The following paper by Andrius Kulikauskas concludes his investigation Definitions of Sequences, Hierarchies and Networks Exist That are Most Natural for Arranging Thoughts at the Minciu Sodas laboratory.

This paper is a starting point for the work of the Infrared Data Association's Special Interest Group Flow of Experiences.  It belongs to the public domain, which means that you may copy it without asking permission.  Andrius Kulikauskas appreciates your suggestions on the content of the paper, and where it might be usefully presented or published.  Thank you to Raimundas Vaitkevicius and Steve Raiff for their comments!


 Developing Import/Export Standards for Aggregates of Notes

Abstract:  Different kinds of writing promote different kinds of thinking. Functional, anthropological, psychological, and philosophical criteria can be applied to define sequences, hierarchies, and networks so that they accord with their use in working with thoughts. The transfer of irrelevant structural information can distort and cloud thinking.  Separate formats should exist for handling unstructured thoughts (emphasizing punctuation), structured thoughts (emphasizing manipulation), and restructured thoughts (emphasizing visualization).  There is a need for a standard for the import and export of sequences, hierarchies, and networks.

1. The Need for a Standard

Arranging and rearranging notes is a remarkable way of promoting different kinds of thinking. A sequence of notes helps us distinguish strong and weak ideas, a tree of notes - broad and narrow, and a web of notes helps distinguish vague and clear ideas.

Writing down thoughts in a computer on a daily basis over several years can yield a collection of ten thousand notes.  Unfortunately, notes get trapped when software products are discontinued, as happened even with the best of products, such as Lotus Agenda and Netmanage Ecco. There is a need for an import/export standard for sequences, hierarchies, and networks of notes.  As things stand, individual and corporate users cannot risk using these products for long term projects, and so the products are not generating the revenues nor satisfying the challenges that might fuel their maturation.

The Minciu Sodas virtual laboratory is devoted to caring about thinking, and is providing leadership to develop and apply a Threebook standard for the transfer of sequences, hierarchies, and networks of notes.  (The Threebook standard is named in honor of Kestas Augutis’ educational project 3 Knygos) [1].  Use of the standard will require nothing more than the design of converters between the standard and the file formats of interested vendors.

It is important to define the standard correctly so that it does not cloud thinking by transfering irrelevant structure.  A correctly defined standard will make users mindful of the choices that they need to make as they transfer notes from one structural context to another.

In this paper, I describe my work to define distinct but related formats for unstructured thoughts, structured thoughts, and restructured thoughts.  I list structural questions that arise, and present several criteria for arriving at answers.  I show how to apply these criteria to answer the structural questions, at least tentatively, and expose the need for defining different formats for unstructured, structured, and restructured thoughts.

This paper asserts the point of view of the writer of thoughts, rather than the reader of thoughts. The main result is that arranging thoughts involves more restrictive definitions of sequences, hierarchies, and networks, than does visualizing thoughts.  Visualizing is based on restructuring, which takes us a step away from the decisions of the authoritative thinker.

2. Structural Questions

In the area of systems integration, where there are many well established semantic data models, there is a wealth of evidence for creating a low-level hypergraph-based data model for inter-model transformations [2].  There also exists a data visualization programming environment, VANISH, which allows one to port information from one visualization domain to another by treating each as a tree, and where necessary, breaking the cycles of networks so as to present them as trees. [3]. However, the transformation of aggregates of thoughts presents philosophical challenges because sequences, hierarchies, and networks of thoughts are not simply varieties of a common structure, but rather different strategies for arranging thoughts that promote very different effects on thinking.  Mathematically, a sequence may be a hierarchy and may be a network, yet the purpose of arranging thoughts in a sequence is very different from that of arranging in a hierarchy, or arranging in a network.  The most helpful definitions will ultimately be those that heighten these different purposes, which are ultimately evident not in the shapes of the diagrams, but in the intent of the author, for whom the import/export standard is being created.

Software for writing and arranging thoughts is underdeveloped, and very few people have used it for serious long term projects involving continuous reflection and reorganization.   Natrificial Software Technologies arguably produces the most advanced commercial product, The Brain, and encourages publishing of The Brain on the web at their Thinktank, but so far only a handful of authors have published personal files with more than one thousand thoughts, the most notable being Jerry Michalski’s, with more than ten thousand [4].  I myself have experience working with a comparable collection of paragraphs with a hypertext editor as well as a hierarchical editor of my own construction [5].  Experience from the use of existing products needs to be supplemented with reflection on our purposes as authors if our standard is to do justice to the difficult structural questions that arise.

