Thinking About Filters

I recently wrote about the idea that one might prefer to use a filter, rather than an inbox.  For clarity, I thought I woud add a few additional thoughts.

There is already one filter which is fairly widely used – Google News.  Its my opinion that an inbox / filter of the sort I am describing would end up looking quite a lot like Google News.  As the filter scoured the web (or took in the results of other web scourers) for content, it would collect similar content together, much as Google News collects news stories together.  As I user I would choose one of these areas to ‘zoom in’ on, which would give me access to a priority ordered list of potential content to read.  My choice would then both help identify the sort of thing I wanted to read more of, but also eliminate identical and nearly identical articles.

With this in mind, I might think a page would have different sections such as ‘incoming mail’, todos, groups of things to read.  Exactly what appeared in those groups would depend on a large number of factors including time of day, day of week, what I’ve been doing recently, where I am physically located, which computer I’m using.

Search would stop being ‘find data in an index’ and would become ‘open certain parts of the filter, and bias towards certain term’.  Search terms would stil be biased towards things that the filter has learned about you (so a UK-centric user searching for Football would find information about soccer rather than american rules – or certainly higher ranked).

I’ve talked about using external services to get more data – in effect it would work like this:  in the general use of the filter, I would be building up a personal index of pages I visit (or read RSS feeds of, say) and ‘close’ pages – pages closely linked to those.  When I search, I would open the filter to find examples of those pages which contain those search terms.  However the filter would also contact some known external sites – lets say Google and Wikipedia – to see what pages they have to offer.  The filter would then read those pages and add them to the general quorum of pages it knows about.  They would then have the chance of showing up in the filter’s search (but would not necessarily show up if you already have content which looks better for your needs)

I said that the filter could run on your home PC, or in the cloud.  In retrospect this was wrong.  It would have to run in the cloud.  I have a large number of devices, and more and more I want all my devices to sync together – the cloud is the place where this can be done.  Similarly, some of my devices are too dumb to run a sufficiently complicated filter, so again, we are looking at something running in the cloud.

When we start talking about things running in the cloud, a threat looms – what makes this different from, say Google or Facebook?  I think my answer is that with Google or Facebook, they hook the user by providing useful services, and in return get lots of data about the user.  This data is then used by google to sell targeted advertising.  In the filter model, things are slightly more complicated.  The filter begins to act a bit like a huge distributed market – people will push an advert (or what I’m going to consider ‘sponsored content’) to the user, offering to pay a certain amount if the user clicks on it.  The user (or more reasonably the user’s filter) returns how much it is willing to charge if the user clicks on the content.  For other content the user may offer to pay for it, and the content provider may set a charge.  In short, we are instituting a micropayment system, one which doesn’t require the user to actually put any money forward, if they are getting enough sponsored content that they are prepared to read…  the filter can increasingly make it clear that watching adverts is necessary if the user wants to continue reading things – or that the user can inject some cash of their own.  In any event, the advertisers will be paying the user rather than the equivalent of Google (the filter service provider).  The user will then pay the filter service provider from their amassed micropayments.  What this ultimately means is that the user of the filter becomes the customer – so the filter service is set up with the customer (and not the advertiser) in mind – indeed it becomes the filter services mission to ensure the customer sees as few adverts as possible, while enabling them to continue viewing the type of content they want (if the customer wants to see newly released movies, they are going to be watching lots of adverts, or injecting quite a lot of their own cash).  A final side effect of this is that whichever company builds a filter like this will become a major micropayment player and clearing house.

It occurs to me that this – rather than me-too plays such as Google Play or Google + is what Google should be working on now.  Whoever does manage to introduce the right type of filter engine could easily out google google, just as Google out altavista’ed altavista.

You must be logged in to post a comment.

© Ben.Cha.lmers.co.uk
CyberChimps