was announced. OpenSocial was supposed to unify social network application developers behind one common API. Revolutionary, innovative, all that shit.

Still nothing? Ah, okay. What about OpenID, the best damned federated authentication scheme the world has ever seen, but nobody in the world can figure out how to use? If not, surely you must have heard of Google Gears. This was the Firefox plug-in that was supposed to establish Google as a first-rate operating system vendor, even before Chrome was supposed to do the exact same thing?

I'm still waiting for these technologies to change the world...any day now.

It's Like A Venereal Disease Transmitted by TCP/IP

If there's one thing all engineers love to do, it's create APIs. It's so awesome because you can draw on a white board and feel like you put in a good day's work, despite having solved no real, actual problems. Web 2.0 engineers, in addition to their intrinsic love of APIs, have a real hard-on for anything having to do with a social network. For example, developing a Facebook application lets them call their shitty little PHP program an "application" running on a "platform," like a real, live computer programmer does. Make-believe time is so much fun, even for adults.

This situation gets really dangerous when you start to involve people from San Francisco. Every person who lives in San Francisco has the intention of starting a nonprofit organization of some sort. Therefore, if you collect a bunch of Web 2.0 engineers in San Francisco, the inevitable outcome is the OpenSocial Foundation: a nonprofit organization that only exists to support an API for programming social network applications.

As a result of this upbringing, programming with the OpenSocial API feels a bit like being bukkaked with tolerance and understanding.

Like every other product Google has released since search and ads, OpenSocial has been a dud. It served its purpose: generating talk about Google in the social networking sphere, all the while being thinly veiled as an act of benevolence. Unfortunately, excitement died off quickly as people remembered that they don't really give a shit about social networks. To date, Google's featured OpenSocial applications are:

My, what a hotbed of innovation.

Greetings Dr. Falken

OpenID is a federated authentication scheme for the Web. There's all sorts of good shit built into it: identity providers, Diffie-Hellman key exchanges, SSL, and a little bit of HTTP thrown in to make it internet enough. OpenID is very secure, too, because it's a bit like a Rubik's cube: It's complicated by design, so that stupid users and developers don't even bother with it. This makes maintenance of the spec very easy for the OpenID Foundation, which is yet another nonprofit formed around an API.

The point of OpenID was for you to be able to use one account across many web services. And of course, because building off of existing technologies like LDAP, NIS, Kerberos, SAML, XACML, SASL, or Active Directory isn't a good enough excuse for creating a nonprofit, what better thing to do than create a new authentication scheme?

Yeah, the kid who made LiveJournal can surely do a better job than those dinosaurs.

Anyway, have you ever actually tried to use OpenID? I gave it a stab once. What a catastrophe. I got lost on the part where my username isn't really a username. It's aURL, and I have to go somewhere else beyond the site I am trying to log into in order to type in my password. The damndest part of it was I was trying to log into some also-ran Web 2.0 service that thought they would be able to get more users by only supporting OpenID, because shit, 250 million people have an OpenID, by virtue of having an account at Yahoo or elsewhere. The only problem is, only 14 people out of the 250 million even know what OpenID is, and maybe 6 of the 14 have figured out how to use it.

Can you feel the revolution?

Spacely Sprockets or Cogswell Cogs?

The media's fascination with Google taking on Microsoft didn't start with Chrome. No, Google Gears is widely understood to be the first flaccid attempt at an application platform to come out of Wonka's Chocolate Factory. Gears was supposed to be a panacea for web developers who wanted to support offline access of their apps. Well, the user would have to install a browser plugin and the developers need to write code to support less than one percent of their user base, but if you ignore all that shit, Google Gears was the second coming.

The tech media ate that shit up too, proclaiming that this was a "new platform war" and that Google is "driving straight at Microsoft's profits." Gee, does this sound familiar?

As a developer, there isn't a significant enough drive to work with Gears. For example, it offers a thread pool feature where you can, get this, run Javascript tasks in the background to improve performance. This is the programming equivalent of putting a wing and racing stripes on your stock Honda Civic. Don't you think it's just a little scary that people who spend their programming careers in Javascript and PHP are now allowed to think about concurrent execution? Well, Google did, and crippled the "threading" interface by making all workers shared-nothing. Nice, now you don't have to worry about hard things like resource synchronization.

If you are looking for a web development platform that goes beyond the capabilities of HTML and Javascript, Google Gears is about as useful as a tumor. It turns out that roughly 99 per cent of internet users already have Adobe Flash installed. With that kind of base and an automatic update feature, it's no wonder that nobody gives a shit about Gears. I guess offline support doesn't have a whole lot of pull.

Google Gears on its own was a complete failure. Now, though, it's integrated into Chrome. If Chrome can accomplish what no other browser has ever done and overtake IE, Gears will have a shot.

I'm not holding my breath though. Never underestimate the disparity between developer excitement and user apathy.

Storm the Bastille

A revolution doesn't mean that you pseudo-sexually "poke" someone from Facebook to MySpace. It doesn't mean that you can type in your password once and have access to every authenticated service out there. It certainly doesn't mean that you can make a word processor program work when you're not connected to the internet (we've had that feature for decades).

No, a revolution means that somebody gets beheaded. ®

Ted Dziuba is a co-founder at Milo.com You can read his regular Reg column, Fail and You, every other Monday.