Imagine arranging ideas in a tree, that is, hierarchy, with the purpose of distinguishing broader and narrower ideas.  Note how difficult it is to visualize the tree without introducing irrelevant structural distinctions.  For example, if the tree is presented in outline form, then the branches end up ordered, even though this was not the intention.  Or even if the tree is presented as a planar diagram, with branches distributed around the root in all directions, we still have the problem that certain branches end up adjacent, and others do not, even though this was not the intention.  It is of major importance not to force a person to introduce irrelevant structural data which will distort the picture of the whole.  The transfer of information to another environment is the time at which such irrelevancies should be acknowledged and discarded.  It is important to let a person discard the irrelevant information, or at least make it clear to them, that if they do not decide which structural information is truly relevant, then they risk distorting and watering down their thinking, and generating unnecessarily awkward structures in the new environment.

With this example in mind, it becomes a challenge to answer the structural question, Given a parent node, may its children nodes be ordered or not? An even greater challenge is the more general question, may the children nodes have any partial order?

It is helpful to review existing standards of information, as in [6], to generate additional structural questions (with S = sequence, H = hierarchy, N = Network) such as:

3. Criteria for Defining Formats

Four criteria have proven helpful in addressing the structural questions above.  The functional criterion considers the global purposes of a thinker arranging thoughts. The anthrophological criterion considers the ways of organizing thoughts that may be found in human practice. The psychological criterion is perhaps the most important, in that it considers the effects that the structures have on the thinker using them. The philosophical criterion considers theoretical models that derive the structures and account for and clarify their observed purposes, uses, and effects [7].

The functional criterion clarifies what we mean by thinker, and her work of arranging thoughts.  Why arrange thoughts?  An arranger composes relations between thoughts.  The thoughts themselves should be isolated! and not place any restrictions on the arranger.  In this sense, thoughts are encapsulations, mental punctuations.  They can, in principle, be anything: text, parameters, images, code.  What is more relevant, a thought should express a mental expectation, the basis for an emotional response.  It could be a single word, or an entire document, but in general would be something in between, such as an idea, a paragraph, a note. The thinker wields cognitive authority, and may delete, especially for the sake of clarity or starting anew.  She is building a system of ideas, and wants to see the overall picture, and how an individual idea relates to that picture. Emotions arise from the difference between how the thought relates with the whole and how she manages to express this by means of relations with other thoughts.  She writes down only a portion of the thoughts that she thinks, organizes only a fraction of what she writes, edits and polishes only part of what she organizes.  The thinker therefore places highest priority on entering thoughts, also high priority on arranging and reorganizing thoughts, and has very little interest in formatting thoughts.  Her system of thoughts is like a closet or garage that is meant to be semi-messy or even super-messy. Even so, all of the thoughts are tending towards a total system.

The anthropological criterion has us consider the ways that people have been organizing thoughts over the centuries.  A survey of these ways revealed that sequences, hierarchies, and networks are the basic external structurings of thoughts, and that in addition to them, table structure is a basic internal structuring of thoughts [7].  Surprisingly, in human practice sequences, hierarchies, and networks do not occur in isolation, but rather in pairs.  What happens is that one of these three external structurings gets used robustly, gets overwhelmed, and gets patched with another of the external structurings.  For example, when a scribe lists historical events, then as soon as the sequence is of any interest, she introduces a hierarchy of eras and ages to get things back under control, yielding a chronicle.  The different ways of restructuring yield a taxonomy with six visualizations, as described in Table 1.  It should be emphasized that this is not a taxonomy of diagrams, but rather of the mental pictures associated with the use of diagrams of thoughts.
 
 
Visualization ( S = sequence, H = hierarchy, N = network )  Examples Psychological Effects Structural Violations
S to H, Chronicle  Adds a hierarchy to a sequence. Chronological events get organized in eras and ages.  Table of contents, photo albums, infantry, decimal notation for whole numbers, indentation of code.  An idea is weak when it needs to be grouped with its neighbors (as in a magazine).  Thoughts in a hierarchy can be labeled by subsequences.
S to N, Canon Adds a network onto a sequence, as when we have reuse.  Index of book, math proof, factory line, concordance, Scripture, sequence diagrams.  An idea is strong when it leaps out away from its neighbors (as in a Scripture).  Thoughts in a network can be labeled by subsequences.
H to S, Evolution  Maps a sequence onto a  hierarchy. Can show trees of possibilities change over time.  Origin of species, recipe book, decimal notation for real numbers, knockout tournament, genealogy, chess openings, activity diagrams.  An idea is broad when it synchronizes variations (such as an Age of Dinosaurs).  Different contents can have the same position in a sequence.
H to N, Catalog Grafts references onto a hierarchy.  File manager, auto parts store, pantheon, codex, academia, thesaurus, bureaucracy, flow charts, design by contract.  An idea is narrow when it links from one part of a tree to another (as in a bureaucracy).  Different contents can have the same position in a network.
N to S, Tour Considers all possible walks within a network.  Correspondence, Markov process, conversation, state diagrams, collaboration diagrams, scenarios.  An idea is clear when we pass through it many times (as in a Socratic dialogue).  The same content can have different positions in a sequence.
N to H, Atlas Adds a hierarchy onto a network. This makes for global and local views. Anatomy, web portal, political map, ecosystem, social network, package diagrams, CRC cards, extending use cases.  An idea is vague when it gets lost within a cluster (as does a small country).  The same content can have different positions in a hierarchy.

