Initiatives around ubiquitous, semantic and personalized computing are a long story. NEPOMUK (SemanticDesktop.org) supported by European Union, Semantic Desktop by Pi Corp, Microsoft’s Semantic Engine and several other projects aimed to deliver a seamless, powerful & personalized user experience on top of semantic storage backend. The problem is that saying that these projects really succeeded isn’t correct. They delivered great technologies, some of which are now shipping in big products like Semantic Search in SQL Server 2012 that was originally the so-called Microsoft Semantic Engine. I can’t dig into details of the Semantic Engine as it was an internal project; still it was presented to the public at Microsoft PDC 2009. But, still, they didn’t really succeed on their own. Why?
As we are developing a system that also can be characterized as ubiquitous, semantic, and personalized, we constantly ask ourselves, what is the secret sauce that was missing from these projects?
Semantic. From perspective of different approaches to store and represent information in the world where Anyone can say Anything about Any topic (famous AAA slogan), use of databases with strictly defined schemas is not applicable, and back in beginning of the XXI century the Semantic Web entered its way as an answer to the question. A combination of graphs of triples representing atoms of information in a subject-predicate-object manner seemed to be the most promising approach to store and represent the information. And yes, in 2011 we see Twitter, and Facebook, and lots of other projects built using these principles, so, to some extent, the approach is correct. And the popularity and attention these social networks are attracting is so loud and clear that it seems that whatever it is, semantic computing isn’t a something that could be a reason for these ubiquitous, semantic personalized computing projects to fail.
Ubiquitous. The original idea was to make computing available from everywhere, to access the big information network from any device, no matter where you are. The devices should become small, almost invisible enabling you to achieve your goals. And, with the rise of mobile communicators – iPhone, etc., – it becomes pretty clear that being connected to the world via these small devices, being in the stream of the information, is something really valued by people. We already live in almost ubiquitous world – you wake up every morning and see a screen of your communicator, you check your emails and Facebook notifications, you send SMS to your loved onces, you see latest news and weather report. You are always connected to the information. Whatever it is, ubiquitous computing is almost already here, and it is a really big trend that we already live within for quite a some time now.
Personalized. This is tricky. A true personalization actually means that a device, and its software should know you really well to be able to successfully adapt to you and your behaviors, know what you like and love, and what you hate, know your history and your current situations, your social connections, almost everything about you. Until recently software wasn’t that personalized, and pretty often you had to put a lot of effort into customization of the software you worked with in order to gain all the benefits of this thing, personalization. Say, rules for incoming email – these are from your friends, these are from your boss, these are from your loved once, these are spam, these you’ll read later, etc. With rise of free email services, we got these rules available for free – Gmail, Hotmail all have them now. Personalization is pretty much about content. If you like music, you probably have a collection of several dozens of tracks you like to listen to again and again. There are classifications of music by genres, artists, albums, years, etc., and you can ask computer to play music by specific artist, or of a specific genres, you even can ask it to play either originals, or mixes of the original soundtracks. You can save these sound tracks to playlists, and computers can later retrieve them back for your pleasure. Services like Last.fm got a lot of information stored within it about classifications that are based on users preferences; if you want to listen to techno music that is useful for programming, you can just turn on Radio channel in Last.fm based on the music track you already feel great for this purpose, and the service will automatically find similar tracks, essentially becoming a new radio station tuned up to your preferences. Similar ratings are done for movies, and books, the list can go on and go on. Whatever it is, personalized computing is a something that is a service people pay for, and is a something that is also embedded into our lives.
So what is the reason some of most visionary projects didn’t succeed when all their components, or pillars already proven their success?
The Secret Sauce
A question of philosophical kind is the one that pretends to shed some light on the problem. It’s a question of difference between a something dedicated for solution of a specific problem vs something focused on solving generic problems. And a question of market niche.
The services mentioned above are trying to solve one particular problem. Similar Music & Videos, or Availability Everywhere, or Search.
But when someone tries to solve too many problems at the same time, and builds a product that is trying to be a really generic one, it almost constantly fails. And when someone tries to solve a problem that is important for a really small audience (and that audience won’t be able to make a really good return of investment by paying somehow for the product), they get a too small market niche, and they also almost always fail.
For the Universe project, this is a really painful question. A strict need to keep the balance between trying to solve a quite a generic problem for a really large audience – information workers, and at the same time to keep us relevant for a really large customer base.
That is, a secret sauce of success. The balance.
And it is a something that is really hard to maintain.