Manifesto (Summer 2001)
Where is this WebLog coming from? Several threads of thought come together here, and the following is a summary as I saw them before I started out on this K-Blog research exercise.
Chaos, complexity and the buzzing, booming
The logical rationality
in most information systems,
Information as communication of intent.
Apart from a few islands, can these ever be brought
Now read on .... I hold these truths to be self-evident ... etc ...
A. BUSINESS ORGANISATIONS - I'm a stickler for ensuring fundamental principles are recognised, but very laid-back about detail, because real life business experiences tell me ....
A.1 That rational (logical reason) versus irrational (gut feel) decisions, and other well documented paradoxes, exist in the functioning of organisations.
A.2 That any business organisation, having a basically sound business proposition, survives on the common sense qualities of well-intentioned humans, rather than any best-laid rational plans to deliver that proposition.
A.3 That the success or lack of failure of most complex physical systems, is often the result of redundancy amongst many aspects, rather than good design of each and any one of them, even sometimes a design aspect that might at first have appeared to be the most critical. According to my calculations, that should have fallen down years ago!
A.4 That rabbits run. A huge amount of resource is spent managing (ie controlling) information publication, because of (well-founded) fear that a small leak of information out of context or without "full knowledge of the facts" (sic) can and does waste a similarly huge amount of resource, pursuing erroneous threads of activity, spreading the misinformation, and recovering from the confusion. This is actually one of the classic management paradoxes - is it better to simply give everyone all the facts or to attempt to give everyone only relevant facts. Answers on a postcard - need to know basis only, mum's the word.
A.5 That the 80/20 rule or something like it, is in action everywhere in business organisations or other complex systems. (So much so, that I have my own perverse view of the "rule" - something like - if on average the important 20% is better than 80% right, the other 80% may be less than 20% right, so in any given complex situation, the chances are that what you are seeing is only ever 32% right overall - frightening thought. ) Almost everything is on a sliding scale, a balance of probabilities, very little is black and white - The 80/20 Slide Rule perhaps.
A.6 What people do, what people say, and what people say they do are
three different things. When describing requirements or procedures of
things required to happen, humans rationalise the irrational according
to preconceptions about what is rational.
B. INFORMATION SYSTEMS - Experience of modelling information, specifying and implementing information systems tells me ...
B.1 That most "programming" in information systems relies on processing of logically defined instructions against a logically defined information model.
B.2 That most computer programs beyond the simplest of applications are unsatisfactory, perhaps for the most part useless in terms of their intended purpose, but being part of complex systems in use (A.2 / A.3), their "failure" either goes undetected, or their true value unrecognised. Controversial or what?
B.3 That more and more sophisticated information modelling, with less explicit and more implicit or abstract meta-modelling, appears to be an antidote to this until the point when an implementation is made, where explicit rules are again applied.
B.4 That paradoxically, what may be a better model, covering more aspects more precisely at the logical level may in fact result in an implementation less likely to achieve the desired results. Modelling more and more fine-grained detail may be obscuring the wood for the trees in implementation. Furthermore with more detail comes more complex relationships and interdependencies, so complexity itself may become the issue. (See D.)
B.5 That even the most capable programmers are often insulated from the woolly paradoxical reality of the application domain, by the logical framing of specifications and models for implementation.
B.6 That removing the woolly "business rules" from one piece of software implementation, simply leaves these aspects to be "programmed" elsewhere. If we're expecting information systems to get anywhere close to automating or even just modelling complex business (I prefer to aspire no higher than "decision support" in any computer application) - we'd better find a way of modelling and programming the woolly 80% (95%, 99% ?) of the problem.
C. HUMANS IN THE LOOP - cogitating on both of the above areas I find ....
C.1 I tend to think of information as active. From the perspective of the transitive verb - to inform, with "intent to communicate intended meaning", rather than anything static like data. Even captured at a point in time as data or "knowledge" it may be considered as a record of "intended messages".
C.2 I tend to think of all communication as human. Even automated messages have human origins in rules for their structure and content and for their sending. There is at least one real human in every loop, and only rule-based approximations to artificial humans elsewhere.