Table 1: Taxonomy of Visualizations

The psychological criterion looks for an explanation of how to isolate and heighten the psychological effects the thinker experiences when using sequences, hierarchies, and networks.  In a crude sense, writing in a sequence encourages the strong ideas to come first, and the weak ideas to come afterwards.  Writing in a hierarchy encourages the placement of broad ideas closer to the root, and narrow ideas farther from the root.  As for writing in a network, my experience is that it helps one write in a private language where vagueries can be left to footnotes that will get written later.  Vague ideas will have many links, and clear ideas need not have any.  A strong, broad, and clear idea will be at the head of the sequence, at the root of the tree, and without any links, which is to say, it will stand independent of all other thoughts.

The philosophical criterion has us prefer definitions of sequences, hierarchies, and networks that accord with models explaining, relating, and clarifying the evidence provided by the functional, anthropological, and psychological criteria.  Introspection suggests that as a sequence, hierarchy, or network grows, we do not visualize it, but instead visualize its restructuring.  For example, as a sequence grows, our mind either starts to group elements (constructing a chronicle), or hop back and forth among elements (generating a canon).  The mind faces a dilemma of which way to restructure, and this dilemma seems to be what encourages us to distinguish between the strong ideas and the weak ideas.  In this case, when the mind groups elements, it treats them as weak, but when the mind hops from element to element, then it treats them as strong.  Similar observations are made in Table 1.  This model is important in as much as it successfully explains and relates evidence from the anthropological criterion and the psychological criterion.  It is particularly useful in that it makes predictions that may be tested, both psychologically and anthropologically.

4. Applying Criteria to Answer Questions

I have yet to find definitive answers to the structural questions posed in Section 2.  However, I can report some important insights gained from applying the four criteria.

The psychological evidence, which deals with structuring (and arranging), appears to be at odds with the anthropological evidence, which deals with restructurings (and visualizing).  For example, from the psychological point of view, two thoughts should never be assigned the same position in a sequence, because doing so evades the effects of arranging ideas in a sequence, and defeats the purpose.  However, from the anthropological point of view, there is a visualization, evolution, the whole idea of which is to have several concurrent thoughts.

Such head on conflicts suggest that each visualization, which is to say, each restructuring, introduces a structural degeneracy, creating a structural violation, as listed in Table 1.  Note that the degeneracies occur in the secondary structuring, for example, in the case of the evolution it is not the hierarchy (primary structure), but the sequence (secondary structure), which is degenerate (different thoughts may have the same position with the sequence).

Note also the symmetries in the violations.  Of course, the table phrases the violations so as to heighten these symmetries, but it is noteworthy that in each case the symmetry turns on the primary structuring.  For example, in the case of the evolution, the primary structure is a hierarchy.  If instead the hierarchy (primary structure) is restructured with a network (secondary structure), yielding a catalog, then the violation is expressed in terms of the network (different thoughts may have the same position in the network).  Specifically, a catalog may have many types of unusual links, such as a bidirectional link from A to B, which may be understood as a link from A to B and link from B to A having the same position, so they are treated as one.

The symmetries in the violations are tremendous help in figuring out the original structurings (which are difficult to arrive at, because they cannot be visualized!)  Working backwards it is possible to propose that sequences, hierarchies, and networks are originally such that:

5. The Need for Several Formats

Acknowledging the conflicts generated by the evidence, I envision three separate formats for unstructured thoughts (emphasizing the issue of punctuation), structured thoughts (the issue of manipulation), and restructured thoughts (the issue of visualization).

Unstructured Thoughts

The format for unstructured thoughts addresses Why are we arranging?  It is supported by the functional criteria, together with the philosophical criteria. No relations between thoughts need to be imported or exported, because it is up to the arranger to arrange them as she wishes, and that is given in her own mind.  The emphasis should be placed on determining, what is a thought?  In Microsoft Word, for example, does a thought equal a paragraph?  Separate fields (such as labels, dates, addresses, Ids) should in general not be exported as thoughts, but rather incorporated as part of the internal structure of a thought.

