[abstraction] [aggregation] [archive] [class, classification] [data] [data architecture] [data warehouse] [generalization] [grain, granularity] [hauntology] [identity] [information] [interference] [leak, leakage] [negation] [object] [ontology] [palimpsest] [ratification] [reification] [screen]
Like many other things, LIGHT has changed in our understanding, from being particles to being waves to being something else.
This "progress" can be regarded as an increase in complexity or an increase in simplicity - or both.
Now perhaps it is the turn of INFORMATION to be rethought. Traditional notions of information (as particles of meaning) fail to account for interference, negation and other important phenomena.
Abstraction clears away some of the specifics, and allows us to see the structure. If abstraction is taken to the extreme, no specifics are left at all. Except for skilled mathematicians, who are trained to understand highly abstract structures with no direct relationship to the real world, most people find such a model incomprehensible. Thus abstraction should be practised in moderation, leaving a sufficient amount of specifics for the model to remain meaningful.
There are three methods of abstraction: aggregation, classification
|Veryard Project Papers||Abstraction Notes on Aggregation, Classification and Generalization|
|Veryard Project Papers||Abstraction by Aggregation|
But Derrida draws attention to the fact that the prefix arche
in both archive and architecture) represents a starting point or founding
act in both an ontological sense (this is whence it began) and a nomological
sense (this is whence it derives its authority).
|Veryard Project Papers||Archive Fever Review of
book by Derrida
Modelling History Notes on information and object modelling for historical and archive data
Information Scientists have usually assumed class membership can be defined monothetically, despite Wittgenstein’s famous counter-example, based on his definition of the class GAME.
Whereas operational entity types (such as EMPLOYEE) can usually be defined monothetically, strategic entity types (such as COMPETITIVE THREAT) often cannot. The behaviour of an entity is usually a characteristic feature, rather than a specific feature. A definition of a class in terms of what the occurrence might do (as with COMPETITOR or DANGEROUS DOG) is even more difficult to manage.
|Veryard Project Papers||Sorting and Classification|
|Recommended Book||Geoffrey C. Bowker & Susan Leigh Star, Sorting Things Out: Classification and its Consequences. MIT Press 1999|
Among the data there will be records of recent transactions and decisions, results of surveys and analyses, mixed up with a lot of much older stuff. But in order for an organization to assimilate these various data, the data must themselves be organized. And for learning to take place, data must be reorganized.
And this is where we slip into infinite loops. The organization that organizes and reorganizes its own data, its own memory, its own archive, is thereby organizing and reorganizing itself.
The traditional IT view of data is of something that can be Created, Read, Updated and Deleted - sometimes known as CRUD. This appears to be true of "physical" data storage, where the binary patterns on a magnetic disk can be rearranged or erased (although techniques exist for reading supposedly erased data - see Palimpsest). But from a "logical" business viewpoint there are only two meaningful operations: Read and Write.
|Veryard Project Papers||Data Modelling|
In my 1994 book, following Christopher Alexander, I postulated the ethic of the data architect as an ethic of repair: on a never-ending quest to mend the holes in the fabric. For the post-modern data architect, in contrast, the holes attract attention in their own right.
|Veryard Project Papers||Information Architecture (pdf)|
This approach separates two aspects of Data Warehousing:
|Veryard Project Papers||Virtual Data Warehouse|
Generalization is a useful way of reducing the number of classes or types in a model. Generalization is unavoidable in building an information model, since without any generalization at all, each type would only have one occurrence.
The key question is not whether to generalize at all, but how much to generalize, and where to stop generalizing.
Some people see the similarities between things more easily than they
see the differences, thus they want to lump the objects being modelled
into relatively few classes, to gain generality. Others tend to see the
differences more readily than the similarities, so they want to split the
model, to divide the objects between a larger number of more narrowly defined
classes, to gain precision. The tension between lumpers and splitters
a perennial one.
The destruction of an object does not result in the destruction of its representation. This is why people believe in ghosts - because they cannot accept that when loved ones die, their names and their memories no longer refer to anything - and because they cannot accept a similar lack of reference for their own ideas after their own death. Ideas may carry on, but no longer refer to anything outside themselves. The life of a concept continues for longer than the life of the physical object.
An image is grainy if the imprecision is visible - in other words, even if you can't see the individual dots, you can see that the image is composed of dots.
In information management, granularity refers to the degree of detail
or precision contained in data. In modelling, granularity refers to the
degree of detail and precision contained in a model.
Derrida invented the term ‘hauntology’ to refer to the logic of the ghost. In French, the word ‘hauntology’ sounds identical to the word ‘ontology’, which it is part of Derrida’s purpose to critique.
Traditional data methodologies have been fairly naïve about ontology, and Derrida’s coinword provides us with clues for deconstructing these methodologies and the systems that have resulted from them. But further, the word points to the fact that data frequent organizations, like the traces of past events.
Data are spectral/imaginary. They provide a view/image/spectre of the organization and/or its environment. Data architecture, on the other hand, is an attempt to impose a symbolic order/logic onto the raw data.