C.3 There are no facts, only interpretations of perceptions <of interpretations of perceptions> etc. of facts. Given humans in the loop the idea of facts being recorded or communicated becomes tenuous to say the least. You are brought back to interpretation of perception at every level from existence of fact, through observable phenomena, through formulation of model of observable facts, through formulation of decision or intent, through framing of message, through transmission of message, through observation of message phenomena on receipt, through model held by recipient, to interpretation and/or recording by recipient. Send three and fourpence is only the half of it !
C.4 Reification rules OK. The reality is traditional implementation of an information system must at some point reify (cast in concrete) a reality (!) which actually consists almost entirely of abstractions and perceptions. In fact A.6 says it is normal human nature to do this, to do otherwise needs to be learnt. The points at which we can physically bump into the assets or feel the money in our pockets are very important, but only a tiny part of the problem we face. There has to be a better way.
D. PROBABILITY, COMPLEXITY & CHAOS - All roads seem to lead to the idea of needing something other than rational deterministic logic in both modelling and implementation of systems .....
D.1 Having said that deterministic logic seems a poor basis for programming business information systems, introducing probabilities seems attractive. This is still rational of course, and simply shifts the rules to those which determine values and ranges of probability, say based on statistics.
D.2 Chaos theories point us to the fact that apparent randomness expressed in probabilities are likely to be phenomena or manifestations of systems (complex or simple) which are actually behaving chaotically - in the scientific sense.
D.3 Systems don't have to be complex to exhibit chaos, but undoubtedly business organisations, particularly viewed as extended enterprises, are about as complex as human ingenuity can create.
D.4 KISS - Keep it Simple Stupid. Having searched for simplifications of (models to represent) our complex systems, we'd do well not to forget that we were dealing with complex systems, and that chaos may well be at work when we choose how to implement them in software. We must beware cutting our own throats with Occam's Razor.
D.5 Described as the central paradox of AI, Jorn Barger refers to the question of how to describe any concept (from an arbitrarily complex whole set of concepts), using a finite (simple) vocabulary. e-Business XML Schema meets AI and gets a strange uncomfortable feeling, or is it just me ?
D.6 Entropy - the noise of non-compliant information beyond the bounds of a chosen simplification (model) may still be very significant because it can represent a drag or degrading influence on the model in use.
D.7 Fractal Thicket (coined also by Jorn Barger) - one way of looking at complexity is to consider a "fractal" arrangement of otherwise simple things. A good deal of the modelling I have been concerned with has been looking at decomposing complex messages or transaction sets into fundamental component "facts", whilst recognising that these could be part of many different hierarchical trees - the grove model, as in grove of trees. Neither is a perfect analogy, though the thicket implies more entanglement of the branches that the grove - both imply that the leaves each grow on one tree/bush only. We need an analogy closer to a symbiotic parasite or fungus, perhaps Jorn's thicket included such things as Ivy and Mistletoe, anyway I digress.
D.8 Suffice to say complexity, chaos and "proper" artificial intelligence are the areas I need to research.
E. SO WHY A WEBLOG ?
E.1 A formal research project looks an attractive way to consider the above issues, but the day job pays the bills. This way it is an online offline hobby, though in fact from what I can see, a large proportion of formal research is conducted via the web these days.
E.2 I've never been good at keeping a journal of significant thoughts. The idea of becoming a "blogger" arose it provides an unstructured means of capturing useful thoughts in a way that doesn't require effort to organise them at the moment they occur, from anywhere where I have web access.
E.3 Seeing other blogs and research projects on the web, from which I have already drawn useful ideas, I have become convinced that the open collaboration and exchange of ideas is a powerful means to explore and resolve complex issues. I guess you always have to suspect whether everyone is playing the game of give and take. The collaboration may be chaotic even anarchic, but you can always draw your own conclusion at any time - the web is not a committee - it's a free country.
E.4 The downside of web exposure is the risk of losing copyright on any valuable, exploitable interpretations across which you stumble. All my sources are acknowledged in the Links page and in the WebLog itself. (With any luck, I'll go back through any of the published text of mine, and include specific references where appropriate too.) I hope you'll do the same with anything you find here.