ThoughtMesh is an argument for changing the way scholars publish their research. ThoughtMesh is also a tool. If you're one of those people who learn more from doing than from reading, then check out this tool right now. (Go ahead, while no one is looking.)
Otherwise read on to find out what I think is wrong with existing publication mechanisms, and why I think ThoughtMesh is a step in the right direction.
It's no secret that today's academics are having trouble keeping up with networked media. The currency of academia remains the peer-reviewed print journal--not exactly the ideal medium for intellectual discourse in the fast-paced age of the Internet. The archaic criteria by which most universities award promotion and tenure mean that even academics who specialize in digital culture find it hard to justify writing about it in a digital vernacular. But if scholars don't want to drift ever further out of touch with the information Twittering and Flickring across the world's browsers and cell phones, they'd better find a way to tap into and redirect these information flows.
Unfortunately, it's one thing for the American Council of Learned Societies and Modern Language Association to urge academics to explore digital scholarship, and another for academics to get it right. Most professors stop at PowerPoint, which is a bit like aiming for the Olympics and stopping at Pee Wee baseball. Of course, whenever I rant in public about PowerPoint's unconnected, unnavigable, and profoundly anti-social interface, some 50-something guy in a tweed jacket invariably asks me what to use in its stead. Usually I recommend learning some HTML or JavaScript, at which point Mr. Professorial's eyes glaze over and I realize I've shifted into a language he doesn't speak.
Sure, there are off-the-shelf HTML cookiecutters like FrontPage and Dreamweaver, though they can be a little intimidating to a writer raised on an IBM Selectric. Blogs are easier and offer nifty networking tools like syndication and trackback, but their diaristic format doesn't lend itself to lengthy analyses. If you're really lucky, your department will hire an ace Web developer to HTMLify your research--but even then your pretty online essays are likely to end up marooned on the island of your university's Web site, semantically isolated from the rest of the Web.
In fact, none of these formats makes it easy to find online scholarship. What are the chances that the Tufts professor you met at a conference on Postimpressionism is going to stumble upon that kick-ass Cézanne paper on your Web site? Googling the phrase "nude bathers" certainly isn't going to help. When forced to compete with porn, paparazzi, and politics, the average scholarly Web page falls too low on the food chain of commercial search engines to be discovered by another researcher. And with the Semantic Web at least five years off--with many saying it will never get here at all--sleuthing out networked scholarship means shadowing a dubious trail of desultory hyperlinks or piecing through disconnected search returns from siloed databases.
Of course, Vectors projects offer scholars an ideal solution. Who wouldn't want their paper semantically tethered to other Vectors essays, with an interface tailormade to the subject matter by a dream team of Flash designers? Sadly, Vectors Fellowships are available only to a half dozen Fellowship winners per year. So what's a new media scholar to do?
When Craig Dietrich and I set out to build ThoughtMesh, we asked ourselves how an ideal publishing tool for scholars would behave. We decided that we wanted a system that was distributed--not siloed away in a single database, but able to be published on any Web site anywhere. We also wanted all the essays to be connected to each other, by something less random than search returns, but more serendipitous than intentional hyperlinks.
To accomplish these ends, we built ThoughtMesh to:
* Allow navigation by tags as well as essay sections.
Tag clouds are new organizational structures emerging in today's distributed publication communities, most famously in popular social networking sites like del.icio.us, Technorati, and Flickr. In a typical tag cloud, clickable words corresponding to user-defined categories mill or float about on a page, their position and prominence determined by an emergent count of the number of times they have been used rather than by some top-down authorial decision. Clouds allow for overlapping, not dichotomous categories. They visualize relevance as a swarming or bubbling rather than a roll-call or rank.
* Allow dynamic re-organization.
Clicking on a tag should reveal a substructure of relevant points--a particular slice through the essay--without leaving the essay behind. And readers should be able to choose topics from a single tag (Cézanne) or from combinations of them (Cézanne + nude + bathers).