In the absence of daylight, we are haunted by ghosts and moths, which pass through walls and create holes in beloved fabrics. In my 1994 book, following Christopher Alexander, I postulated the ethic of the data architect as an ethic of repair: on a never-ending quest to mend the holes in the fabric. For the post-modern data architect, in contrast, the holes attract attention in their own right.
For many purposes, two things are the same if we cannot tell them apart.
Identity amounts to a lack of difference. And we can define information
as "a difference that makes a difference" (Bateson
- see below).
|Veryard Project Papers||Identity and Difference|
information equals data plus meaning
|The equation information equals data plus meaning remains important,
but it is not universally true, and needs to be demonstrated for particular
If meaning is the result of an interpretation, then the process of interpretation needs to be visible. If we accept Wittgenstein's slogan: meaning is use, then we often want to think of information as data in the context of some human (or at least conscious, intentional) use/purpose.
information as the unlikely
|This is a definition, taken from the communication theory of Shannon and Weaver, which lots of people cite -- and then ignore.|
all that is the case
|Wittgenstein defined the world in terms of information.|
a difference that makes a difference
|Bateson defined information in
terms of difference. Our decisions are based, not on the absolute value
of a given attribute of a given object, but on a comparison between values.
Information (as opposed to raw data) requires a context that makes such
Within traditional information modelling, an attribute that makes a difference is known as a state attribute. Differences are made up of states. States, in the form of conditions, can be composed into rules.
In school physics, you find out (with some surprise) that if you make two pinholes in a card, you don't just get twice as much light through. Similarly, if you have two items of information from two adjacent sources, that doesn't mean you've got twice as much information.
If I buy two newspapers, does that mean I get twice as much information? Of course not. And if an organization receives two different messages on the same subject from different sources, they may sometimes reinforce one another, sometimes cancel one another out, among other possibilities. As far as I know, these interference patterns in information have never been properly studied.Thus, like light, information isn't additive. Whereas data is additive: 50 bytes plus 50 bytes equals 100 bytes.
|Veryard Project Papers||Information Leakage|
Suppose that an announcement is made on the television news. There has been a fire at a sugar warehouse. Shortages of sugar are expected. Consumers are requested not to hoard sugar, and only buy enough for their short-term needs.
What's the likely response to this announcement? People will rush to the nearest supermarket and clear the shelves of sugar. The prediction of a sugar shortage is self-fulfilling.
Now suppose an opposite announcement. There has been a minor fire at a sugar warehouse, which was quickly put out. Stocks of sugar are unaffected, and no shortages of sugar are expected.
What's the likely response to this announcement? As before, people will
rush to the nearest supermarket and clear the shelves of sugar.
|Stubborn information - refuses to be erased, resists eradication. (Try to get rid of a bad credit report, a hostile press statement, or even a completely groundless rumour.)|
|Information goes underground - it's still there somewhere, even though it's not visible any more - and may come back to haunt you. (Something may seem to be forgotten, but it suddenly pops up again - and always at the most inconvenient time.)|
The OO view of object is remarkably similar. They have a public reception
area, known as the interface, and a private area, known as the internals.
The world consists of interactions between these objects.
|Veryard Project Papers||Reification as a form of Materialism
Software Objects and Mediaeval Thought
|Internet Links||Cetus Links: Objects and Components|
For practical purposes, ontologies are only interesting when they're different or volatile -- in other words when two people or organizations each have a different conception of what things and relationships there are in the world. Or perhaps the conceptions are themselves changing or unstable. These differences or changes have important implications for communication and collaboration between business and systems.
Establishing the ontology of a strange tribe – such as “users” or “customers”
– is fraught with error, misunderstanding and
misinterpretation. The American philosopher WVO Quine argued that we can never be entirely sure we’ve fully grasped someone else’s ontology – and that translation can therefore never be proved correct. Despite this, I still believe that better communication usually justifies practical effort.
Where feedback loops or learning loops are involved, punctuation also refers to the start point – where to break into the loop.
Punctuation is often subjective or arbitrary. Different analysts may punctuate a communication process in radically different ways. This critically affects the perceived structure of the situation. Is A helping B, or is B coaching A? Is C commanding D, or is D controlling C? Who is controlling whom, who is proactive and who reactive?
Punctuation is therefore an essential component of sensemaking.
When relationships are regarded as things, this usually focuses attention either on the bridging mechanism, or on a static snapshot of the relationship, as for example represented by a legal contract. When processes or services are regarded as things, this usually focuses attention on the deliverable or end-result.
The object-oriented way of describing components is extremely useful, especially for designing and managing components. It is also useful for describing the behaviour of components, and their performance in complex environments. But there are limitations to an object-oriented view of systems and components. Sometimes we need the reverse procedure - to understand things as dynamic clusters of activities and relationships. We call this ratification.
|Veryard Project Papers||Reification and Ratification
Software Objects and Mediaeval Thought
The computer offers information as services through a screen. The screen
is both literal and metaphorical. It is a surface on which the data are
presented, and also a filter that controls what the user sees. The screen
is a two-sided device -- it both reveals information and hides
Copyright © 1999-2003 Veryard Projects Ltd