Monday, March 17, 2008

Tesla Production Begins


They started production today. St. Patrick's Day is a good day to start making a green car.

Still haven't decided what color I'm going to choose. I'm thinking Electric Blue!

Sunday, March 16, 2008

Adam's Station: Smart Playlist for my iPhone

I was discussing Vietnamese iPhone imports with my friend and Madagui teammate Laurent last week. He said he was interested in getting one of the new 16GB iPhones. I didn't know they had 16GB iPhones. My response was that my 8GB iPhone was rarely full. You may think that I have a lame music collection, which may be true, but I do not have a small music collection. So how do I deal with the fact that I have a 30GB music library on my computer but only a an 8GB iPhone? The trick is just keeping the good stuff on your iPhone or iPod.

Smart Playlists to the Rescue

I would argue that a large percentage of any music collection (especially mine) is crap. The trick is just keeping the good stuff in a playlist and then syncing that playlist. iTunes has a really great feature called Smart Playlists. A smart playlist allows you to create a playlist that is automtically updated. In the mood for Madonna? You can make a playlist of all your Madonna songs. Just create a Smart Playlist of all songs where is Madonna the artist. Smart Playlists are dynamic, this means that when you buy that new Madonna album, all those new Madonna songs are automatically added to your Madonna smart playlist. Pretty cool huh?

Rate Often

Of course you don't want to really listen to Madonna all the time. You probably just want to listen to the good stuff in your library. How do you find the good stuff? Originally I solved this problem by creating a Smart Playlist of all songs that were rated 3 stars or higher. You can rate songs while listening to them on your iPod or you can rate them directly in iTunes. This is simple and works pretty well. You just need to spend some time rating songs and anything with a decent rating will show up on your playlist. There are a couple of problems with this approach. You probably have a lot more music in your library than you have time to rate. As a result, the playlist gets stale quickly. For a while I had two playlists, one with everything and another with my top rated songs (rated 3 stars or higher). When I got tired of the top rated lists, I would just listen to all the songs in random order. When I found a song that I liked, I would rate it with 3 or more stars, automatically adding it to my top rated list. This worked when I carried around my big 30 GB iPod with all my songs on it, but it was still annoying because so much of the overall library was made up of songs that I just didn't want to hear. I have a lot of kids music and classical music that I don't typically listen to.

My current solution is based on this article by Adam Knight. It involves four playlists. The first is called the Core 2 playlist. This is anything that you might reasonably care about. I create this by excluding music I know I won't want to listen to on a regular basis. You can see the definition of my Core 2 list below. First I exclude anything that has a one star rating. This allows me to banish a song by rating it with 1 star. This allows new, unrated music, to get onto the list. I also exclude non-music items from this list.


It took me a while to fine tune this list. I suggest creating a similar smart playlist and then looking it over. Try sorting it on Genre and then scanning down the Genres. Are there any weird Genres you don't want to have on your core list? If so, exclude them. You'll probably find a bunch of stuff that doesn't have a Genre setting. You may end up spending some time cleaning up your music collection. Note that you can change the genre of a whole group of songs by selecting the songs in your music library and then typing cmd-I and editing the genre field. (You can probably do the same on Windows by using control-I.)

Now that you have your core list, create two new playlists based on it. One is called Infuser. The Infuser list is made up of songs from your core list that haven't had much play. You can see mine below; it's made up of songs that have been played fewer than five times.


The next list is called Sprinkler. It has songs that you haven't listened to in a while, but you seem to like. My current Sprinkler list has core songs that haven't been played in the last two weeks but have a play count greater than 5. If I've listened to it five times, I probably don't hate it.

Finally let's mix these together to create Adam's Station. You can call yours what ever you like of course. I select songs from either list and order them by least recently played.


Now I setup my iPhone to simply sync the Adam's Station playlist. It will complains a bit since I'm syncing a Smart Playlist based on another playlist which isn't on the phone, but it still works. If I'm listening to a song on the road and I don't like it, then I banish it by giving it a rating of one star. If I find a song I like, I typically rate it with three or more stars, ever improving the old My Top Rated playlist.

Wednesday, March 12, 2008

TwitterIM Down?

Normally you can use AIM to send a Twitter update to TwitterIM. Just tried it and got this message.

Monday, March 10, 2008

Veoh Monetizes Snake Video


While watching a video from a local snake restaurant, a catchy ad for the Cayman Islands popped up as an overlay. I quickly grabbed a screenshot before it went away. This is a great example of the perils of trying to monetize video. While more and more video is coming online, it's difficult to figure out what video is appropriate for ad placement.

I'm guessing the Cayman Islands chamber of commerce didn't expect to have their ads placed on top of a video featuring the decapitation of a live snake!

Here's the original video. Please comment if you experience some interesting ad placement while watching the video.


Online Videos by Veoh.com

Sunday, March 9, 2008

NRK uses Miro, Amazon S3, and Bit Torrent

I was reading this article about Norwegian Public Broadcaster, NRK, using Miro to distribute HD Television and was surprised to learn that Amazon Web Services supports bittorrent as part of it's S3 service.

I came up with an idea over the weekend (computer says no) that would require S3 and maybe EC2 as well. It has to do with a lot of compute power and storage, just the kind of thing AWS is meant for. Now that I know about the S3/Bit Torrent thing, I might actually do some work on it.

I'm a big fan of Miro, behind Firefox it's the most uses application in our house. Especially after I figured out how to get prime time TV on my Miro....

Saturday, March 8, 2008

Future of Search Talk

Here's a talk I gave back in the Spring of 2007 at the Future of Search meeting in Berkeley.

I just ran across it while looking at the webmaster tools for this blog. Some searches that I would expect to lead you to my blog, also lead you to this video. Not surprising I suppose.

It's short and sweet. Enjoy.


Sunday, March 2, 2008

Ancestor Arithmetic

Here's an interesting question: How many ancestors did I have at various points in history? As we all know, each human has two biological parents, at least so far. This means we can approximate the number of ancestors we had num_generations ago like this:

num_ancestors = 2 ^ num_generations
If we make the simplifying assumption that a generation is about 20 years, then this means 200 years ago (about 10 generations) I had 1024 ancestors running around when the Unites States of America was just a startup. Huh, I wonder what they were all up to? My home town of Irvington, Illinois only had around half as many people in it when I was growing up there.

OK, so what about 2,000 years ago, back when the Romans were masters of their domain? That would be 100 generations ago. Using the formula above, we see that:
num_ancestors = 2 ^ 100 = 1,267,650,600,228,229,401,496,703,205,376
Hmm, something's not right here. How can I have had 1.2676506 × 1030 ancestors? There are only around 6 × 109, 6 billion, people on the planet today, let alone a couple thousand years ago. The problem is with our model. It doesn't take into account common ancestors, ancestors that are counted more than once.

Let's say, for the sake of argument, that I married a cousin. My wife and I would then have one set of grandparents in common, assuming it's that kind of cousin. Using the formula above, my children should have 8 great-grandparents. However, since my wife (cousin) and I share a pair of grandparents, our children will only have 6 great-grandparents instead of the normal 8.

The point here is that there will be overlap, a lot of overlap, in the old ancestral tree. The higher up the overlap, the larger number of branches will be eliminated.

I wonder if there is some way to get a reasonable estimate of the number of ancestors one actually had running around the planet at some point in the past.