* Pull in related texts dynamically.
Thanks to John Bell's Telamon software, within the essay interface readers should be able to skim a stream of similarly tagged articles created with the same tool, whether they are hosted on Vectors or Geocities. Unlike a typical database search, the tool should fetch thematically related excerpts or articles without requiring a page refresh.
* Let authors choose automatic or manual transmission.
An ideal system would let fastidious authors set their own tags for each section, while authors pressed for time could let ThoughtMesh auto-generate their tags based on the most frequently used words in the essay.
* Encourage live conversation.
A presenter using this tool should be able to skip around from one topic to another outside of a pre-ordained order, so as to respond better to live conversation and feedback from others in the room. (Want to see a PowerPoint presenter squirm? Ask her to go back to a slide in the middle of the stack.) And make the tool browser-based, so if anyone asks a question the presenter doesn't know, she can Wikipedia it in another tab!
* Be easy to use.
The must successful Web 2.0 applications (Flickr, del.icio.us, Basecamp) have clean, easy to use interfaces. Inside of fifteen minutes, authors should be able to copy-paste a bunch of paragraphs and watch a navigable cloud of tags emerge.
* Operate in standalone or networked fashion.
One week you're presenting on a laptop in an igloo in Nunavut; the next you're at MIT, where even the bathrooms have WiFi. Either way, the tool should let you strut your stuff--it should be easily networked but not require a network connection.
* Be easily shared with others.
The tool should be lightweight to download and easy to host on your own domain. For maximum reusability, the application should be open source and ideally able to run without a compiler.
Craig and I have designed ThoughtMesh to accommodate as many of these features as possible, but we have more goals for future releases. In approximate order of completion, these include:
* Mesh more essays.
ThoughtMesh needs a critical mass of essays for its tagging system to generate results; it's nice to see all sections of the current essay tagged with Cézanne, but nicer to see some other essays in the mesh tagged with that term. Fortunately, the generous folks at the Vectors program have agreed to seed the mesh with the articles from past issues of Vectors, which will provide a wide-ranging initial pool of insightful and polished essays.
* Enable multiple meshes.
Adding hundreds of essays to ThoughtMesh could introduce noise in the signal--how do you know that article tagged with Cézanne isn't really a Viagra ad?--but Craig and I have a couple plans for offsetting this tendency. One solution is to enable community-based "meshes" that authors can create and control. So if you're in charge of the "Victorian Literary Criticism" mesh, you can decide what's in and what's out. Each reader can then choose to view 1) only related sections from a particular mesh/es, or 2) related sections from the entire system (all meshes).
* Article rating and smart sorting
Article ratings offer another way to boost the signal-to-noise ratio. Fortunately, Still Water Research Fellow John Bell has been at work on a prototype for a distributed commentary and evaluation system, called RePoste, that would dovetail perfectly with ThoughtMesh. In a future release, a RePoste-driven mixing agent might also reconfigure the position of each essay excerpt returned by the mesh. This would prevent the circular "rich get richer" bias of many interfaces, whereby the most-frequently cited essay is most read because it is the most frequently cited.
* Single-field essay input.
To make it even easier to mesh an essay, an alternative version of the input page might allow authors to paste an entire, unseparated text and add wiki-like markup instead of having to cut and paste text sections separately.
* Mesh tracing.
A longer-term goal is to offer readers an interactive map showing how tags connect articles in a given mesh. This is more than eye candy; as I've argued elsewhere, clouds of influence are a much more nuanced way to recognize achievement in academic networks than journal "impact factors." Giving an essay's connections visual form could help evaluators understand the Big Picture of that author's influence within an academic subculture and the broader online community.
Craig and I are always looking for ways to improve ThoughtMesh, so please email us at TK if you have any questions or suggestions. In the end, we hope ThoughtMesh's paradigm of distributed connection will prevail as much by enactment as by argument. That, after all, is the new media way.