No No Netflix

[UPDATE 3/10/10  - The problem at Netflix that prompted this article appears to be fixed. Checking the rating predictions today, they are back in line. Whatever caused the "growth spurt" in ratings has been taken away. I was not seeing UFOs - it happened right here on my same old computer that I was using exactly the same way as I always do to access Netflix. Now I do not have to waste time modifying my personal movie database file]


What's with Netflix?  A few days ago they changed the look of the details pages for the movies. Things are moved around on the page and it looks less colorful, but it still seems to have the basic info I want, so I did not think much about it. Then I started noticing the predictions for how I would like the movies seemed to be a lot higher all of a sudden. I don't follow news releases or similar information at Netflix, so I don't even know where to look for information about such a change. So I did a Google search and did not come up with anything from Netflix itself, but I did come up with some forum discussions by subscribers.

The people on the forum were overwhelmingly irritated with Netflix, but their complaint was that Netflix had made changes to the friends features which seemed designed to phase it out. The complainers also agreed Netflix was non-responsive to customer input and as an egregious example cited how their numerous complaint postings about the matter at hand are being systematically removed from the Forum by the Netflix moderator.  I did find one stray posting asking about predicted ratings jump ups, but the poster said he did not have a record of the prior predictions. Well I do.

Since the Netflix queue cannot be sorted by column (a glaring deficiency in my opinion), I maintain my own database of the queue for sorting purposes. When I enter a movie in my database, I put the date of entry and the Netflix prediction as of that date as part of the movie title. Then later, as I update the database, I change the entry date and put the current prediction in a separate column. I have been doing this for over two years, and I have never noticed the Netflix predictions changing much over time.

So now what I have done now is go through about 100 movies in my queue to see how much they have changed, and the changes are astounding. Of the 50 movies in my queue added most recently or up[dated within the last three months, 34 increased by up to almost one full star, 5 decreased mostly by just one tenth of a star, and 11 stayed the same (but one of these I had already rated and 5 were not predictions for me but rather all member predictions). The average increase for these 50 was .366 stars, quite significant for me since my own ratings usually only span from 2.7 to 3.5. Of the oldest 50 movies in my queue, last updated by me between May and November, 2009, 40 increased by up to 1.2 stars, 5 decreased by just one or two tenths of a star, and 5 stayed the same. The average increase for these 50 was .392 stars. For me personally, I will be discounting the Netflix predictions by .4 stars; my old cutoff for putting a movie in my queue was 2.8, but will now have to be 3.2.

Netflix has always skewed the features of its web site to encouraging rentals more than making the site useful to subscribers, but this unexplained, significant padding of the recommendation ratings is going too far. I read that the company stock has been running high, based on projections of adding new subscribers, but some financial analysts think the stock is overpriced. If the Netflix predictions of new subscribers re as inflated as their new projections of how I will rate movies, I think the analysts are right.

My Top Rated Shows Watched in 2009

Sister-in-law Shirley suggested I do a year end piece on my top ten or on academy award style winners. I still have Siskel & Ebert lists that Shirley sent me each year, going back at least to 1992, on which I periodically check off my viewing. A personal list of favorites of an individual, like Siskel or Ebert, is helpful if you are familiar with the tastes of the individual. Lists and winners determined by group voting, like the academy awards, are less specific guides, but I still consult them for leads to movies to investigate.

NPR did a fun piece this morning about the academy best movie category being expanded to ten films. Seems they want to make sure enough blockbusters are being promoted to help sell tickets and also to pump up the ratings for the academy awards TV show.

I don't go to theaters to watch movies. What I watch comes from Netflix and the public library. I like a variety and have lots of historical interest in movies. During 2009, I watched 420 DVDs, in 17 different genres, in 17 different languages other than English, released in years from 1924 to 2009.  I rated them for myself as explained elsewhere in this blog. Below is a list of the dozen top rated ones, rated 3.7 or above. When I compiled the list from my database, I was surprised to see that eight are documentaries and only one of the 12 is from before 2000.

Scottsboro: An American Tragedy. From PBS American Experience in 2000, excellent documentary on this fascinating case from the Jim Crow era.

John Adams. The 2008 HBO miniseries on our second President.

Anne Frank: The Whole Story. From ABC in 2000, this drama tells the story of what happened to Anne after their hiding place was discovered.

Beyond Belief. Documentary from 2006 about two 9/11 widows who travel to Afghanistan to meet with widows there.

What I Want My Words To Do To You. Documentary from 2003 about a writing program for women prisoners who read from their writing and have them performed for other inmates by professional actresses.

Freedom Writers. Drama from 2007 with Hilary Swank as a teacher inspiring at-risk students to keep journals about what violence does to their lives.

War Made Easy. Documentary from 2007 about how in the last 50 years American Presidents have propagandized us into wars.

Bigger, Stronger, Faster.  Documentary from 2008 about the pressure of sports and competition and steroids.

Cats of Mirikatani. Documentary from 2006 about a Japanese artist who was visting America at the time of Pearl Harbor and the bittersweet life he has led since then.

Leila. A drama from Iran in 2000, telling the story of societal pressures on a couple having difficulty conceiving a child.

