Bloomsday

During breakfast yesterday I was talking to Elif about what Harold Bloom had to say about Flannery O’Connor in his Short Story Writers and Short Stories and she asked me if this was the same guy who was Saul Bellow’s friend (and on whom Bellow’s Ravelstein was based). I said yes. Well, it turns out (as she gently pointed out to me later that day) there are two Blooms, one Herald and the other Allan, totally unrelated, but both literary critics with relatively similar intellectual bents. I had read parts of The Closing of the American Mind (by A.), liked the literary criticisms about various short story writers, and was quite impressed by “Bloom’s” depth of knowledge. I still find the Bloomian literary reach quite amazing, even if it is the sum of two minds.

Curious questions come to mind: Have they ever met (past tense, since Allan Bloom has passed away)? Quite possibly, since they were/are academic stars in their literary circles. What is the chance of two eminent critics with same last names to be born in the same year (1930), two months apart? What would have happened if they taught at the same university?

In his analysis of Jorge Luis Borges, Carlos Fuentes has famously said that Borges’ fantastic literature has four components: “The dreamer, the metaphysician, the time traveler, and the double.” Could the Blooms be doubles of each other? That would be an interesting story.

Using Game Theory to Pick Up Blondes

I recently read Prisoner’s Dilemma, a history of game theory that also covers John von Neuman’s life. I wished it was more technical, still a great read, though. One of the interesting nuggets of information was that conservative and liberal approaches (in the American sense) correspond roughly to the cooperation and defection strategies in the Prisoner’s Dilemma (PD) game. I also came across this great quote by the mathematician Jacob Bronowski:

You must see that in a sense all science, all human thought, is a form of play.  Abstract thought is the neoteny of the intellect

In Chapter 11 there’s a classification of two-person symmetric games (i.e. payoffs are the same for each player under similar circumstances), which are both the easiest to analyze and the most common. There are just four payoff values in such games, DD (when both players defect), CC (when both players cooperate), CD, and DC. Each game can be represented by the ranking of these payoffs, e.g. DC>CC>DD>CD. There are 24 such orderings (you should check that). However, not all of these are interesting. Prisoner’s Dilemma is interesting because of the conflict between the common good and individual rationality. Each player wants the other to cooperate but wants to defect themselves.

So, what does that imply? To have interesting games, we should have CC>CD and DC>DD. This leaves only six possibilities. Not all of these are troublesome, either. For interesting paradoxes, like PD to emerge, we require that there should be either an incentive to defect when the other player cooperates (DC>CC) or an incentive to defect when the other player defects (DD>CD). This eliminates two more possibilities and we are left with four classic 2×2 games that all turn out to be very common socially.

• CC>DC>DD>CD Deadlock
• DC>CC>CD>DD Chicken
• DC>CC>DD>CD Prisoner’s Dilemma
• DC>DD>CC>CD Stag Hunt

You might be thinking this is all dry, theoretical stuff. However, such nerdy mathematical knowledge can come in handy in the most unexpected circumstances. Case in point: the pickup scene early in the movie A Beautiful Mind, where Nash and gang see a blond and four brunettes walk into a bar. While his friends are getting ready to strike it with the blond, Nash explains to them that if they all try to talk to her, the worst outcome will occur: nobody will get her. And nobody will get the brunettes either, since they would hate being seconded. They should, therefore, cooperate (a more in-depth analysis of this scene can be found in this blog). This works perfectly in the movie, the friends go for the brunettes, Nash gets the blond and the theorem. In real life this might be harder to pull.

Of course, if you look like Russel Crowe, I think you can safely omit the math and proceed as usual.

Day 2 at ETech, The Evening

At lunch I had an interesting chat with a librarian from UCSD. His theory why libraries (and librarians) are getting out of touch with the public in the age of Google was that in the 80s libraries made big investments in the (then) newest technology so it was hard for them to switch. Now teh first thing they ask of any data management system is: can we get the data out.

Afternoon Talks

Bo Cowgill talked about what they’ve learned from their prediction market experiment, which is the largest in the world. Main result was that the best predictor for how people traded online was spatial proximity. I was expecting more analysis and interesting details. I bailed out of that towards the end and caught the end of Violet Blue’s (no, that’s not a pseudonym) talk about sexual identity online. Blue is a sex educator and maintains an explicit blog. She talked about her experiences and some of the problems with using her real identity online. I was surprised to learn that her biggest trolls (who even sent her death threats) were women, who were evidently too dumb to notice that they can be tracked back with email headers.

