sardonick

Giving a talk at NCCU while I am in Taiwan this summer, entitled Creating a Virtual Language Tutor that Understands Learner Speech . Not the most exciting of titles, I know. Feel free to ask me if you are interested in attending. Here is the abstract.

The recent combination of getting a lot of pseudo-conversational-type email spam, and getting to know an online friend whom I’ve never met in real life has triggered an interesting thought:

When (if) we solve the AI-hard thing, when we’re able to make virtual agents that can carry on conversations and social interactions just like real humans, what’s to stop advertisers from creating armies of Turing-Test Passable AIs whose sole purpose is to trick humans into relationships so that, once trust is garnered, pimp whatever products the companies are selling?

I can imagine such an apocalyptic nightmare in vivid detail…

It all depends on which set of glasses you use to look at reality, I guess…

Another entry from McSweeny’s lists:

Three Items at the Supermarket I See in a New Way, Now That I’ve Read Some Books on Literary Theory.

This is brilliant.

// TODO: insert witty and sublime metaphor about Supermarkets and Life here

Backchannel was good, and backchannel was fun, but was I the better or worse for having participated in it? I usually take personal notes at these types of conferences, and with IRC up the whole time the backchannel became my offboard note-taking file. This is good because my notes got to synergize with other peoples’, but it’s bad because I didn’t get a chance to write any tangential thoughts that wouldn’t be pertinent to the rest of the group (self-consciousness of the group’s net signal:noise ratio, call it social temperance) but that would be useful to me. Had I not had the backchannel and had my personal notes instead, I would have written the contents of this huge blog entry in some text file instead of here, after the fact. And I probably would have remembered many more ideas than are listed in my braindump.

More to follow…

The waves that tower over us
Betray tomorrow

Tonight marks the end of an entirely filled weekend, and really all I want to do is soak my brain in a big bucket of ice. Much more intellectual stimulation than I usually get in such a short amount of time. The cause of all this was a conference workshop on Social Software in the Academy. Dynamic people, personalities, theories… It’s at this point of mental exhaustion and computer-use-oversatiatedness that I want to turn off electronic screens and curl up in a comfy chair by a fire and read poetry alone… but I also know that, in reminiscing, it’s always these points that I most regret not capturing and summing up.

So, a short flow of consciousness stream of thoughts and subjects, to capture what I can and hopefully prompt more organized reflection later:

• social software turning the world into Small Town, where grassroots reputation matters again, more than big top-down messages the media distributes
• social software as a watershed technology that changes academics in a way that is COMPLETELY different than any technology-wrought change that even the tenured-generation professors have seen (this includes video, etc)
• and what makes this difference is the empowerment of a wide range (intergenerational, different walks of life, even, dare we say, outside of academia) of people
• questions about the future of academia because of this changing: will it shift and adapt and adopt, or will it become a hollow shell that has lost its prior authority and societal function?
• current academic reputation is maintained on a “publish or perish” meritocracy, with publishing filtered by a longstanding tradition of peer review in controlled journals. blogging allows the everyday man (or the everyday professor) to be a wide-reaching mouthpiece, allows the possibility of peer-review-by-hoi-polloi-many instead of peer-reveiw-by-few-on-the-paper-acceptance-committee
• change in paradigm as we relate to information: before it was the scent of information we would try to follow, now it’s the stench of information we’re overwhelmed with
• psychological patterns honed in the old scarcity of information world that lead to harmful/self-destructive patterns in the new world of information (book-hoarding, infornography, obsessive bookmarking, …)
• future of intellectual property in a world where content and meta-content is easy to create (do people always have the right to create metacontent? or does it ever infringe on the “rights” of the primary content creators? and should those primary content creators even have exclusive rights to ideas?)
• when content is housed by universities (e.g. campus-wide blogs) does the student or the university own the student-generated intellectual property
• what role will new social technology play in the teaching world (when are blogs/wikis the right tool for the job, when are they not?
• when professors can establish authority by means of blogs instead of the traditional paper publishing machine, what happens when competent authorities emerge who are outside academia (e.g. Thomas Vander Wal), how do they fit in to the picture
• how are those extra-academic authorities dealt with in attributions and citations?
• how are blogs and wikis (changing, maleable, created by MASSIVELY large groups or created by anonymous groups) to be attributed and cited in the academic publishing machine?
• and lastly (perhaps most importantly?) when will we get RSS readers that don’t suck?

more on any (all? none?) of these later. each sentence in here deserves at least a half day of mulling over. This will keep me busy with thought-fodder (heh like i’m ever at a loss for that) for…hmmm… half of the summer? that’s conservative.

p.s. Micro-bounty-note-to-self: I owe Leonard a dinner at Chanos if he ever codes the personal lifetime web cache that never forgets the content of a site you visit. This will be a powerful punch combo when paired with ubiquitous desktop searching (beagle, spotlight, google desktop search, msn desktop search).

p.p.s. Note-to-self-2: continue getting to know the other USC Social Software Researcher Fellows (Richard, Daniel, Aram, David, et. al)–they are both nearby and awesomely brimming with neat ideas.

p.p.p.s. More thought fodder

????????,???
??????June 19?July 5?????
???????????,??email????????????????

