Friday, October 12, 2007

The Best 3 Hours I've Spent in a Long Time

This will be part one of a three part discussion about the new gaming collection by Valve called the Orange Box.  The Orange Box contains Half-Life 2 and it's two expansions, Team Fortress 2, and the game that I will be talking about today, Portal.  The Orange Box is available on the PC (recommended) or for the X-Box 360.

I would say that the Orange Box would be worth buying if it only contained the Half-Life games and you hadn't played any of them.  If you have played Half-Life 2 and Episode 1, I'd still say it's worth it just for Ep 2 and TF2.  But the icing on the cake (The cake is a lie!) is certainly the final component of the set.

Portal is one of the most innovative games I've played in a very long time.  It mixes a puzzle game with a FPS.  It teases the brain and the reflexes.  It is hysterically funny (I literally laughed out loud several times).  It is just a tremendous amount of fun.

It is hard to do the game justice by describing it.  Portal is a series of 19 puzzles that place you in rooms with switches, moving platforms, hazards, and later on, turrets.  Your objective is always to figure out how to open the doors to the elevator that will take to next level, or in the case of the last level, to your eventual prize (The cake is a lie!).  Your only weapon is a portal gun.  Its primary fire will shoot a blue portal, secondary fire is a red one.  Step through the blue portal and come out the red one, or vice versa.  Portals will only stick to certain surfaces, however.

So you will use portals to get around hazards and to move boxes or energy ball around the level.  If you drop a portal onto the floor and put another one on the wall, when you jump through the one on the floor you will maintain your downward momentum, except it will change to horizontal momentum and you can launch yourself over wide chasms.

The best part of the game, however, is the computerized voice that is guiding you through each level.  The voice and dialog (or I guess monologue since your character doesn't speak) are full of such a brilliant, sardonic wit that will continually propel you to the next challenge, just to hear what she will say next.

The only criticism I have of the game is that it is too short.  I beat it in one sitting of about 3 hours.  You will fly through the first 15 fairly simple puzzles.  The next 3 are a bit more complex, and the final level is actually quite long and challenging before you can earn your just desserts (The cake is a lie!).  There just should be more like it.

There are a series of bonus levels that I have not tackled yet.  And hopefully more level packs will be released in the future.  However, while the puzzles are fun, they just won't quite be the same without the computer narrating your progress.

Oh, and one last thing.  This game may contain the greatest ending credits sequence of any game I've ever played.  No spoilers, but the song that plays over the credits is genius.  I want to get it into my iTunes library.  Might even give it 5 stars.

Oh, and even one more last thing.  The cake is a lie!

Wednesday, October 10, 2007

My Latest Project

Last week, I wrote a post about the lessons I have learned in more than three years of rating movies using Netflix. You can read the full article, but I will also provide a short summary. Because I generally select movies to watch that I think that I will like, my rating list is very top-heavy, with more than twice as many 5-star ratings as there are 1 and 2-star combined. I also realized that there is very little difference in my opinion of 2-star movies over 1-star, while there are actually tiers within my 5-star ratings that are only known by me and not reflected in the rating.

So that brings me to my latest project. I have recently started at the top of my iTunes music library and will listen to all 5240 songs (and that number will surely grow by the time I am done) and rate them using the 5-Star system that iTunes affords. And I will do so applying what I think is a better strategy than I have used for Netflix.

This song rating project is actually just a smaller part of an ongoing project that I've been working on for about 8 months now. Since I finally gave in and am using iTunes as the sole platform for managing my music collection, I decided that I needed to get my iTunes library into as good a shape as possible. I have already sifted through my music correcting inconsistencies in band and album titles ("Beatles" vs "Beetles" vs. "The Beatles" vs. "Beatles, The"), supplying correct album titles were there were none, and tracking down album artwork for every track in my collection. I am specially proud of my solution for the large number of live bootlegs that I possess. I grouped them by band and gave each set an album name of Live X ("Live Pearl Jam" for instance). I then found a picture from a concert by that band and used it for album art.

So the latest phase in that ongoing project is rating every song. I quickly realized that the problems I would have using a rating scale from "love it" to "hate it" would be even more severe than in Netflix. Here's why. Netflix is a catalog of every movie that I've ever seen. If I do watch a 1-star movie, I can't un-see it (and believe me, I wish I could get rid of the 20 minutes that I watched of Pink Flamingos). But iTunes is a collection of music that I possess. While there might be any number of Britney Spears songs that I can't stand, they aren't in my library. Of the 5240 songs currently in my collection, I would hazard a guess that I would say that I "like" at least 5000. Which, in Netflix terms would mean 3 stars.

Therefore I have started using a new ranking system that will hopefully result in a much more natural and even curve. I present that system to you, my loyal readers, for your consideration:


First an foremost I will apply this ranking to any song that I never like listening to. As I said before, there aren't many of those in my library, but sometimes you rip a cd and there is one song you can't stand, but you don't want to delete it and screw up the track numbering ...

But I will also use the 1-star rating for another purpose. Any track that I would label as "album filler" will get the 1-star. By album filler, I mean that there is no circumstance that I would wish to hear that track outside of the context of the album. While I think that the helicopter chase between Another Brick in the Wall parts one and two is really cool, I wouldn't really want it popping up in a playlist.


2-star songs are ones that I don't really have anything against, they just don't do a whole lot for me. White noise tracks. While I wouldn't skip them listening to an album, I also wouldn't ever seek to them. For one of these songs to go on a playlist, it would have to have a very specific concept, and even then it would be just because I am trying to pad the length.


