(Originally envisaged to cover 2004-06!)
Enough things have come together that it is now appropriate for me to ensure I can dedicate the next three years to completing the software I have been describing as TransForum 2 (TF2) and applying TF2 to an independent research project within the broad program outlined in Stephen Wolfram's A New Kind of Science (NKS). This page outlines the objectives of the Three Year Plan, opportunities for commercial and other involvement, and the broad current and emerging context in which the plan is situated.
TF2 is a highly customisable platform for databased collaboration implemented in an open Internet systems environment realised by five interdependent primary functions:
While similar functions are provided in various groupings by other approaches, TF2 is designed to focus its users on achieving and presenting collaborative outcomes.
TF2 is informed by a design heritage that I started work on in 1981, initially in the form of a system for Public Information Communications and Access (PICA), along with earlier and contemporary work of others, especially on hypertext, collaboration and human-computer interaction. Over more than two decades, those design ideas have been applied to a broad range of real world applications, including practical experience with the development of Internet communities across the past decade.
Theoretical grounding for the use of computer communications for collaboration were brought into particular focus by Winograd and Flores 1986 publication of Understanding Computers and Cognition and more broadly by the first Computer-supported Cooperative Work (CSCW) conference the same year. Within a year of the first graphical Web browser being made available on personal computers, my own MSc research report, Why The Web?, expounded the reasons the World Wide Web was delivering what earlier efforts had aimed to achieve without success. TransForum is also influenced by my privately presented 1997 paper, Conversation Piece.
On a then separate track in 1983 I became involved with cellular automata (CA), my Pattern Breeder CA being described in Scientific American in 1986. Stephen Wolfram became even more involved in CA around the same time and I got to know him both through that interest and through my then company, PICA Pty Ltd, representing Wolfram's Mathematica® software in Australia. For more than a decade Wolfram has used the success of Mathematica to allow him to focus primarily on further research into CA and similar "simple" systems which resulted in his very successful publication of NKS in 2002. During 2003 the return of focus to CA and the maturation of available CA software has enabled me to obtain significant preliminary results from accelerated research in areas I first explored during the 1980s.
The ultimate objective of the three year plan is to have TF2 and its successors accepted as a project within the open systems community as the best way of guaranteeing ongoing user support. The complementary objective is to retain control of design and implementation only for as long as is needed to ensure conceptual integrity.
There are several likely milestone objectives concerning CA research, uptake of preexisting community applications, development and deployment of new applications, and presentation of results to the NKS, CSCW and open systems communities. Milestone objectives identified in the remainder of this part should be considered more indicative than definitive.
A preliminary implementation of TF2 should suffice to support the interactive dissemination and critique of the results of my current CA research projects, Life in a Tube and Trapper. A more mature implementation might be needed to deploy a proposed public collaboration experiment, Hive, results from which are likely to continue to come in well beyond the period of the three year plan.
Amongst the very first objectives for TF2 will be the systematic upgrading and continued operation of several still operational original TransForum sites. After that it will be appropriate to develop and deploy uptake mechanisms to allow at least one other legacy Internet community format to be upgraded systematically to TF2.
During 2000-2003 working through our premium Internet hosting business, Webmastery, a trading operation of TransForum Pty Ltd, we have deployed client applications built on more and more components which will, in their next iteration, form much of TF2. It is likely that long planned work on Mietta's restaurant database will be based on a preliminary implementation of TF2 and that at least one larger client system will be converted to TF2 once it starts to stabilise. Such applications will involve other developers and/or the clients themselves in much of the customisation facilitated by TF2.
There should be ample material for presentations at NKS conferences in each or the three years and at CSCW and open systems conferences by the second year. In some cases there may be opportunities for journal publication complementary or as an alternative to conference presentations. It is also of the very nature of TF2 that it should be able to host the ongoing work of at least one such community.
In keeping with its focus on developing collaboration, this Three Year Plan should both provide first mover commercial opportunities and fit well into the framework of a computing science PhD program. The claim to be tested in such a research program would be that genuine innovation can be achieved within the framework of the open systems and standards detailed in the final part below.
While it would be possible to detail numerous examples of potential first mover advantage from early application of TF2, those examples might only serve to get in the way of the larger picture. As well as those explicitly mentioned or alluded to above as milestone objectives, I have long lost count of the potential applications for data intensive collaboration on which I have done significant work as a developer or as an expert user, to say nothing of those I have become aware of through the work of colleagues. So rather than detailed examples, I should try to briefly characterise a typical TF2 application.
The primary target for TF2 is custom rather than mass market functionality. It is aimed at the smallish business or group which is based on unusual or unique expertise which a keen user community will not just turn to, but also contribute to. A TF2-based implementation will manage base data and ongoing data processing as well as ad hoc and often impassioned discussion of issues at hand.
Such applications cannot afford the kind of large scale dedicated software development that is the norm for big business and mass markets. They need a very low cost of entry and a system with which they can grow and further customise their use as requirements evolve, most often with contributed assistance from members or even customers. The primary point of TF2 is to provide a conceptually light weight bootstrap system which provides a minimal obvious set of initial functionality specific to the individual application, but that bootstrap also provides a solid functional foundation upon which services can grow as needed.
Although only minor enhancements have been made to the original TransForum since 1999, through the 18 month period of its initial development and deployment and through the reapplication and extension of underlying technologies during subsequent years, a handful of other developers have utilised at least some parts for their own projects and clients. It is expected that those developers will be the first of a growing group who expand the application of TF2.
