Internet commentary

Citation

Andrew, A.M. (2000), "Internet commentary", Kybernetes, Vol. 29 No. 1. https://doi.org/10.1108/k.2000.06729aag.001

Internet commentary

Keywords: Cybernetics, Systems, Internet

Abstract Comments on the Internet, giving details of new facilities and provisions that are of specific interest to systemists and cyberneticians. They include, in Part I: Cybernetics Discussion Group (CYBCOM) on the status of some contributors to cybernetic thought and on the basis of cybernetics and general systems theory, and distinctions between the two. Part II reviews the Hermitage Museum Project undertaken by the museum jointly with IBM, and in particular the plan to form a digital collection of the art works of the museum, available over the Internet. Special features are also highlighted including search facilities utilising graphic input.

Part I - Cybernetics Discussion Group

The month of June 1999 has seen an exceptional burst of activity in this discussion group, with no sign of an abatement at the time of preparing this commentary in early July. The activity is of special interest since it reflects on some fundamental issues in cybernetics and general systems theory. The Cybernetics Discussion Group is the title of the extremely useful automatic listserve facility associated with the American Society for Cybernetics and with the George Washington University. As mentioned in an earlier commentary, the facility can be joined by sending an e-mail to the address: listserv@hermes.circ.gwu.edu containing in the body of the message the line: SUBSCRIBE CYBCOM firstname lastname

The recent burst of activity seems to have been triggered by a circulated message from John Warfield, on 24 May, nominating the writer Michel Foucault (1926-1984) as a "Founder of Discursivity" (a term due to Foucault himself but not used self-referentially by him) in the context of cybernetics. A Founder of Discursivity is someone whose ideas come to pervade a subject area, sometimes colloquially indicated by saying the ideas "have legs".

John Warfield is himself an established figure in cybernetics and systems. (A convenient, though somewhat arbitrary, criterion of venerability is inclusion in the Bibliography compiled by Klir and Rogers (1977), which gives Warfield no fewer then 11 entries.) In his contribution to the discussion group, he is a little cavalier in the matter of giving precise references, saying that the books can all be found in the Amazon list. No doubt such references to published books will become increasingly acceptable, at least within communications by e-mail or on the Internet, since the Amazon company is a pioneer in providing an on-line book searching and ordering service. In Britain, the Website at which the Amazon lists can be consulted is: http://www. amazon.co.uk .

The Amazon search facility produced no less than 185 matches to the keyword Foucault, all but a handful of them referring to Michel Foucault, either as author or as part of the title. The particular books to which John Warfield makes reference are The Archaeology of Knowledge, by Foucault and The Foucault Reader. The former of these has appeared as a paperback of 224 pages, published by Routledge in 1990, ISBN 041504537 1, and price from Amazon £14.99.

The Foucault Reader is also available as a paperback, either in a 1984 edition by Foucault himself, or as a 1991 edition attributed to Michel Foucault and Paul Rabinow, both available from Amazon at discounted prices below £10.

In the same e-mail message, Warfield also refers to the contributions of Geoffrey Vickers (who has the distinction of having become rich by applying his theories). Vickers is quoted as criticising the use of certain terms, including system and feedback, in the theoretical context, insofar as they tend to be unduly narrowly defined. The term "linguistic pollution"is used to denote this, and some other examples of unproductive misuse of language.

The contribution of Michel Foucault, at least as far as cybernetics is concerned, was contested in a later message by Klaus Krippendorff, of the University of Pennsylvania, another well-established figure in the field. He contested the view that Foucault deserves recognition as a contributor to cybernetics, at least in the modern sense of second-order cybernetics. His arguments, and those of other contributors, raise the question of the distinction between cybernetics and general systems theory, which stimulated further interesting contributions. Both Foucault and Vickers are more directly under the banner of systems theory than of cybernetics, but the distinction is acknowledged to be fine and possibly controversial. Ashby chose to be seen as a pioneer of cybernetics, according to his choice of titles for his publications, but his theories depend on clarification of the idea of a system and he is often quoted by systems theorists.

As remarked by one of the contributors, the difference between the two subject-areas, at least in their early stages, could be said to amount to the prime attention paid to communication and control in cybernetics, de-emphasising the origin of the signals or commands to be conveyed or obeyed. This seems to make cybernetics less general than systems theory, since in systems that are not man-made it is not necessarily obvious, in a re-entrant net of influences, which point in the net, if any, is best seen as exerting control. It is acknowledged, however, that the gap between the two subject-areas became narrower as cybernetics developed and attention was increasingly paid to complex systems exhibiting self-organisation.

