Kid Tetris

Metacritic uses 40some publications from which they cull reviews that go into their "Metascore". How does any invididual publication compare with the overall Metascore? I have convinced myself that Elvis Mitchell and A.O. Scott get about the highest weighting, but actually reverse engineering this to prove it (they won't give you the weightings for Metascore) is sort of hard. I like the Metascore or Rotten Tomatoes average rating so much better than the compilation of crazy user reviews which are either 10, 9 or 0 (the best ones start with "I didn't see this..."). I'm glad the polarization of our country can get taken to the logical extremes, where people can divide into camps about based on how much they loved or hated an admittedly schmaltzy movie.

I would buy season tickets, but I get bored looking at the home schedule. The 1st, 2nd and 3rd place teams in the conference last year are away games.

The elevator in my apartment gets recalled to the lobby after every trip. This is great if you are just coming home, but if you are on any other floor than the lobby, you are basically guaranteed a wait. The alternative model is that the elevator doesn't get recalled to the lobby, so you may or may not have to wait for it on any given ride.

Which is better? Well, if you assume that all trips originate or end on the ground floor and whether a trip is going up or down is random, the ground floor recall model will always yield a lower queue time (but with a higher variance than the other). This is appropriate for my apartment, I think. And, obviously, the elevator recall will take a much longer time if the elevators travel randomly across floors (something like 50% longer).

However, in an office building it needs to be more sophisticated. If it's the morning, the elevators should be recalled to the lobby (my office building does not do this). If it's the evening, they should be recalled to the middle floor (my office building does not do this, either - in fact, after six, they get recalled to the lobby, where you must wait the longest for it).

The expected time savings (using some assumptions about patterns in trips) is about 10% versus strict lobby recall and 60% versus the model where the elevator doesn't do anything. However, if you believe that time gets more 'expensive' - i.e. you would pay more to cut your wait time from 5 minutes to 4 than you would to cut it from 3 to 2 (and the function in question is quardatic), the savings are about 15% for lobby recall. If it's the opposite - you would pay the most for a 1 minute to zero minute wait, then it's around 5% for lobby recall.

It should be clear that I've never actually studied queueing theory and I'm doing this because I want to bitch to the property management firm that the elevators are too stupid for me to have to deal with.