The New Religion - Handling the Classification Serpent

I had taken on the new the religion but it wasn't sitting well. I was uneasy and very woozy. The old religion was still messing with my mind and for good reason. I had had one of the apostles in my office preaching me the virtues of Web2.o, specifically folksonomy versus taxonomy. Even though I argued Mendeleev and Kingdom, phylum, class, order, family, genus, species it was to no avail. He explained that if users classified content themselves you did not need to manage a hierarchy or enforce one or any other legacy from the past. Oh and it was just plain cheap!

As a bit of background the early search engines learned very early on that taxonomies – eg. hierarchical classification systems – were both limiting and very costly. LookSmart and Yahoo had to pay people to classify websites manually. After all the people who submitted sites for indexing could not classify them and metadata was useless. It was too obsessed with keywords and sex, drugs and rock and roll. I was working for Alta Vista at the time via Digital Equipment Corporation, God rest their soul, and I had suggested using colour coded listings as a classification method. I called it TrvialPursuit-onomy. We never did get to implement it because it added to much time to result page display but we did classify listings from Australia in green and those from NZ in red. Definitely not a folksonomy but direct and useful. I was also working for the Australian Yellow Pages online directory, the first Yellowpages directory on the net in the world. The problem with Yellowpages taxonomy (heading structure) which is really the product of directory businesses is that they are not logical. So its translation to the web made searching rather problematic. During this time, and don’t bother asking me what year because most of the Web2.0 people were still in diapers, we had master degree librarians on staff. Why you ask? - Because they have a great deal of expertise in classification theory. Can you imagine a library being classified using a folksonomy? Hey authors, you choose the classification that you want and hopefully people will be able to browse around and find your stuff. If I am sounding cynical about folksonomies it’s no accident.

So let’s get back to the main game here. My main focus in this blog is to talk about both the evolution and integration of networking structures into classified sites. I am talking cars, jobs, real estate and general classifieds. Now cars are easy because it has a fixed taxonomy – eg. make / model / version/ colour / options and usually make and model are enough. Real Estate is pretty simply too. My argument with jobs is that a resume is the taxonomy of job based classified site. Oh yes there are 6000 standard job classifications and etc etc (in fact at the last US Census there was over 100K discrete job classifications) – but the big problem is the nomenclature is not fixed and the big promise of keyword or even fixed classifications on a job site delivers nothing but pain in terms of finding anything quickly and reducing manual post search refinement. You cannot just search for a nurse and get any value. Wouldn’t it be great if every resume that came through could be sifted through and attributes of these resumes could be used to drive the classification system, the menus by which we find specific opportunities? Oh my God I think I have just become a convert to folksonomy. User defined classification, driven by users, owned by users but wait my mind hurts. Yes I am sorry Mr. Spock was the father I always wanted. All that logic and that stupid human half side ruining everything… but I digress.

The point is that neither taxonomy or folksonomy will work in context and purpose specific classified websites wishing to convert to the new religion. One must be very careful about choosing how to classify and how to enable search. The apostle of the new religion in his mad spouting of books I had to be reading and relevant other facts mentioned that humans can maintain 150 concepts in their head at any one given time. I am sorry, call me cynical, but I am thinking this is more like 7- at most. With this in mind I suggest that taxonomies will be profoundly more successful especially considering that Cletus and Byullah know nothing about the new religion at least not at this point. Maybe once the snake handlers start to tame and control the serpents of the new religion they will embrace it but at this point just give them shallow classifications systems. Make / model / price works for me and if it ain’t broken…..?

E-Bay probably does have some soul searching to embark on very soon. General classifieds will have to embrace the new religion sooner because finding a specific item may become more and more difficult but my advice is go back to old school and try hiring a librarian. One other thing worth mentioning is search. Your search tool should return results no matter what the user has searched for. There are several search tools which generate a taxonomy on the fly so that click based refinement is possible. I am currently working on enabling many different verticals with the new religion and I will be documenting the details of my torpor. I mentioned Mendeleev at the start and I will finish with Darwin. The key to success with the new religion is evolution. One must move slowly so that the value equation of finding stuff and buying it is not completely lost. My final word to the apostle was "if clothes pins are cheaper on site X who cares about blogs?"