John Waxfield is dismissive of much of the content of both subject-areas, and in a message entitled "Well-formed subjects?" of 1 July, he makes the provocative assertion that the content of both subjects is almost a null-set. In defence of this he points out that artefacts embodying feedback were invented and understood long before Wiener and Ashby came on the scene. (To show how the true pioneers have fallen into obscurity he mentions a development published in Russian in 1875, which he still leaves unnamed.) Since Warfield has edited two systems journals, over a total of 13 years, and has made major contributions, his viewpoint merits attention, though the vigour with which he argues various points in the present discussion suggests his disillusionment is not so complete as he claims.

In trying to be precise about the growth of "systems language", a quotation from the philosopher Heidegger is invoked. A case is made for a language that stresses relationships rather than objects, and it is argued that certain Andean and also Australasian languages have this property. The idea of a distinction between outside and inside, as an elementary perceptual event, receives attention, and this has a connection with the theories of Peirce and with the calculus of Spencer-Brown. The nature of "identity" in mathematics is treated and shown to be more complex than is generally supposed. The contributions of Willard Gibbs are mentioned, and a definition of "system" due to him is shown to embody the same subjective component that has been stressed by Ashby and others.

Much of the discussion is directed to the difference between the approaches of cybernetics and general systems theory. Cyberneticians are criticised for adhering too closely to "constructivist epistemology". There seems to be a feeling that the differences are largely artificial, and that the distinct formalisms of the two subjects are an artificial barrier to cross-communication. Stuart Umpleby, one of the administrators of the discussion group, enters here to consider what would happen if cyberneticians and systems theorists changed roles and argued from the other viewpoint. There is the suggestion that both sides communicate more effectively with the general public than with each other.

A wide-ranging discussion of this kind is impossible to summarise effectively, but perhaps something of the flavour has been conveyed. The amount of interest aroused can be demonstrated by giving numbers of messages under some of the headings. The name of Foucault occurs in the headings of nine messages and that of Heidegger in the headings of 35. A further 21 refer to "Peirce and distinctions" and 38 to "relational languaging" or "relational languages". It can be seen that membership of this discussion group is well worth while and that, at times, significant material emerges at a formidable rate. The possibility of this kind of discussion, free of geographical limits and occurring with rapid response because of e-mail and an automatic list server, is a major development facilitated by the Internet. Virtually instantaneous response allows interaction that is enormously more vigorous than is possible in the columns of printed journals, and is convenient and almost free of charge.

Part II - The Hermitage Museum Project

A real feast can be found by opening the Website: http://www.hermitage museum.org . In a note dated 15 June 1999, it is explained that a multi-year partnership between the museum and IBM has produced a digital library of masterpieces of art and a Website claimed to be the first of its kind. A number of developments local to the museum itself are also mentioned, including multimedia navigation kiosks within it, an image creation studio, and an education and technology centre. These result from an IBM technology grant of nearly two million dollars, and the entirely believable claim is made that the developments position the Hermitage as one of the most technologically proficient museums in the world. The date of 15 June is that of the unveiling ceremony.

The Website allows viewing of more than 2,000 works of art, ranging from ancient Russian icons to French Impressionist paintings. It is intended that eventually the whole Hermitage collection of three million works of art will be included. These will be listed in 12 categories (paintings, sculpture, jewellery, etc.). Java-based applications allow electronic "visitors" to zoom in on parts of an image at will. It is also possible to select a feature called PanoramIX which allows virtual reality tours of halls that have been photographed in three dimensions.

A further feature is comprehensive search facilities, by which a particular work of art can be found by inserting incomplete details of title, artist's name etc. An intriguing addition to this is means of searching based on pictorial recollection, which can be of the proportions of a selection of colours, or can be expressed in a rough coloured sketch produced by the "visitor" using simple tools of the kind provided in graphics packages such as paint. The facility is termed "Query by Image Content" (QBIC) and the two forms of search are respectively QBIC colour search and QBIC layout search.

The virtual tour facility, or PanoramIX, depends on an item of software called the IBM HotMedia Java applet, which downloads automatically to the "visitor" computer, with no manual installation necessary. The applet allows for smooth panning and zooming controlled by mouse movements.

The non-Web facilities

Although the Web-related aspects of the project are likely to be of most interest to people who do not often visit St Petersburg, the features that will operate more locally are also important innovations. (It is worth noting in passing that the Website is also of interest when a physical visit to the museum is contemplated since it shows plans and gives advice on optimising a tour, as well as showing items from the museum shop.)