Gina Trapani’s Personal Productivity Session

This one saved the day for me, it was great. Trapani has created the uber-popular blog Life Hacker and talks faster than an excited football announcer. One of the main ideas, that resonated hugely with the audience was: It’s good to be efficient but do not lose sight of why you want to efficient, sometimes shopping at a supermarket with your sister for an hour is better than splitting up and finishing the shopping in 10 mins. The seven habits she advocates to increase productivity were:

• Have a system: e.g. always put car keys at the same place
• Get things out of your head: Classic GTD. Do stuff immediately, e.g. answer email immediately after you read it
• Piling, not filing: Don;t create complicated folder structures, e.g. for mail, use tags, like GMail doe
• Park on a downward slope: Leave a note for yourself when switching tasks so it’s easier to pick it up later
• Build strong filters: Only look at important stuff.
• Clear the clutter: You should clear your field of vision, e.g., desk space, inbox.
• Have doable todos: Don’t sabotage yourself with huge or vague tasks, e.g. “clean the garage”

She also recommended the followinf software: textexpander (only on Mac but a Firefox plugin for Windows is available at Life Hacker) , quicksilver (Mac only), and KeePass.

Partying The Night Away

ETech provided drinks and food in the evening while we looked at the sponsoring companies who have booths. A guy from Yahoo showed me their new stuff at the Yahoo developer network. Great site with tons of documentation. He told me that that was the documentation, i.e., they don’t keep internal and external versions. Sweet. However, Sun owned the floor with their new Sun Spot thingies. It’s not clear where Sun wants to go with these things but they are cool. They use ZigBee to communicate, programmable with Java, and can be combined like Lego bricks.

Google was there, too, but they only showed the Android simulation on two screens. Lame. They can do better than that :-)

After the ETech reception there was another party thrown by Disney at 8:30pm, oh the joy! They even had a raffle but they announced the winners at 10:30 and only had 5 prizes. I never win anything from these but the night was not a total waste: I talked with a manager from Intel and a IT staff person from Purdue University (go Boilers!). Our main topic was, as you might have guessed, home schooling and the crumbling American school system. Interesting discussion, but that is the stuff for another blog.

Day 2 at ETech, the Morning

Collaborative Mobile Gaming from MegaPhone

Day at the conference started with a great social game demo from MegaPhone where people dialed a number and played a shoot’em up game on a large screen. It was great but with 20 people on one screen (they said they normally allow only 10) got somewhat chaotic and old after 3-4 minutes. What happens if you want a more interesting game with lots of players? Next demo illustrated this, people signed up by calling the number and random pairs were selected among them. These pair then try to find each other and enter each other’s phone number by using a secret code that is sent to both of them as an SMS message.

Keynote Speeches

The first keynote was by Saul Griffith about Energy Literacy. The world’s power consumption rose from 9.5TW in 1980 to 15TW in 2005 (compared to 1TW in 1890); how are we going to cope with this extra consumption without wrecking the environment? Griffith thinks raising to this challenge is well within our technological capability. He went through an in-depth calculation of his energy footprint, i.e. how much energy he is using. His estimate was 25kW. I wish he talked more about how economical factors can be used against global warming rather than the “let us all be good and use less” type of argument.  The recent surge in energy startups was probably caused more by the fact that VCs saw opportunities to make money rather than the goodness of their hearts (everybody knows they are heartless).

Other interesting keynotes were given by Chris Melissanos, Sun’s Chief Gaming Officer about Sun’s Project Darkstar (shardless, open source JSE design that is game and client agnostic that will make it easier for small companies to enter the online video gaming business that is expected to be $11B by 2011),  and Elizabeth Churchill from Yahoo (use of displays in public spaces).

Peter Norvig’s Session : How Billions of Examples Lead to Better Models

Norvig, Google’s Director of Research, talked about how having a lot of data (e.g having a collection that of a couple of trillion words). Particular applications he talked about were automatically generating annotation from images (paper), segmentation of ambiguous domain names (e.g. whorepresents.com, therapistfinder.com), correcting typos in queries, and, of course, machine translation. Most of the talkw as standard machine learning stuff but his comment about machine translation was interesting: Difference between computer and human translation accuracy is very close to the differences between human experts, in a short while we have to find a better way to rate the translation algorithms. He is expecting that newm real-world data sources (GPS, etc.) and handling video is the future in this line of work.

Day 1 at ETech, The Rest

Marc Hedlund’s Debugging Hacks Session

Marc talked about principles distilled from their day to day activities at Wasabe, which is a personal finance site that is becoming popular. Some common mistakes he listed

1. “That doesn’t look right, but it’s probably fine” If you think there’s a bug, most probably there is a bug.
2. “It seems to have gone away” If you haven’t fixed it, it’s still there.
3. “I bet I know what it is” Wait to form theories until you have data.
4. “That’s impossible” Set up logging/exceptions/asserts and make sure you get the report.
5. “Beats me…probably a race condition” This one is ridiculous.
6. “I’m just going to fix this other bug quickly” This will suppress the original bug.
7. “That probably the [server/client/not my] code” Don’t guess, prove it.
8. “I think I found a bug in the OS” Most probably, it’s your code.
9. “I found a bug and fixed it. Done” Did you find the bug or a bug?
10. “I’ve got to get this out now — no time for ____” Stick to a good process even if urgent, break down suppression and closure.