There are limits to what so-called bleeding edge customers will be prepared to pay to obtain first mover advantage. However, experience with the original version of TransForum and with subsequent client projects utilising developing TF2 technology suggests that there will continue to be some income from client development work over the Three Year Plan. The challenge is to not allow the need for a bit of extra income today to get in the way of the integrity of TF2, so it is highly desirable that we also attract other forms of commercial support. It is therefore hoped to attract some practical support for this Three Year Plan from some information technology vendors and even from one or more entrepreneurs who can see the first mover benefits to be gained from their association.
On the academic front, the oft repeated claim that open systems development imitates rather than innovates has been hard to prove or disprove based on available evidence. This is an important question for computing science and a question which might be best tested by supplementing practical approaches from a long ago BSc and a career's worth of experience and professional development in computer science with theoretical perspectives from my more recent MSc studies in science-technology-society.
There is more than one way in which a computing science department would stand to gain from providing academic oversight of relevant aspects of this Three Year Plan. While much of the remaining development work on TF2 is novel mainly in the combination of services rather than the services themselves, further enhancements are planned for TransForum's innovative method for presenting and managing hierarchical indexes. As mentioned, TF2's combination of application services is particularly geared towards ensuring outcomes from conversations.
The Australian government's recent prioritisation of complex systems research may provide an even stronger justification for a computing science department wishing to stake out territory in the heartland complex systems areas defined by NKS and the CA work that is a key component of this Three Year Plan.
Without trying to look too far into the future, given three years progress with TF2 and with the open systems and standards detailed below, there promises to be no shortage of potential follow up projects. And while computing science departments are not usually overly concerned about the application of computing to the core learning and research activities of their institutions, TF2 draws on a heritage of involvement in education technology policy formulation during the 1990s, and so is intended to be able to support group interaction related to specialised course delivery as well as it will support research collaborations.
The technical context for this Three Year Plan is provided by several of the most prominent ongoing open systems projects and a still developing set of Internet and Web standards. In part, the success or otherwise of NKS will be measured by the degree to which CA graduate from their original categorisation as recreational mathematics to a serious academic pursuit.
TF2 is being implemented in Apache ModPerl with a MySQL database running on Linux and other Unix servers through standards-compliant client Web browsers. It is geared to take advantage of newer versions of the server software, especially Apache 2, Perl 5.10 going to Perl 5.12 or, if ready, Perl 6, and MySQL 4 going to MySQL 5. The Mozilla project is expected to continue to be the dominant source of standards-compliant browsers.
It would be too much detail for our purposes here to try to identify all the Unix flavours and services that are assumed part of a modern developer's tool kit. By its very nature, TF2 will become more and more self documenting, potentially to the point of serving as its own front end to the likes of CVS or equivalent and to e-mail. It is also expected that Mac OS X will continue to have a role to play by virtue of its combination of user convenience and Unix services.
The World Wide Web Consortium (W3C) is steadily fleshing out the next generation of Web standards based on the solid foundation of XML. Even beyond its use in subsidiary standards, XML provides an implicit style guide for configuration files, typified by Apache, and for content templates, such as our own TransForum::Proforma, even where it would not be appropriate to try to process such data as formal nor even fully compliant XML.
The previously mentioned original methods for presenting and managing hierarchical indexes have been designed to take advantage of whichever practical standard emerges for browser-based manipulation of 2D schematics. While SVG should deliver XML-compliant services in this area well within the three years, we have an acceptable fall back position in the form of MacroMedia's support of SWF as a public standard for its Flash technology, comparable in their respective roles to Adobe's public yet proprietary PostScript and PDF formats. The important point here will be to be able to deal with key structural data in the DOM using ECMAScript however the components are combined into a "page" by a browser. Dynamic manipulation of similar schematic diagrams is also expected to enhance the presentation of results from some CA experiments, alongside the venerable animated GIF format now coming free of patent restrictions.
Perl was chosen for the original TransForum and continues to be the main development language for TF2 because it facilitates, without obligating, separation of functionality from content, presentation and layout. Separation and customisability are further improved through the use of style sheets together with XHTML based Proforma templates, the rendering process for which provides a base set of database query support without need for custom functionality. Setup and customisation of a TF2 installation will be through a basic template editor, a generic database table editor and a style picker, all of which will be available to the installation administrator(s).
During and following his research and writing of NKS, Wolfram extended the functionality of Mathematica to facilitate his and others' related work. It will be desirable to strike a balance between ensuring results of further work can be efficiently exchanged with those for whom Mathematica will remain the platform of choice, while also looking outside Mathematica, and any constraints it may inadvertently introduce, at other suitable platforms for "simple" systems. My initial research into Life in a Tube and Trapper has been done using Andrew Trevorrow's excellent shareware 1/2D CA program LifeLab and my own original Perl programming. The planned second phase Hive research will require some efficient custom algorithms for CA generation, rendering and user interface as well as a significant TF2 installation for coordination, data collection and analysis.
This Three Year Plan should produce and/or facilitate:
It is hoped that the commercial and academic value of these outcomes will justify not just the further investment of time and resources that will be required over the three years but also a fair share of the many years and many threads of prior work on which the Three Year Plan builds.
[ Memes index ] [ Early TransForum deployment ] [ A simple evolving network ] [ Selected writings 1992-97 ]