Archive for February, 2011

poor man’s bandwidth control

Consumer broadband modems and routers typically cannot deal with a large number of connections from multiple users, and the cheap firmware has no settings to restrict the bandwidth of each user. This is when you resort to the physical layer to help you out.

Solution: Restrict wifi to the really slow 802.11b, maybe even slower, and let the physical layer radio contention fake the effects of fair bandwidth control. Works like a charm, no more dropped connections.

watson v. mit

So being at the event captured in the image, I got to ask a question toward the end. Actually I asked two questions. The first was whether Watson would ring in and use the remaining 3 seconds or whatever to continue to compute. Gondek said it would if it helped. In actual competition it doesn’t appear to be the case, as the buzz-in thresholding condition ensured that further computation would not have been helpful. The second question was a follow-up on the identified weakness of Watson — learning on “common sense” knowledge. I asked what path AI research would take to tackle such knowledge, which are by its very definition, “not in the books.” Gondek said that IBM is building up semantic information (e.g. a “report” is something that can be “turned in” and “assessed,” etc.) from corpus. That wasn’t exactly what I was asking, however.

My point was whether all “knowledge” is written down. There is such a thing as experiential “knowledge,” and humans take years to learn it/be trained in it through parenting (i.e., to “mature”). If only there were a handbook on life, or that life could be learned through reading a series of textbooks, then perhaps I’d believe that the kind of general-purpose AI that most people are probably imagining (rather than an expert/Q&A system) can be achieved along the lines of current methods.

nchoosetwo and collaborative ranking

Walking around campus these days, there are cryptic-looking things like

\(\binom{n}{2}\mathrm{.com}\) and \(\binom{n}{2} \ni \binom{i}{u}\)

obviously referring to a dating site — currently it’s restricted to MIT and Harvard students. This one tries on an idea that I’ve heard discussed numerous times in different contexts, but apparently nobody went and did it in all these years. Instead of running a matching algorithm, it asks third parties (i.e. matchmakers) as well as the interested parties themselves to suggest matches. The thing that is supposed to keep this low-risk is anonymity: a match isn’t revealed until the two primary parties involved mutually accept or their lists intersect.

As with all things that involve anonymity, this asks for trollish and antisocial behavior. I’ve already registered three aliases on moira for exactly this purpose — ok, ok, so they’ve suppressed that antic after people raised concerns, though these and other ramifications should have perhaps been worked through a bit more carefully pre-launch.

The spam potential remains. A matchmaker’s identity isn’t revealed unless both people accept her suggestion, so pranks and insults can be conducted to an extent. One way around this may be grafting social graph data onto the system for collaborative filtering (if they manage to obtain such data…). And if they do, perhaps the suggestions of more closely related people should weigh more, along with those of successful matchmakers. Perhaps there should even be more weight if multiple matchmakers concur. This is extremely intriguing, because eliminating spam is equivalent to predicting who is a likely match, and collaborative filtering for this problem is an unexplored direction.

The more fundamental question is why such a site is even necessary.
(Read the article)