??????????blog??????????????????????????, ???????????????????

(p.s. sorry if my Chinese is bad… I have forgotten too much since I came back to the U.S.!)

As of Monday evening, I finished the last final exam I’ll most likely ever take in my life. Wow.

It really was a whimper-type rather than a bang-type ending. The class was Algorithms, the one that I’d put off taking for all of my academic career ’til now. Not because it’s not an interesting subject (dynamic programming is pretty awesome, powerful stuff if taught right), but because it’s rarely well taught. The particular class I took was taught by an instructor who, while being a good lecturer, didn’t spent enough time at all in preparation. Book, problem sets, lectures, and tests were all completely disjoint from one another–if I could hazard a guess as to class organization, the book was chosen because it was a famous one in its field, the instructor’s notes were taken from a prior class he had himself attended, problem sets were plagiarized from algorithms courses in other universities, and test questions were written by the TAs. Not exactly a cohesive learning environment

Preparing for the final this weekend was like pulling teeth. ~20 hours spent pretending to study, ~4 hours actually studying.

Really, this class has shown me the distinction between masters-level and doctoral-level classes. Honestly, and unfortunately, it seems like most of the classes in graduate-level computer science here at USC are built to cater to our Masters-Degree-Diploma Mill. Granted, this diploma mill does much to fund PhD students like me, but… the resulting classes are so large-sized, so banal in content, that it feels more like an extension of undergraduate coursework. There are exceptions to this, of course. Arbib teaches an amazing class on Human-Brain-Inspired AI, and Knight/Marcu/Hovy teach a set of mean NLP courses, but these are exceptions rather than standards.
</rant>.

The important thing is it’s over now.

Now that course work for my PhD is done, all that remains is a year or two more of research. And then I get to call myself “Doctor Nick”, and feel happy about myself because I have a paper diploma with my name on it, and face the vast unknown that is the future =).

Getting married in December, and graduating a year or so after that, that’s the plan. Then it’s hopefully a bit of post-doc work, and then emigration. Honestly, my dream someday is to be a professor in some country/area that doesn’t have as much academic infrastructure/resources compared to the U.S. Call it holistic tentmaking. Or reverse brain-drain. Or something, I dunno yet. We’ll take it one step at a time, for now.

A few weeks ago, Sarah Lohnes put out a call for lunchtime discussion topics for the upcoming Social Software in the Academy Workshop.

Well, the more controversial and future-looking the better, I thought, so I suggested the following:

What will happen to the Ivory Tower as social software makes advanced research more accessible to the armchair academian?

Well, as of a few days ago, I got to taste the repercussions of this firsthand.

As a Doctoral Candidate student, I’m not exactly an “armchair academian”–at least not in my areas of focus (Linguistics/Second Language Acquisition, Artificial Intelligence/Natural Language Processing). When it comes to Information Architecture, though, I’m as novice and {self-,half-}educated as they they come.

In his blog, Jess McMullin takes a few lines to beat an earlier post of mine to a bloody pulp saying

“The New School of Ontologies is just so off the mark… The article reads like a jumble of classification buzzwords stirred once and regurgitated.”

Ouch!

His complaints about things being “off the mark” seem to be more directed toward folksonomy in general. And the “jumble of classification buzzwords” bit?

By nature, the Ivory Tower has always been exclusive, and therefore filtered. The good part about this filtering is that it filters by quality. PhD applications/screening/quals, submissions to papers, conferences (well, most conferences), submissions to journals… these are all good quality filters. But there’s also a filter on content, too. Anyone involved in upper academia has seen the ebb and flow of “acceptible” (read “fundable”) theories. Research in neural networks was dead for decades because some of the fathers of AI didn’t think it was feasable. And then there’s the Cult of Chomsky, and its strangelhold on Linguistics–just try giving a talk at USC that doesn’t agree with the tenets Generative Grammar, X-Bar Theory, or UG. On the remote possibility that they even accept you to speak, asbestos underwear would be a must for the after-talk discussion.

As social software puts windows (phone lines? cracks? what’s the metaphor I’m looking for here?) into the ivory tower, it’s going to break down this exclusivity. Whether this is benefit or detriment I can’t say for sure. On the downside, it definitely increases the noise:signal ratio. And it empowers people like me who will then {pontificate,spout off} without full knowledge of either the jingo or the theory. This can, of course, be annoying to the educated reader. But on the upside it encourages cross-pollination of ideas, and exposes the idea-filtered elite to new (or out-of-vogue) ways of looking at the same old problems.

This isn’t anything new. We’ve seen it happen over and again, fifty million different ways, every time new technology is introduced. Blogs vs mainstream media sources. Home video cameras, VHS, and cassette tapes allowing amateur distribution of music/video. Copy-machine zines. But this might be the first time the Ivory Tower has felt the crumbling

Note: This wasn’t the place to address McCullin’s criticism of my folksonomy paper. Perhaps in a later post…

Stefano’s Linotype has a good essay on emergent folksonomy, especially how it applies to del.icio.us and the different-people-use-the-same-word-to-mean-different-things problem.

His solution to colliding semantics is to augment syntax to document things.

This is good, but it overlooks the fact that I myself might use the same tag to refer to different things.