My committee members have asked me to write up my research question.

The difficulty I have with expressing my research is deeply related to the latency I have permitted myself with describing my practice.

For me, research is synonymous with learning. In the social sciences and arts, 'discovery' is not as it is in basic science, or even applied science.

For a scientist, whose 'method' is nailed down by 100+ years of reified practice, research is experimental design, execution, and reporting.

For a social scientist who is trying to play the part, method is less adulated, venerated, and gilded calflike. It is various, not homogenous. It is a big question-mark.

If naturalism, empiricism, scientifism are not the basis of your method, what approach offers an adequate foundation?

Long ago I began exploring phenomenology and more recently have been trying to understand how it can be used as method.

To date, my research/research-practice/practice has been exploring my own experience, actions, and motivations. Becoming aware, cultivating 'mindfulness', and developing capacity for meta-cognitive 'observation' or journalling all have a part in my thinking and practice to date.

Yet it is still hard for me to learn new habits and integrate them into my daily routines. I find systematic records of practice particularly onerous to maintain. Cultivating the practice of observing & recording doing is not nearly as much fun as discovering, using, and refining tools, processes, and procedures. I enjoy doing more than recording.

But at the end of my multi-month trajectory of exploration, I have little to show, save for a few more bookmarks, eArtilcles downloaded, or other files here & there. The narrative is lost to the new moment that flows along, flushing all traces of old discoveries from recent memory.

The learning and discovery, of which tools to use at what junctions in a given workflow; the tradeoffs, strengths and other comparative merits of various tools 'contending' in a given class; the insight and knowlege gained by the practice remains tacit, unsaid.

In my view, this is okay. I have several reasons for thinking so. But articulating them is difficult, thanks mostly to a recent moment of enlightenment: I am defined by what I am.

Though it makes both 'the big picture' and 'the end game' a bit more difficult to apprehend for both my prospective committee members and myself, my current projects fit more in the foundational, infra-structure building department than in some scientific discovery department. It is consequently hard to articulate beyond these massively intertwined projects.

I have tried to define my 'space,' 'role,' & 'identity' in others' terms for too long. In most cases, I have found it seems easier to start with what one is not. But this mostly just leaves my listeners feeling like the one thing they understand clearly is that my work is not what they do. Since it has not proved that effective a communication strategy, I have foresworn this lazy approach to self description.

So I will begin with what I am.

I am an experimental philosopher. I am an end-user developer. I am a transdisciplinary analyst. I am a workflow designer. I am a cognitive scientist. I am a discourse linguist.

So my research questions will present windows of inquiry into each of these domains.

It makes sense that the conceptual, technical, and technological infrastructure required to develop and execute a research program with such transdisciplinary scope and objectives should come first. Since organizing research resources is a big part of research, having (or in my case developing) the dedicated environment, calibrated toolset, and practiced knowhow for the core tasks necessary for a doctoral dissertation (or any other knowledge-intensive analytic role) is a good strategy.

So I have begun with an operational problem: Analysts spend 'most of their time' (some even claim over 90-95% of their time) getting data ready to analyze: finding, collecting, cleaning, converting, massaging, and otherwise monkeying around with it to get it into the analytic environment and ready to analyze. I would add from personal experience, much of this time is spent in frustrating, apparently sisyphatic effort; straining against system's inbuilt limitations.

So, given that an analysts' computing environment (system) changes so extensively and regularly (e.g. Complete OS changes, Installation of new Applications, Updates to existing OS & Applications), what are some first steps towards solving the 'data ingest' problem?

My answer to this research question range over 3 or 4 related research/practice projects.

Can we streamline the process of collecting, preparing, (and even predigesting) document data, to enable analysts to focus on analysis rather than research & file management? Can we do so in a way that does not require analysts to become OS gurus? Ultimately, can we reduce the cost of data ingest transformations to analysts?

My 'Mindful Media Management' and 'FlowSpaces' projects kind of bleed into my 'MetaData³ (munging, mananging, & mobilizing) project...

Can we capture and extract the content and formatting style of original data without sacrificing or damaging or altering original data?

For data-provenancing purposes; Can we automatically maintain connections between the abstraction and original?

For data annotation and markup purposes; Can we collect and layer (overlay, or otherwise connect) external data and metadata onto documents, in a coherent system that respects both data provenance and original presentation?

Some technical questions that must be answered along the way, which may offer a contribution to the UI and software engineering domains: Is an external, proprietary database of metadata preferred to file-embedded metadata, or to file-system-attached metadata?

• What is the best way to facilitate the extraction and updating of internally embedded (file-embedded) metadata at the level of the system or whole collection? (Yes! e.g. PDF Aerialist, DSOle enabled VB Macros)
• Can we facilitate the extraction and updating of file-system-attached metadata? (Spotlight Comments e.g. Leap, Yep, TagBot, Punakea; OpenMeta e.g. Tags, Tagit, Tag Folders)

For interoperability purposes: Can we make proprietary metadata database applications draw from embedded and attached metadata? (e.g. Leap, Yep, Papers, Sente) And vice versa:* Can we make attached metadata methods draw from and edit both embedded and proprietary metadata stores? (e.g. Tags, Tag Folders, Leap 2 Beta)

For automated meta-data retrieval purposes: Can we leverage existing meta-data stores that are both structured (e.g. ISBN data, Library of Congress Data, SFU Library MARC data, MedLine keyword data, some Journal-assigned keywords, CDDB data, Amazon product data, Whitepages data, OpenGIS data etc.) and unstructured or loosely structured (e.g. some Article Author-assigned keyword data, Individual computer-users' idiosyncratic and organically evolving directory structure, Folksonomy tags, Individual users' tags, Google Scholar RIS data, etc.) so as to automatically layer metadata onto documents?

Can this metadata then be used to autofile documents? (e.g. iTunes, iPhoto, Yep, Hazel, MetaDataMover Automator Workflow)

My committee members have asked me to write up my research question.

Unfortunately for us all, my projects right now fit more in the foundational infra-structure building department than the scientific discovery 'Im a genius' department. Oh well, right?

So what I have is an operational problem: Analysts spend 'most of their time' (some even claim over 90-95% of their time) getting data ready to analyze: finding, collecting, cleaning, converting, massaging, and otherwise monkeying around with it to get it into the analytic environment and ready to analyze. I would add, much of this time is spent in sisyphatic effort, straining against system's inbuilt limitations.

So, given that an analysts' computing environment (system) changes so extensively and regularly (e.g. Complete OS changes, Installation of new Applications, Updates to existing OS & Applications), can we determine first steps towards solving the 'data ingest' problem?