The Problems of Ontology

Never mind "What's Stuff?", what's a knife?

Easy peasy. A knife is... er... a sharp edged, hand-held tool for cutting.

Or, to quote the OED (hope this counts as Fair Use) "A cutting instrument, consisting of a blade with a sharpened longitudinal edge fixed in a handle, either rigidly as in a table-, carving, or sheath-knife, or with a joint as in a pocket- or clasp-knife..."

Do you see it? The wood?

The essential observation to make is that a knife is defined in no small part by what it does, how it is used, and not by intrinsic, objective characteristics: the purpose or application of a thing would seem to be a defining characteristic. I could sharpen the edge of a credit card and use it as a knife, but would it be a knife? Maybe. Certainly if I bind it soundly to a good, strong carrot.

What about a knife that has lost it's handle? Is it still a knife - a handle-less knife? What about a knife that has lost it's blade? How can I even refer to a blade-less knife?

[I pick these examples for several reasons, one of which is that I do recall a Radio 1 DJ running a phone-in for words defining things that don't exist (strangely high-brow for Radio 1, but it was a long time ago), and - although I can't remember the word that was coined - I do remember a definition: "A blade-less knife without a handle."]

The Ontology of Everything must take such things into account.

The military like(s) to talk about Capability - we have a Rapid Response Capability, a Persistent ISTAR Capability, a Nuclear Deterrence Capability, u.s.w. - but, alas, one cannot have (or not have) an unqualified Capability.

What one can have are degrees (qualitative or quantitative) of Capability, where Capability becomes the ability to achieve a certain specified result (or perhaps class of results) under specified circumstances (e.g. environmental conditions, available resources, etc.).

An example I have used before involves the comparison of a Knife and a Screwdriver in terms of their "Capabilities".

A (good) knife is good for cutting and probably stabbing too. A screwdriver is not only good for tightening or removing screws, it too is probably a good stabber. And for certain types of screw (usually small slot headed screws) a knife can be an acceptable screwdriver.

So, what do you need for a Stabbing Capability? There are (at least) two choices. And what differences between a knife and a screwdriver should or could be used to separate them in an ontology... or what similarities would put them in the same category?

It would appear that an Ontology of Everything would be hard to keep orthogonal.

How can one construct an ontology of Things when Thingness is so fuzzy?

Ha!

Not telling (yet) But here's a clue: don't.

Stuff in All Generality - An Ontology of Everything

The discipline of organising Stuff into categories and saying such things as "Stuff B is a type of Stuff B" is called Ontology - particularly in the domain of computer science. [Of course if one is to be truly rigorous one needs to say also that "is a type of" is a binary operator relationship between "Types of Stuff"... which probably means that you have to say what Type means in order to define Type... is Type a Property of Stuff or... and before long a whole Chicken and Kettle of Fish Eggs Situation looms large...]

But, back to the beginning.

Because I had terrible trouble making sense of some of the MODAF, ISSE and UML models I encountered while dabbling in modelling and architectures for Peter Burge (before he became a god) and for Vega (who now provide the high-priests of the modelling/architecture temple that is - or was - Peter's ISSE Lab in Malvern) I decided to create an Ontology so that I could re-interpret the various concepts I was supposed to be working with and - hopefully - disambiguate and clarify them.

I like to think I was moderately successful, but it was only "Julian Stuff" so it has had little impact on the development of MODAF, which is I think a shame - because although MODAF can reasonably be described as "rigorous" (as indeed Ian Bailey describes it in the Wikipedia article), it is also in my opinion non-orthogonal - which is to say that definitions overlap and that you can create two models of the same thing (or two separate but related models) faithfully following MODAF and still end with descriptions that either look completely different or are generally incompatible (or both). Call it the inevitable result of design by committee.

To similise, modelling with MODAF (and I bet, DODAF and - to be fair - many other metamodels too in all probability) is a bit like writing in accordance with a grammar, but just because two authors can write grammatical English it doesn't mean that their writings can be merged seamlessly together. Unfortunately, since Enterprise Architecture (lots of coherent models created by a large number of independent modellers) is what MODAF is supposed to be all about, that's a shame.

Anyway, based on its relative success (it worked well in the very limited trial I was able to give it - almost but not quite a sample of one - and it seemed to accommodate the UJTL, METL's, JETL's, JTL and numerous other military Task Lists neatly), and had considerable utility for me in making sense of otherwise mutually unintelligible things, I thought it was time to start thinking about doing it properly and with greater scope in support of an absolutely wizard business idea, about which I shall be saying nothing more for the time being.

Of course, many have said that An Universal Ontology a.k.a "An Ontology of Everything" is impossible, and they may be right... but I shan't let that stop me trying, no siree!

However, I think I can say - without giving too much away - that the foundations of this Universal Ontology shall be Space, Time and Stuff...

... or is that the title of my autobiography?

Perhaps it should be Time, Space and Stuff... or Stuff, Space and Time? Careful with that comma, Eugene.

Who knows? I certainly don't.