You have to be relentless against bugs, never give up, but have a method. That was the basic message. They use FogBugz to track their bugs and Campfire to discuss them. Another interesting point: Do not blame people for bugs, deal with chronic bugs in private. If bugs cause punishment, reports will be killed.

The Evening 

At 7pm we had the AppNite where developers talked for 5 mins about an app they’ve created and the audience voted for the best (a la SMS, like American Idol),  here’s the whole list of apps. First were 5 facebook apps, followed by 5 Open Social apps on Orkut and MySpace. Most were pretty mundane stuff. The ones I liked were Living Social from Hungary Machine, extending their Visual Bookshelf to restaurants, beer, etc. and Chirp Screen, which creates an interactive screen saver from photos that your buddies have posted.

Then the man himself, Tim O’Reilly, took the stage with a talk entitled “Why I Love Hackers.” He did not disappoint, it was a fascinating talk touching on a lot of things such as the fact that increasingly apps are being driven by new kinds of sensors, cool companies (Immi, Willow Garage), and body and brain augmentation. But what really amazed me was that he finished with a poem from Rilke (“The Man Watching“), which he interpreted as an ode to hackers. It was really excellent. BTW, this must have been the first exposure of a lot of alpha geeks to German poetry!

That was followed by a round of Ignite Talks. Yours truly did one, too , though I wasn’t great. The 15 second per slide format is hard to master and the topic was a little too technical. You be the judge when the videos are posted to YouTube soon.

Day 1 at ETech: The Morning

Came to Marriott at 8pm, got registered. There are about 100 people, today is the day for tutorials so I guess not everybody’s here. Interesting crowd. Attire varies from cool-print shirt-with-jeans hip hacker to good old khakis-and-polo (these are rare though).

Nice breakfast. I know that’s a lame thing to say but you’ve got to see what passes for a continental breakfast at some other conferences. The iPhone tutorial is in 10 minutes, at 8:30. Too bad I can’t be at two places at once, missing Kathy Sierra’s tutorial is a shame.

Nate True’s iPhone Development Session

Oh, Nate just said that the directions he’ll give are just for Macs, shoot. He’ll post instructions for other platforms later in his blog. So I’ll just watch along. He gave a bit of history about iPhone hacking: First people who worked on unlocking started to talking on OSX86.hu, this is still a good place for info. The initial hacks exploited the fact that the ramdisk was not encrypted and got the key for the main disk. The earlier jailbreaks reconfigured AFC as a first step so all the disk is writable, breaking the “chroot jail”; hence the name jailbreak.

Apps compiled with the hacker SDK are likely to be compatible with the Apple SDK; no guarantees, though. After compiling, pass the app to the phone using AFP (uses wireless connection) .

Tom Carden’s Web Visualization Session

Since the iPhone tutorial finished early I had a look at this one until lunch. He showed lots of cool Web data visualizations. Things I learned about:

• EveryBlock: Localized news (down to the block level), currently only for three cities;
• flare: AS3 tools for building interactive visualizations;
• Edward Tufte: Pioneer in visualization and interface design

An interesting thing he said: A good visualization starts with simple questions, answers those, but generates more interesting questions.

The lunch was served outside, awesome, even got a slight tan. Had a fascinating discussion about simulating single neurons. Note to self: How about this idea: put 1-10 neurons in a box so that they live 2-3 weeks, connect them to an USB interface, and provide software so that people can play with the input and output. Sort of like Sea Monkeys. When those die, sell them new cells :-)

Countdown to ETech08

I arrived at San Diego around noon, the weather is GREAT, sunny, around 65. This past Friday we had the 36th round of snow in Chicago so this came as a relief, it certainly was weird to see people in shorts jogging at this time of the year. Rebecca Goldstein writes “I feel an immediate closeness to anyone who loves New York or hates Los Angeles. Either condition is sufficient, but I’ve found that satisfaction of the one usually entails satisfaction of the other.” I used to feel the same about pretty much the whole of CA. Not anymore!

ETech starts tomorrow with cool tutorials. I was really looking forward to iPhone Software Development: Past, Present, Future by Nate True, but Apple has not released the iPhone SDK yet, so I wonder what we will cover. I will also talk at the Ignite Session tomorrow evening about Natural Language Watermarking. This will be my first time with the 15 seconds-a-slide format. We’ll see how that goes.