Portal Primer II

Here comes Part II of the Portal Triptychon. Stay alert for things to come.

PORTAL PRIMER

Version 0.67

Part II

Lorenz Lorenz-Meyer

VI. Social Web

So let’s continue our quick stroll through the Garden of Portal Wisdom! Basically, there has been very little progress made in the last years in this area. Most portals still look more or less the same. Probably the most advanced portal in the English-speaking world is Yahoo.com. The people at Yahoo have been very quick in responding to every hype and trend in the Internet world, some of their attempts successful and some not. For example, Yahoo bought Launchcast radio, a learning Internet radio station based on the MIT MediaLab’s influential collaborative filtering and recommendation technology (cf. Amazon’s recommendation system) as early as 2001. Also, their recent acquisition of Flickr.com, a highly praised and successful photo sharing platform, shows that they are very much on alert regarding the social technologies of Web2.0. On the other hand, Yahoo hasn’t yet managed to integrate Flickr into its portal environment, and their own photo and video publishing tools are somewhat less than perfect. So let’s see whether the Social Web and Portal Culture are a good match.

What exactly is the Social Web? Mostly, the terms Web2.0 and Social Web can be used interchangeably. Both concepts focus on a set of technologies based on user communities on the Internet. Community, the net-based interaction of many to many, is no new phenomenon. BBS conversation, chat and instant messaging have been around for many years. But Web2.0 is community with a twist. There are several features which in combination make it much more important than just another Internet hype.

First and foremost, the Social Web provides for a set of technologies that allow extremely easy publishing. Take Wikis and Weblogs for an example. Both are publishing formats – a hierarchical, inter-linked set of web pages the former, a chronologically arranged, diary-kind of Website the latter. But both can also be viewed as defined by very simple, easy-to-use content management systems. In both cases, publishing is not seen as the act of an isolated individual. Wikis are structurally built for collaboration, and weblogs are essentially giving the writer’s audience a voice in form of ongoing commentary.

But even this lowered threshold for publishing and interaction would not have guaranteed the far-reaching consequences of the ‘social web revolution’. Another set of features and services helped to make the existing communication space far more helpful and interesting than it used to be. First, there is RSS, a data description and transmission protocol that allows for an easy-to-use subscription method. Frequently updated sites can thus be checked automatically without the reader having to call them manually through his browser. The RSS feed of the sites is updated regularly and any change subsequently highlighted in the subscriber’s feed reader – a software that is as intuitive and practical as any E-Mail client.

Second, there are Ping and Trackback. These are methods that actively inform other sites about what is going on on your site. Imagine you have added an article to your weblog, commenting upon an article on another blog. A ping to a search engine or directory pro-actively tells these services to update their databases with your new contribution. A trackback message to the weblog you’ve commented upon informs this weblog about your comment and normally is quoted there as if it were an on-site reader’s comment – with an added link to your site. Trackback basically complements the traditionally unidirectional Weblinks with a backlink: I not only set a weblink to your site, but I also notify you of this link and thereby allow you to explicitly register the link that points to your site.

These methods, RSS, Ping and Trackback, are like a glue that keeps the individual contributions in the social web interconnected. Reader-Writers subscribe to each others sites, they inform each other about their activities and relation to each other, and they make sure that their work is to be found in the relevant directories and search engines.

Then, third, there is Tagging. A lot has been said about the dream of a ‘semantic web’, even WWW’s creator Tim Berners-Lee has spent years on the attempt to better organize his path-breaking invention with a set of suitable Metadata. Organize the chaotic web, that is, like a good library, with a standardized set of descriptors and keywords which help to better categorize, catalogue and search its content. But exactly this professional zeal has made the Semantic Web a hopeless task. Even with professional archives and libraries it’s extremely difficult to agree upon a common set of standards. With millions of users and micro-publishers on the Web it’s simply impossible.

So when Joshua Schachter, operator of the social bookmarking site del.icio.us, allowed his users to add to their entries any keywords of their own choice, disregarding whatever professional standards, this was like a democratic revolution against the tyranny of the professionals. Other popular services like Flickr followed suit. Many experts welcomed the development and ‘tagged’ the idea of unregulated tagging with the term ‘folksonomy’ (from ‘folk’ and ‘taxonomy’). They acknowledged the fact that for the idea of a Semantic Web you have to sacrifice perfection if you want to gain any practical use. Like Flickr, del.icio.us was bought by the ever-alert Yahoo – but has not yet been integrated into its portal structure.

A fourth contribution to the Social Web are the already-mentioned Recommendation Engines based upon collaborative filtering. This is a method nearly everybody knows from Amazon: “Customers who bought this book also bought the following…” The method is based upon profiling and similarity relations. One user is showing his preferences (explicitly or implicitly, through monitoring his behaviour) and is then mapped upon a set of other customers similar to his profile. Doing this you can either map him to these other customers, thereby creating a cluster of like-minded people, or you can map him on a set of products originating from the preferences in the cluster of his ‘buddies’.

An obvious application of this method are Learning Internet Radios: I listen to a random choice of music, continually rating the tracks and asking for tracks of my own choice. The service analyzes my preferences and relates me to other users with a similar taste. Then music of their choice that I might not know is offered to me as a recommendation. The result is breathtaking. People report that they never felt better informed and understood, even by the most expert and attentive music vendors.

This is the sophisticated way Internet services like last.fm and Yahoo Radio work. But more primitive versions of recommendation engines are also to be found in social bookmarking and other areas. The future applications of the method are numerous. In the long run it opens up the possibility of ‘virtual media brands’: Imagine an Internet user asking an automated agent to collect Chinese language media content as similar as possible to the profile of the New Yorker!

As RSS, ping and trackback help to connect the fragmented Internet content, tagging and collaborative filtering help organize its abundance. All of these features and methods have one thing in common: they are not driven by technology or business, but by the individual intentions and actions of thousands or even millions of Internet users, wherever they are.

Incorporating these insights into Portal Wisdom seems to be inevitable. But it turns out to be surprisingly difficult, as can be seen with the difficulties Yahoo has with integrating its recent acquisitions of del.icio.us and flickr into its portal structure. Part of the reason is that the Social Web is distinctly anti-portal in at least two respects: it is essentially decentralized, and prima facie it seems to be non-commercial. This latter observation actually is an illusion, but it turns out that the pre-dominant business model applicable to the social web is incompatible with the model that traditional portal websites make use of – a model that is essentially based on control of a few relevant marketing channels.

We will later get back to this topic when we compare the Yahoo (i.e. traditional portal) paradigm with the Google paradigm. Also I will, as promised, say a few words about the integration of multimedia content. But first it’s time for another break.