October Road - Season One.  Drama series from ABC in 2007 about an author returning to his small town roots and confronting old friends upset aboout how he portrayed them.  This is an example of a good one season series that became grossly inferior in its second year.

Frost-Nixon.  The documentary of the original interviews from 1977.

How about letting us know what shows you watched in 2009 that you rated tops?

Deciding What Movie to Watch

Is there a difference between watching a movie and seeing a movie? To me, seeing seems deeper, involving a conscious effort to find the message or meaning, while watching is more akin to TV channel surfing. I watch a movie and if there is a message or meaning that I get, then I see.

By nature, I over think, and so it is with deciding what to watch. In the years before computers and film guides, movie critics were the best informed source of advice. I found the recommendations of John Hartl of the Seattle Times a good match for my tastes. Then Siskel & Ebert came to TV, with the tastes of Siskel the better match for me (I still like his guideline, "Do you care about the characters?"). Maltin's film guide was handy and extensive, but the accuracy for me was mixed, probably because of the team approach that had to be used to review so many movies.

I used to subscribe to the magazines Films in Review and Sight and Sound, but by the time the magazines arrived, the films were usually gone from the theaters. Foreign films were only shown at the Ridgemont in Greenwood and sometimes at a couple places in the University District. The University put on a quarterly film series which concentrated on films from one country each quarter. The only movies shown on TV were old Hollywood ones.

The most reliable referral source for me now is the Netflix algorithm, based on my own ratings compared with those of other members who rate like me. After watching a movie, I use the Netflix filmographies of the director and actors to find other films of interest. I also look at the reader reviews and sometimes the critics reviews (though Ebert rates so many movies good that I suspect he is influenced by the Netflix desire to keep people renting). I still check out any other recommendations I get from Netflix, family, friends and various media sources, by seeing what the Netflix algorithm predicts for me.

My Netflix queue usually has about 200 movies. If a movie is available from the King County Library (which a little over half are) I get it there, so my Library queue has about 250. I tried to maintain this second queue at Netflix, but found it cumbersome, so now I keep a database with both queues and can sort movies by genre, date, length, language and Netflix prediction. I watch 4 or 5 a week each from Netflix (which is making a smart business move during a down economy by working extra hard to keep the movies coming fast) and the library, with the mix designed to go with my mood and to provide some variety.

Susan is my resident movie watching companion, but she works full time and our tastes differ somewhat. She retires at year end, so may become more involved in movie picking after that.

In November I watched 33 movies, dated 1936 to 2009, all dramas except for 3 comedies, 3 romances, 2 documentaries and 1 sci-fi, all in English except 4 Spanish, 2 German and 1 each Hungarian, Italian and Japanese, predicted by Netflix from 3.4 to 2.8 and rated by me from 3.8 to 2. For my mathematically inclined readers (especially John), the Netflix predictions were accurate on the average to within .1, with only 4 being off by more than .5, and the median being an exact match.

My Movie Ratings

Here are my ratings of almost 1,500  movies (including TV shows, etc) I have watched on DVD in the last few years, and rated for my own purposes. References to seasons or disk numbers are not meaningful unless more than one rating is made for that title. Some of the movies include the year of release. Numbers following some titles are Netflix or other predictions of how I would rate the movie, as of the time of viewing. The Netflix star system is designed to make three of the five stars positive ratings (and thus encourage rentals). My reaction to that tactic was to hold my ratings down to mostly three stars (liked it), with a few four stars (really liked it) and a rare 5 star (loved it).

My rating system is built around the Netflix five star system, but I have added decimals to make it more accurate. It essentially has evolved into a four star system. A 2.7 or lower movie is one I would not recommend. Ratings of 2.8 and 2.9 are marginal.  From 3 to 3.2 are all right for those interested. From 3.3 on up are recommended. About 50 of the older ratings were done before I started using decimals and they are identified as "no decimal" and should be taken with a grain of salt, since a 2.7 and 3.3 movie would both have been rated three stars.

I have rated about 850 other films at Netflix, mainly for the purpose of helping the Netflix predictions of what I will like. I don't have a decimal rating record for them. I could make a list here of those which I top rated, but it might be best for me to first watch them again to see how they stand the test of time and then rate them on the decimal system.

Remember, these ratings are my current opinions of what I want Netflix to recommend to me, based on my most recent viewing of the movie, and also on my mood of that day. In years past I may have rated a movie higher, but recent viewing reduced my opinion. I am not rating based on the historical merit of the movie as of the time it was released. I chose movies to watch that I did not remember seeing which I thought would at least be marginally acceptable (2.8). Of the movies listed through 2009, I rated 84% at 2.8 or above. I think the toning down of my ratings at Netflix has enabled the Netflix prediction algorithm, which is calculated to one decimal, to quite accurately guide my choice of what to watch.

My movie ratings lists are now dated December 31, 2010, and are linked to in the blog side bar, both alphabetically and from high to low rated. The lists may be updated in the future. Some of the movies do not have a year listed, because the year was not included in all my lists I used in compiling this list. In the case of a TV series, sometimes the series is rated as a whole and sometimes the individual disks are rated. Direct any requests for clarification to me, Tom Blake.