I don't see anyway to technically sidestep the privacy issues to tell you the truth.. other than to have all the providers send their FULL databases for us to sift and act on, which is impossible.

Whatever we do will be done through their systems and in such way we'll always be sending our actions and intentions to them.

^ Right, which is what I was alluding to when talking about how domains and gestures of trust have been the stumbling block here... and may continue to be.

Whoever manages your metadata wields great power...

^- Right.
So the first company that offers that service and can technically prove that our 'data' won't be misused will be the one to use. Google's "Do no evil" policy is not good enough anymore.

Social Networks are sort of doing this now (think about places like Last.fm or any of the Twitter trending services), although viewing the "attached" user is fairly integral to those systems -- but it wouldn't have to be. People, places, interest, keywords, etc., all get aggregated into a "menu" of sorts, where the user can pick and choose.

Tim

It's good to see someone on the same wavelength.

I'll be perfectly honest here, mustering as much objectivity as I can regarding the context: I really think Nokia *can* be that entity. As an employee I found the corporate ethics to be of a higher standard than any other employer for whom I had worked (although TI is a very close second, or was a decade ago).

IF Nokia can quit stumbling in Internet services, and get the Ovi act together, the company possesses the potential to even outdo Google in certain areas. The breadth of coverage offered by its device share is tantalizing--translate that into portal share and WOW. But it keeps coming down to execution matching vision... and I still see a disconnect.

Man I wish I was back inside...

It must happen non-intrusively, with a certain level of context sensitivity.

For the web, I guess that current methods such as those used in Online advertising (on various large websites, Google Adwords) or on shopping sites such as Amazon (suggestions, ohter buyers bought...) are a good way that works quite well, can be helpful if I want to look at it but still doesn't push itself too agressively. The next step are the non-web applications on the Desktop, where these concepts have come into play in the last years. Music players like iTunes and WinAmp are probably the "first generation" of these concepts. While you're listening to Music, the players display information about Musician and Title, then including additional information (other albums available, a way to buy them, perhaps song lyrics, current events/concerts of the artist, news items... ).

For the future, these concepts could become more ubiquitous, more intelligent and less intrusive, but more practical. The mobile devices coupled with location-aware, social and augmented reality content services are the optimal area to develop such concepts, as they possibly enable a much broader scope of interest to be integrated - not just based on a category of information (web) or music (players), but on the full spectrum of Maslows hierarchy of needs thanks to the additional "sensor" information they can provide:

• If a user is roaming (not in home SIM country), he has quite different needs than a user very close to his usual tower.
• When looking at the location context, daytime might be a factor (changing needs during the day)
• Observing user's speed and travel route might give a clue if he is going by foot, bike, car/bus or train. Content might change on that, too.
• Observing general location (city, suburbs, hinterland) might give additional factors.
• Read the current weather information about the actual location, compare to average data for the same place > interpret the current environment (too wet or hot summer day etc.)
• The device could sample audio input to get an idea about the general noise level by using the microphone of the device.
• Integrate social services such as Ovi contacts, Google Latitude

Mobile devices also allow a different range of uses due to the added (or differrent) usage concepts. One example are ebooks - no one reads ebooks on a desktop, but reading books is (ore at least can be) an important part of a mobile device usage. Now the content discovery concepts seen in the web and music players could well extend to the ebook application (as seen in the services around the Amazon Kindle and Sony (Kindle booksho, Sony eReader). Another example is the whole navigation/maps concept and the POI context. Or imagine Ovi Maps offering free routing (currently a 1.29 EUR value) in exchange for three audio advertisings (such as heard on local radio) being played 100m before passing the relevant location.

Finally, mobile devices also offer much richer interaction not only because of their sensory advantage, but also because of their mobility (with me 24h a day) and their usability: E.g. touchscreens are a great step forward for the quick/casual interaction needed for this type of applications to succeed. Touchscreens are quickly becoming a must in mobile devices, while the desktop lags behind.
Therefore, content discovery "on the web" might well become but a limited subset of what content discovery will be on a mobile device, because the latter offres a much richer environment.

In the end, success might well be defined by offering a flexible state-of-the-art hardware platform coupled with great software service integration. The Iphone shows how that can be done today (albeit in the limited scope of the iTunes concept) - tomorrow, a company either needs to offer at least the same usability, but for a much broader range of services. This could be done in various ways:

• In the closed monocultural Apple tradition (One hardware vendor, few closed devices, mostly closed services) . Challenge: Build/Buy large service landscape. Advantage: Full control.
• Or by offering a open plattform that integrates multiple services, either in a software- or hardware-centric way:
  • In the service-centric way (Google as the dominant service, but Android on a wide number of hardware platforms). Advantage: Centralized service platform. Challenge: Make hardware vendors follow the service provider.
  • In the hardware-centric way (Nokia). Build the hardware, open the plattform (Maemo), provide core services yourself (Ovi), integrate other open services (plug-ins, applications). Challenge: Integrate various dispersed services in a usable way. Advantage: Best chance to build broadest service landscape?

Perhaps (hopefully) the Nokia way might be the most successfull way to reach the broadest content network. Let's try with Maemo 5 :-)

I guess that if there would be a service that would... (bear with me ) be able to create and abstract a massive amount of metapersons (automatic personas, looking at a weird way), then each individual user would send and contribute a part of his own usage patterns for one (or more) metapersons, based on how well these metapersons current interests match your own interests. Data of your own interests would be initially only on your device, and this would be used to match and identify against a set of metapersons.

After this identification would be done, each metaperson would basically take in a set of data from multiple persons and blend it in a way that you couldn't reverse engineer the behavioural patterns of any single person even if you tried, but that would still be valid (as an average of a set of people with similar tastes) to have predictive power in what you might be interested in in the future.

... This to solve the privacy issues as well as to combine tastes of multiple persons into one entity.

@ragnar: let me get this straight.. Are you talking about setting up a proxy of some sort and implements a "genius"-like algorithm (like the onein iTunes)?

Hmmmm

but I think the ultimate value of this system is realized when you are 'honest' with it. It's simpler if the abstraction is implemented at the provider's side through some regulation or a marketing requirement.

Not a bad idea... Let's say all ~7-billion people on earth were connected to the interverse (ah, now my post-cyberpunk tendencies are showing). I'll bet, by using this method, there would only be ~200,000-metapeople in the database (not many at all). Each of these metapeople could be fed by all of the information that's being input into the system and, at some point, spawn new metapeople based on criteria that was triggered if a certain metaperson developed something that looked similar to multiple personalities. Initially, each new metaperson might look quite similar to their parent, but would quickly begin developing into something quite different...

A book in the making, eh?

Tim

William Gibson's netbots just found this discussion