Structured Thoughts

The format for structured thoughts addresses How are we arranging?  It is determined by the psychological criteria, together with the philosophical criteria.  It describes the structural environment for creating arrangements, which is to say, for creating relations, for acting on thoughts, and for being psychologically affected by these actions.

Here the three structures reveal themselves at their most basic way.  Presumably they all satisfy three basic conditions: Different contents may not have the same position, different positions may not have the same content, and thoughts may not refer to each other (in particular, no distinguishing labels, and no subsequences).  I imagine that there is only one hierarchy, only one sequence, and only one network.  The positions in the hierarchy are not ordered, nor are they labeled.  The sequence does not allow access to any absolute order.  There may be directed links from one thought to another, but there are no unannotated, nondirectional, bidirectional, multiple, or bundled links, nor are there links from a thought to itself.

Restructured Thoughts

The format for restructured thoughts addresses What are we arranging?  It is determined by the anthropological criteria, together with the philosophical criteria.  It describes the structural environment for porting visualizations, for stepping back and viewing the structure.  I imagine that:
 


In summary, I imagine that in this format, position constraints (of the hierarchy and of the sequence) are subject to content constraints (of the network).  The hierarchy and the network can use sections of the sequence, which is the origin of labelling.  There is an algebra of structural collapse, both in the network and in the sequence.

Note that even the format for restructured thoughts prohibits many structural possibilities, allowing only those that may be visualized.  There are all sorts of phenomenon involving multiple restructuring, but I claim we cannot visualize them. Consider an XML document, which is founded on sequential processing (all data is processed in a single pass through), is organized hierarchically, and lastly, supports links.  Please try to visually absorb all three layers, and then I think you will agree that we are forced to choose whether to visualize an XML document as a chronicle (sequence restructured with hierarchy) or catalog (hierarchy restructured with network) or canon (sequence restructured with network).

This paper presents only a preliminary vision, but this is enough to express the need for separate formats, and show how they might be related.  The kind of argumentation employed may yield a preliminary format that can be tested by more traditional means.

6. Conclusion

Software for organizing thoughts has existed for more than twenty years [8].  Market forces have not yielded a standard for the import and export of aggregates of thoughts.  The transfer of irrelevant structural information can distort and cloud thinking. Functional, anthropological, psychological, and philosophical criteria can be applied to define sequences, hierarchies, and networks so that they accord with their use in working with thoughts.  Separate formats should exist for handling unstructured thoughts (emphasizing punctuation), structured thoughts (emphasizing manipulation), and restructured thoughts (emphasizing visualization).

A standard could work with existing software as it could be implemented through converters, perhaps using XML.  Well designed formats, verified and corrected through use, could become accepted as a de facto standard.  It would make possible the development of thinker toolkits, and encourage the design of user interfaces based on organizing thoughts instead of documents.  Thinking management - knowledge management by the individual - could lay a sound foundation for knowledge management within the community.

Acknowledgements

Some of the ideas in this paper, especially regarding emotion, came from praying to God.  I also benefited from debate with Saulius Maskeliunas on how to answer the structural questions.

Bibliography

[1] Kestas Augutis,  Effective use of a computer at school, first place entry for the contest. Open Society Fund - Lithuania, Vilnius, 1997. (in Lithuanian).
[2] Peter McBrien and Alexandra Poulovassilis, A Uniform Approach to Inter-model Transformations, Proceedings of the 11th Conference on Advanced Information Systems Engineering, (Heidelberg, Germany) Springer, June 1999, pp. 333-348.
[3] Rick Kazman and Jeromy Carrière, An Adaptable Software Architecture for Rapidly Creating Information Visualizations, Proceedings of Graphics Interface '96, (Toronto, ON), May 1996, pp. 17-27.
[4] Natrificial Software Technologies,  The Brain, 1999. http://www.thebrain.com
[5] Andrius Kulikauskas,  Will computers help us concentrate ?  InfoBalt Laikas, Nr. 3, InfoBalt Association, Vilnius, Oct. 1998, pp. 38-40. (in Lithuanian).
[6] Frank Manola of Object Services and Consulting, Inc., Towards a Web Object Model, February 10, 1998, http://www.objs.com/OSA/wom.htm
[7]Andrius Kulikauskas and Saulius Maskeliunas, Organizing Thoughts into Sequences, Hierarchies, and Networks, Proceedings of EMMSAD’99, (Heidelberg, Germany), June 1999.
[8] Dennis Shasha, When does non-linear text help?, Proceedings from the First International Conference on Expert Database Systems, Larry Kerschberg ed., April, 1986