One of the innovations is the four multimedia navigation kiosks to be installed at points in the museum, which will obviously be well equipped to inform visitors about the works of art around them. The Education and Technology Centre is to be equipped with the means of recording and displaying works of art digitally, with techniques for zooming in from different angles. One area of research will be the detailed comparison of the art of different eras and religions and civilizations, so as to detect similarities that can suggest cultural connections.

The other facility mentioned is the image creation studio, which is to be a centre for the high-quality digital colour photography on which almost everything else depends. Colours will be recorded with fidelity never before achieved in photography, using multiple images taken through colour filters and subsequently combined. The images will be compressed and stored without loss of fidelity.

The virtual tours

At the time of preparing this commentary in early July 1999, only two virtual tours were on offer, but these were quite magnificent and served very well to illustrate the principle. One tour was entitled "The Jordan Staircase" and the other "The Concert Hall". Both were impressive - to my mind the Jordan Staircase especially so as it shows the characteristic architecture and decoration of the Hermitage. The view seen was as though from a video camera that could be panned, tilted and zoomed at the command of the viewer. Panning and tilting were controlled by moving the mouse, with the mouse button depressed, sideways for panning and forward and back for tilting. This is the scheme referred to as HotMedia, and an introductory note on its use can be displayed on request.

The introductory note is actually slightly wrong in one respect since it gives the same instruction for tilting as for zooming, suggesting that both are achieved by forward and back movement of the mouse with the mouse button pressed. On a PC with a standard two-button mouse it turned out that the effect of such movement was to produce tilting when the usual (left-hand) mouse button was used, and zooming when the right-hand button was used. This, however, was not made clear in the instructions. The HotMedia instructions also explain, with an animated display, how to zoom in on any part of any of the static images of art works.

Query by image content

The QBIC scheme is experimental and presumably its usefulness in helping users to identify dimly-remembered pictures will only become clear after some time. In response to any input, the system finds the 36 pictures that give the best match, and lists them in order of goodness of fit. The listing is in the form of catalogue entries, with the title of the picture and the artist's name, as well as a "thumbnail" version of the picture. Clicking on the thumbnail version with the mouse pointer produces a larger representation.

When colour search is used, the user makes a series of colour selections. This can be done by moving a point around within a rectangle displaying all possible colour combinations. The brightness of the colours is controlled by moving a slider along a scale of degrees of blackness. Whatever colour is chosen, three separate sliders indicate the contributions of three primary colours, and the user can, if preferred, adjust these sliders directly.

The various colours specified in this way are allocated space in another rectangle, with means of adjusting the proportion assigned to each. The search process is then begun and delivers its 36 candidates.

For layout search, colours are generated exactly as for colour search, but they are then used to form a crude representation of the recollected picture. Again the search process delivers 36 candidates.

It is of course amusing to insert arbitrary inputs to both kinds of search and to see what emerges. I have to report that my attempts to make a more serious test of the scheme were somewhat disappointing. I was initially over-impressed by the large number of art works included in the digital collection and forgot how small a proportion this is of the total in the museum I inserted details, through both versions of QBIC, that were intended to identify the picture of a lady's head on the cover of a book of reproductions of Western European paintings in the Hermitage. The cover picture is in fact a part of the portrait of the actress Antonia Zárate, by Goya. My attempts to identify the picture using QBIC were unsuccessful, and a straightforward search by artist's name showed a good reason for this, namely that no works by Goya are currently included in the digital collection.

However, my attempts to prompt the search schemes to include a particular Picasso that is included in the collection were also unsuccessful (in that the target picture was not included in any of the sets of 36!). Fairly certainly, one reason for this was that there were substantial colour differences between the reproduction that I used to generate input to the search schemes, and the picture as displayed on the computer screen. I was now using a Hermitage Guide dated 1979, and the colour printing, although showing detail quite well, does not give accurate colour reproduction.

The matter of colour representation is complex, and presumably even the colours of the originals will change as they age and are restored. Undoubtedly, all the aspects are being studied in the Education and Technology Centre, where one of the results should be a great deal of data on long-term behaviour of paints and pigments.

Commercial security

In making this vast collection of high-quality reproductions of art works freely available, the members of the partnership would seem to be wide open to various forms of commercial exploitation since computer images are easily turned into high-quality prints. It is mentioned, although full details are not given, that the images are in fact tagged in a way that lets them be identified, though nothing is apparent to visual inspection.

The whole project has many exciting novel features, and for the many computer users who seldom, if ever, find themselves in St Petersburg (Russian Federation, not Florida!) full access to the collection will be a marvellous boon.

Alex M. Andrew

Reference

Klir, G.J. and Rogers, G. (Eds) (1977), Basic and Applied General Systems Research: A Bibliography, State University of New York, Binghamton, NY.