Monday, August 01, 2005

Ontology is Overrated- Clay Shirky's talk

Clay Shirky writes about how categorization, as applied to the e-world, is more of a mis-fit than a helpful tool. He discusses Ontology in general and goes on to discuss it as applied to the modern electronic world. Read his very interesting piece at:

1 comment:

  1. I've been through Clay Shirky's piece: "Ontology is Overrated: Categories, Links and Tags" with some care - and find there is a considerable amount of good common sense in his contentions. For instance, right in his lead paragraph, he writes:
    > In particular, I want to convince you that many of the ways we're attempting to apply
    > categorization to the electronic world are actually a bad fit, because we've adopted habits of
    > mind that are left over from earlier strategies.
    You will observe that this is the exact thought behind the way we create categories in a Field Representation: "To begin with, cleanse mind of all pre-existing categories" - I don't know if it is even possible to accomplish that, but the basic idea is to prevent adopted habits of mind left over from earlier strategies to run our minds while we try to develop new strategies.

    In fact, to start with, I create FRs with something more general and more flexible than 'categories' - and this is reflected in the current OPMS software, which uses modeling questions about 'similarity'. However, I have found that most people find this level of abstraction difficult to use, so in future versions of the software, we shall have 'categorization' as the default, and 'similarity classes' as a specific choice to be made if desired.

    I'm not claiming that all tricky and difficult issues regarding categorization have been resolved through FR - I believe FR would require something more for it to become a truly effective tool to enable us to see the 'dimensions in the system'; however, I don't know what that 'something more' might turn out to be.

    All that said, I do believe that 'ontology' could well be a very useful concept for design as a whole (and for software design in particular) - but as Shirky claims, a whole lot has to be done.

    In general, regarding 'categories' at the most abstract level - in pure math, 'category theory' (along with the mappings in that theory - I believe these are called 'functors') is one of the most unsettled and difficult areas of pure math - which currently yields many more questions than answers.

    In any case, I have added Shirky's site to my reading list (also Paul Graham). Thanks.

    Broadly, yes, Shirky is probably correct in his remarks about 'ontology' - but I'm not knowledgeable enough to pass any kind of judgement on his views about 'Links', 'Tags', etc. I do know that the kind of categorisation enabled in software is generally less than adequately useful. Example: check out the categorization enabled in the 'Favorites' facility of MS Internet Explorer - I have found it to be pretty worthless. I am pretty certain the other browsers are equally unsatisfactory. This may appear to be a trivial example - but no, the categorisation enabled in MSIE is extremely inconvenient for users - and this is one feature that, in fact, prevents effective research being done on the myriad subjects that are available on the Net.