Ubicomp at UW

This is a lab that I am very psyched about.  Here's their website.  I was lucky enough to see a bunch of these guys give talks the other day on campus.

I feel like they've got 2 main focuses: house-level stuff, and person-level stuff.  House-level stuff is really mind-blowing: you put a single sensor by your main electrical box, and it tells you which device you're using.  Or your gas meter.  Or your water meter.  And shoot, you've got water rushing through here, why not use it to power the measuring device?  Or maybe you want to put out a bunch of sensors for some task, but you want them to be wireless, but it takes a lot of power to transmit to the central base station.  Why not transmit to the nearest wall instead, and let the power lines in your home take it to the base station?  Whaaaa?  This is some kind of magic.

Slightly less magical, but more interesting to me in its applications, are their personal things.  What if you wanted to get some info from your phone, but your eyes and ears were busy?  How about squeezing it?  There's a surprising amount of info you can get from that.  Or maybe something that generates its own power?  Or, okay, we have a powered computer in our pockets... maybe you forgot something you just heard a few minutes ago.  It'd be nice to press rewind on your auditory life, wouldn't it?

What if a human could wear something like this all the time?  What could you do then?

It's all low-power and low-setup-cost, and it all hacks the physical world in cool ways.  Getting information that you wouldn't think you could get.  Allowing more sensors to be deployed more easily.  Maybe with personal health applications.  Maybe with personal mental health applications...

And about that last post

In rereading it, I feel like I'm computerizing humans.  I don't want to do that.  The ultimate goal of my research is to make us humans the happiest, not to make us the best at computing.  And it's a tricky trap to fall into, because you start thinking "well, if I always make the best decisions, I'll be the happiest", which is not necessarily true.

That's all for now.

Emotions as quick mental computing

Here's a neat idea, and I'll jot it down even though it's not very polished.  (I think I'll start doing that more here; seems a good place for it.)

Emotions offer quick and sloppy reactions.  You see something that looks like a snake, you run away; only later do you realize it's just a hose.  You could sit down and concentrate and take in all the details of the scene and mentally label everything and say "that is a hose", but that takes too much time.  Emotions are quick at the cost of being precisely correct.

Similar things are going on in the computing world.  You hear about how SQL is getting used less and new sweet techniques (BigTable here at Google, for example) are taking its place.  They're not always 100% correct, for some value of correct; for example, when you search for something, it doesn't tell you the exact number of pages in the Internet that contain that word.  It gives an approximation, which is fine, because the number is about 12 billion anyway.

When I'm playing Dominion, there are 16 cards, and I don't want to exhaustively research every possibility every turn; that'd take forever.  So I pick a card that "feels" good.  Even more when I'm playing a long match of rock-paper-scissors, I have to just go based on gut feeling.  It's not 100% right, and sometimes I make mistakes (playing rock against someone who always plays paper, for example), but it is quick.

So what if we could train our emotions better?  You'd get the kind of tradeoff you get between trying to use SQL for everything and using new better storage solutions.  It's like the difference between Google search and, well, some slower search engine.  Fractions of a second?  Totally worth it.  Instead of trying to put them aside and compute exactly, take advantage of them and learn to make better quick calculations.

No deferrals? Plus, search by brute force.

First of all, it seems harder than I thought to defer grad school admission (apply now for 2011, and then say "I'm coming in 2012 instead").  It was a thing you could do in undergrad admissions, and my advisor seemed to think it was pretty reasonable for grad school too, but maybe not:

Stanford CS: "A student admitted to the Ph.D. program can defer admission for one year only. However, very few deferrals are granted, and then only for compelling reasons." (I'm assuming "I want to travel" is not a "compelling reason.")

CMU CS: "This policy may vary from program to program. After you are admitted, please contact your program coordinator."
The MIT Media Lab, UW, and U. Michigan all say nothing about it, and Tufts says flat out no.  Maybe I'll be applying next year instead.  That's too bad, because it feels like I was getting to the point of sending out apps, and then I'd be done with the application process.  I guess on the plus side, I've got another year to prepare.

Well, s'okay, it's been a bit of a slog anyway.  How do you find the right professors at the right schools when your goal is specific, interdisciplinary, and without a good name?  "Improving our minds in not-necessarily-cognitive ways with small computers" doesn't have a department at most universities.  So I've been looking mostly at HCI departments, but also media/etc programs, learning departments, neuroscience, and stuff as far afield as communications.  It's almost a buzzword search (which I'm not entirely proud of, but meh): "ubiquitous", "cognitive", and "persuasive" all get +1; "affective", "emotions", "brain-computer interface",  and "attention" all get +2, and "mindfulness" or "neuroplasticity" would get +5 if I could find them.

I was going from conferences first: reading the last 2-3 years of CHI, Ubicomp, and ACII, finding some papers that interested me, and looking up their authors.  But this clustered around MIT, CMU, Stanford, and UW, with a LOT of one-offs from other universities.  It's like trying to figure out what baseball teams are good, when only given a list of home runs and who hit them.  I already know those 4 schools are all-stars; where else can I apply too?

So I went bottom-up.  (or "top-down"?  I dunno.)  I went to schools that I heard were good at HCI, and looked through all the HCI professors.  It's borne a little fruit, not a huge amount of fruit, but some fruit.  Here's what I've found:

UC Berkeley: surprisingly little that's interesting to me.  Big HCI lab, but it's big on collaboration, new interfaces, and vision.

U of Maryland: probably not in HCI.  Again, interfaces, visualizations, collaborations, work with kids, physical devices.
There's interesting stuff in other labs on the AI side, but then this veers into the whole AI can of worms, which I'm not interested in.  I want to understand how we think (and feel and act), and help ourselves currently, not make computers think how we think.

U of Toronto: nothing in HCI very interesting to me.

U of Michigan: some work going on with a project called D-Sense, to try to help with the management of depression.

VA tech: again, not so much in HCI.  Maybe the Mind-body lab, more psych-ish and less computery.

Tufts: surprisingly interesting.  They've got a lab doing brain-computer interfacing, within a larger HCI lab, that sounds pretty neat.  Harvard too, doing some BCI work.  Either of these places would be cool, because I know that they've got solid programs in many departments, and they're in Boston.  (I imagine MIT collaboration is, well, not unheard-of.)

UC Irvine: has a cool ubicomp lab, but I'm not so interested in just straight-up ubicomp anymore.

Georgia Tech: I haven't actually looked at them, because I don't really want to live in Atlanta.

... perhaps with another year I'll find some more options!