On Tuesday this week I went to the London edition of F8, the “on tour” edition of Facebook’s flagship developer conference. I have to say (and I did say it), it was the most useful day at a conference I’ve ever had.

Despite spending a lot of time reading through the docs for the new open graph, there’s nothing quite like hearing it from the horse’s mouth. Not to mention seeing some of Facebook’s finest engineers live code on stage to show you how it all works.

The big topic of the day was how to use the BETA open graph, a new way to get your apps distributing content to Facebook’s ticker, newsfeed and timeline. It also gives you a hitherto unavailable level of control over how that content appears, is structured and linked. In absence of any personal imagination, I’ll copy their example:

The Open Graph allows apps to model user activities based on actions and objects. A running app may define the ability to “run” (action) a “route” (object). A reading app may define the ability to “read” (action) a “book” (object). A recipe app may define the ability to “cook” (action) to a “recipe” (object).

This is effectively a lesson in grammar. You get to define an action (verb) and an object (noun). But the really useful part is that you can then create aggregations of those verbs or nouns that appear on a user’s timeline (which, incidentally, I think is a really great feature).

So from a developer or website point of view, you need to think about what sentences make sense to people and whether they will be interesting enough to be clicked by that person’s Facebook friends. Because as with most Facebook integrations, you need to think both about how things work for your user, and your user’s friends. Ultimately, you want something posted to a newsfeed or ticker that is compelling enough to be clicked on and bring you some extra visits. It’s a bugbear of mine that people often focus on one of those audiences, but not always both.

I’ve spent a lot of this week trying to map out all the relevant objects and actions on JG and how they fit together in an almost database like structure, but as this is so abstract, it’s been easier to focus on the sentences we want people to share and what aggregations will be interesting to our audience and then work backwards.

So if you start with a description of what someone does on your site (like, I don’t know, sponsor a friend ;-)) then you can work back and define the relevant objects and actions on Facebook. Once you get your head around this, it’s actually relatively straightforward to set up your mapping on Facebook and then add the relevant meta tags to refer to your custom objects, as per another example below:

<meta property=”og:type” content=”mydemoapp:recipe” />

<meta property=”og:title” content=”Stuffed Cookies” />

Once that’s all set up, you need to ask for permissions to publish actions at some point in a flow on Facebook or on your site. This is a one-time ask, and one reason why this is so good for content publishers is because it removes the barrier of asking to share (frictionless sharing is what Facebook call it). In the case of spotify, once you authorise the app, each song is shared without the user having to do anything. Whether that’s good for a user or not is another story (and I’ve been caught out a couple of times by spotify sharing dubious song choices…).

Aggregations

Finally, you can create aggregations of any combination of objects and actions to give your user something interesting to show on their timeline. For example, here’s the slightly random collection of music I didn’t realise I’d listened to on spotify in October until I looked on my timeline…

This is where you add some really interesting value, and make someone’s interaction with your app be a part of their social identity. These was a theme mentioned a couple of times on the day. It’s like those boxes you used to be able to put on your Facebook profile to show you liked something, except this is an opportunity to show the user something new, something different, something interesting that they didn’t know themselves – the launch partner apps show top playlists on spotify, or most-read authors on the Guardian, but the potential for this is really quite exciting.

Advertising – the scary/awesome bit

One last thing to add is that Facebook said you would be able to advertise to people who had taken custom actions related to custom objects in your app. So if you have listened to an artist on spotify, that information can be used to target you with an app. This is awesome in that you can advertise based on custom verbs and nouns you define, but scary in that the level of ad targeting Facebook can use has just upped a notch (and it was already more targeted than any other form of advertising already). Fast forward a bit, say an app shared that someone was “buying” something, an advertiser could then target that buyer in near real-time with another offer, and all that person’s friends could see they were in the process of buying something and suggest something else – a scary/cool type of social commerce…

And if you look at it another way, they have managed to build a way of allowing you to add rich customised content to their network, which adds value to their network, and then allow you to pay to advertise to people on their network, based on the content you’ve added! But that is the trade-off, nothing is ever for free. But as far as I’m concerned, the value you will get from integrating so deeply into Facebook is worth it, so why shouldn’t they get some value back too.

Ultimately, though, given that the open graph is in BETA, won’t be live until Timeline is released and the ticker is still new and bedding in, it’s difficult to say how this will pan out and how successful it will be. But I suspect we’re about to embark on a new wave of innovations on Facebook’s platform – one that will create tons of value for websites, app developers and publishers, not to mention Facebook themselves.

I don’t hear people talk about the semantic web much these days, but Facebook are about to release a platform where anyone can create a machine readable summary of not only their site, but what people are actually doing on their site in real time in a way that can be used to target those people with relevant ads or content. As TechCrunch memorably put it, “Share Buttons? Ha. Facebook Just Schooled The Internet. Again.”

More reactions from F8

If you like your conference content curated (who doesn’t!), then I point you in the direction of this F8 storify, a collection of exquisite tweets, and a report from .net magazine.