Monday, April 28, 2008

Rock of Ages

This last weekend my 14 year-old son Nathan went to see Def Leppard, Styx, and REO Speedwagon in concert at the Cricket Wireless Amphitheater. I was younger than he when Styx and REO were starting out, and when i first heard Def Leppard i was 16 (Rock Brigade). This would have been a kick-ass concert in 1980, although then Styx would have been headlining.

I think it was all sort of kitschy and retro for my son and his friends, as if i had gone to see Dion and the Belmonts when i was a kid. He appears to have liked it at least as a technical experience (he was very critical of the Def Leppard sound-- too much bass covered up the guitars he said). He also genuinely seems to like the Styx song Come Sail Away, which i have to admit holds up better than some of the Styx songs i liked when i was in high school (although Renegade still kinda rocks). It's their equivalent of Bohemian Rhapsody, which my son's generation knows from watching the video of Wayne's World. It's sort of odd that kids from the 2000s know about music from the 70s due to a movie about the 80s made in the 90s.

Monday, April 21, 2008

Frontiers in Data Visualization

I like this graph, which comes from a tool that will remain nameless. It's just so Zen.

Tuesday, April 15, 2008

Free Range Kids

About a week ago there was a buzz in the various media about a mom who let her 9 year-old son take the subway home alone from Bloomingdales in New York City. I thought it was pretty cool, though i admit it's very scary also. I'm not sure i'd want my 10 year-old son to rely on San Diego's public transit system, and I have to confess that i don't think either of my kids have had an equivalent test of their independence.

Now, what prompts me to write about this is that when i was 9, i broke my arm while driving an all-terrain vehicle on my grandfather's farm. I was riding down a path with another kid on an ATV, we bumped into each other and went flying. He suffered a concussion and i had my left humerus snapped in half. Needless to say we weren't wearing helmets. I'm not even sure they made helmets at that time. When i look back on this now it seems amazing that my parents allowed my sister and i to ride anywhere we wanted. I get nervous when my kids ride their bikes in our suburban neighborhood.

We had other freedoms that seem odd in this day and age. My sister and i were often at home alone, even at night, and we never had baby-sitters. When i was in kindergarten (in Arizona at that time) i walked about a mile to school each day with one of my classmates. I was 5. I owned my first knife when i was 6, and by 1o or so i was free to use the collection of rifles and shotguns we had inherited from my grandfather.

I can't really say that this turned me into a more independent or risk-taking individual than i would have been otherwise. My life seems fairly mundane by the standards of my generation. I left home at 18, and have lived on my own since, but that was both normal and expected at the time. I suppose that compared to most people in the rural Indiana community that i grew up in, living in California and working in the Internet industry is a slight departure from say, farming, but it's hardly like i became an astronaut or something.

While i think that developing a sense of independence in kids is a valuable thing, i'm also inclined to think that it's probably better to instill them with a legitimate sense of non-conformity. I hope that my kids feel comfortable taking chances and doing things on their own, but i really hope that they pursue their dreams without regard for external expectations.

Thursday, April 10, 2008

Probability Confuses Me

My boss asked me the question "If a track has a 1/15 probability of playing, what is the probability that it will play in a sequence of 15 tracks" (yes, this has to do with on-line radio). The simple answer is the basic binomial case, or what is the probability of 1 success in 15 tries given that the probability of success is 1/15, or (15!)/1!14!*(1/15)^1*(14/15)^14. That works out to about a 40% probability.

However, the answer desired could be the probability that the track plays 1 or more times, which is closer to 65%. However, in the specific case of radio the rules say that a track can't play more than once in a certain span, so we're back to the original case if the separation is large enough. On the other hand, the binomial case assumes that the probability of a track playing stays the same, but if the track is chosen randomly from a fixed pool, and that pool become progressively smaller as tracks are chosen, then the probability changes. For example, in the case where you've only got 15 tracks and 15 spots to put them in, then the probability that the track will play assuming repeats aren't allowed is 1.

This is why probability confuses me: many times figuring out the question is harder than finding the answer.

Monday, April 07, 2008

Mr. Henry Goes To Washington

Henry Drumming in Washington DC
Originally uploaded by mikemull
My son Henry is in the Washington, D.C. area this week with other 5th graders from his school. This is a picture of him drumming during something called "The African Experience".

Tuesday, April 01, 2008

Popularity vs. Preference

One of the things that dawns on you after a while working with music recommendations is that popularity and individual preference are two ends of a spectrum. Say for example that you have N votes to allocate among the N musical artists in the known universe. How you allocate those votes is one expression of your personal preference. If you take everybody's votes and add them all up, you get popularity, which is an expression of the preference of the population.

This is essentially the model of sites like Digg and Reddit too, which percolate submissions to the top of a list based on the number of people that vote for them in a certain span of time. The model works pretty well when precautions are taken to prevent people from gaming it. On the other hand, it's not really recommending things in a personalized way even if it seems like it sometimes because the user base of such sites is overwhelmingly biased toward libertarian uber-geeks.

In a sense music recommendation systems work somewhere between the two ends of the spectrum, trying to find a subset of listeners similar enough that it's safe to assume they have common tastes, but not so similar that there's nothing left outside of the intersection of their preferences. This happens in practice sometimes also, by grouping users or items or both into related sets (i like A, A is related to B, so recommend B; or i like A, she likes A and B, so recommend B).

I'm not a big fan of this approach in music recommendation because it seems both self-limiting and subject to feedback loops. But it's kind of interesting in a personalized radio context. For example, if you're building a station around a particular artist you'll present tracks by that artist as well as tracks by related artists. You'd expect that any listener who has requested the initial artist would like the presented tracks about as well as any other listener. But the variation among listeners even in this narrow context is fairly drastic, especially with bands that have a large catalog. One notable example is R.E.M., a band that clearly has two sets of fans-- pre-Out Of Time and post-Out of Time. In the case of radio where you're presenting a set of tracks that is supposed to be personalized to a particular type of listener, the variation of track popularity among users is rapid feedback on the quality of your recommendations.