A 3-star song is one that when I hear it, I really enjoy it, but I probably would never seek it out. Most of the full albums that I own, even ones that I really like, will be full of 3-Star songs. 3 star songs might turn up on a playlist if it fits the theme really well, or maybe if I wanted to some music to play in the background that I will enjoy but won't overly distract me.

3 stars will also probably be my default rating for song that I'm not sure about. I will keep my eye on 3 star songs when they come up in the future and may bump them up or down a star as I see fit. In fact, I might create one or more playlists for just this purpose.


Now we're getting somewhere. Whenever I've said "Oh, I love this song", that's an automatic 4 stars. Note that under my Netflix system this would be five stars, and you'll see why that would be a problem in a minute. 4-star songs are ones that I will hunt down when I'm in a mood for a certain type of music. Most of my general listening playlists will be full of 4-star songs.


To fit my song ratings under a normal curve, I should have very few 1-Star songs, and equally few 5-star songs. That's why 5 stars under this system is a special category. I said songs that I love will get an automatic 4 stars. I certainly love 5-star songs as well, they just have to have a certain extra quality to them.

I said that 4-stars are ones that I want to hear when I'm in a certain mood. Well a 5-star song is one that will PUT me in that mood. A 5-star song can make you cry when you're happy. They can bring you out of a funk. They can make you feel like you're sixteen again. They're powerful. Dangerous. A misplaced 5-star song could ruin a playlist. But it can also be the cornerstone of a great playlist.


So there you have it. It's too early to say (I'm still in the A's), but my hope is that this system will produce a much more even distribution of stars, with thousands of 3-star songs, hundreds each of 2 and 4 star tracks, and just a few dozen each 5 and 1 stars. I will probably still skew slightly toward the upper register, but hey, I really like music.

My ultimate goal with this ranking system is to be able to make better, smarter playlists. I also have some other plans to assist with this task that I would like to discuss, but I will save that for another time.

Thursday, October 04, 2007

Rating Systems and Other Trivial Matters

As I have addressed many times in this blog, I am a huge fan of Netflix. And one of my favorite features about the service is how spot-on the recommendation service is. I can't tell you how many times I've been interested in a movie that Netflix said I would give a 3-Star or lower rating, watch it anyway and wound up disappointed. On the flip side, I will receive movies that I don't remember why I ever added them to my queue in the first place and which don't sound terribly compelling, but Netflix suggests that I will think it is a 5-Star film, and damn if I don't end up loving it.

So, as well as the Netflix system work, I do have one issue with it, or rather, with how I use it. According to the Netflix Ratings Legend, giving a movie 5 stars means that I "Loved it", 4 I "Really Liked It", 3 I "Liked It", 2 I "Didn't Like It", and 1 star means that I "Hated It". I have rated almost 1300 films using this system, and a pretty clear trend has arisen. It is very top-heavy. My average rating is 3.8. I have given 407 movies 5 stars. And that's compared to only 186 movies with less than 3 stars.

If were to watch every movie ever made, or a true random sampling of movies, I sure that my average rating would balance out to a healthy 2.5 and fit under a pretty little normal curve. But the fact is that I don't watch films at random (though it does seem like it some times). I usually watch movies that I'm pre-disposed toward liking. At this point, I am fairly comfortable with my tastes, and before I even watch a movie, I can guess within a star or two what my eventual rating will be. So, of course I normally only watch movies that I think I will like.

If you look through my 1 and 2 star rated movies, you will mostly find bad action/comedy movies by the likes of Steven Segal, Rob Schneider, or Frank Stalone, which I generally can't stand but that my stupid friends drug me too at some point. Only a very small percentage of these are movies that I actually wanted to watch and completely missed the mark on.

Notice above that I said that I was "disappointed" by a 3-star movie. According to Netflix this means that was supposed to have liked the movie. That's the issue. Because of Netflix's monstrous catalog, I am no longer content watching a movie that I merely "like".

So, movies that I self-select start at 3 stars. This basically changes how I use the Netflix Rating System from a 5-Star system to a 3-Star. Instead of "Like It", "Really Like It", "Love It", to me now it is closer to "Ok, but kinda disappointing", "Pretty Good", and "Great".

Among my 1 and 2 star rated movies, I basically have no strong feelings. When it really comes down to it, I'd no more rather re-watch a 2-star movie than a 1-star. But on the other end, there are clearly tiers among my 5-stars. I may "love" both The Godfather and Shaun of the Dead, both by my definition of the word love and by the fact that I've given them both five stars. But I would never pretend that there wasn't a difference in how I feel about the two movies. The problem is that only I know what that difference is, to Netflix they're both the same.

If I were to redo all of my ratings, I would merge all of my 2-star movies in with my 1-stars. I would then slide all of the other categories one star down (3's become 2's, 4's become 3's, 5's become 4's). Finally I would go back through the list of now 4-star movies and select the ones that I thought were truly great, and bump only those back up to 5-star status.

So why am I not going to do this (at least anytime soon)? Well, first and foremost, this would be a pretty time-consuming project. Free time is not something that I have an overabundance of. Secondly, like I said in the opening paragraph, I think that the recommendation system works really well. I'm a little afraid that if I change something that's working, it might break it.

Therefore, instead of changing my Netflix movie rating system, I am instead applying the lessons learned above in a new project, which I will discuss very soon in this very space. In fact, this post was originally supposed to be about that new project until my preamble about Netflix ran into its second page. So come back sooner than my more normal 10 day / 2 week update schedule to hear about that.