In which a group of graying eternal amateurs discuss their passions, interests and obsessions, among them: movies, art, politics, evolutionary biology, taxes, writing, computers, these kids these days, and lousy educations.

E-Mail Donald
Demographer, recovering sociologist, and arts buff

E-Mail Fenster
College administrator and arts buff

E-Mail Francis
Architectural historian and arts buff

E-Mail Friedrich
Entrepreneur and arts buff
E-Mail Michael
Media flunky and arts buff

We assume it's OK to quote emailers by name.

Try Advanced Search

  1. Seattle Squeeze: New Urban Living
  2. Checking In
  3. Ben Aronson's Representational Abstractions
  4. Rock is ... Forever?
  5. We Need the Arts: A Sob Story
  6. Form Following (Commercial) Function
  7. Two Humorous Items from the Financial Crisis
  8. Ken Auster of the Kute Kaptions
  9. What Might Representational Painters Paint?
  10. In The Times ...

Sasha Castel
AC Douglas
Out of Lascaux
The Ambler
Modern Art Notes
Cranky Professor
Mike Snider on Poetry
Silliman on Poetry
Felix Salmon
Polly Frost
Polly and Ray's Forum
Stumbling Tongue
Brian's Culture Blog
Banana Oil
Scourge of Modernism
Visible Darkness
Thomas Hobbs
Blog Lodge
Leibman Theory
Goliard Dream
Third Level Digression
Here Inside
My Stupid Dog
W.J. Duquette

Politics, Education, and Economics Blogs
Andrew Sullivan
The Corner at National Review
Steve Sailer
Joanne Jacobs
Natalie Solent
A Libertarian Parent in the Countryside
Rational Parenting
Colby Cosh
View from the Right
Pejman Pundit
God of the Machine
One Good Turn
Liberty Log
Daily Pundit
Catallaxy Files
Greatest Jeneration
Glenn Frazier
Jane Galt
Jim Miller
Limbic Nutrition
Innocents Abroad
Chicago Boyz
James Lileks
Cybrarian at Large
Hello Bloggy!
Setting the World to Rights
Travelling Shoes

Redwood Dragon
The Invisible Hand
Daze Reader
Lynn Sislo
The Fat Guy
Jon Walz


Our Last 50 Referrers

« Elsewhere | Main | Trump U, Too »

May 23, 2005

Weird Netflix Recs

Michael Blowhard writes:

Dear Blowhards --

I'm a completely contented Netflix subscriber. What's not to like? A wide selection ... Prompt service ... A demento suggestions engine ...

Recently The Wife and I have been watching Japanese horror, '60s and '70s erotica, and blues and C&W music documentaries. Based on our rental patterns, here's what Netflix thinks we might want to watch next:

  • "Anne of Green Gables"
  • "Jonah: A Veggie Tales Movie"
  • "Agent Cody Banks"

Question: if you knew that a buddy of yours was into Takashi Miike, Merle Haggard, and Radley Metzger, would you cheerily suggest that he give "Agent Cody Banks" a try?

Hmm. I wonder if Netflix is trying to tell us something ...



posted by Michael at May 23, 2005


I've written a few recommender algorithms in my day, and it's an example of what I call the "Beetles Problem". Because most everyone likes the Beetles.

Every kid learning basic Collaborative Filtering Algorithms writes a music recommender. When you have sparse data (i.e., eclectic/rare taste combinations), most of the default algorithms can't get particular high scores for anything. So, they'll end up picking whatever's most popular.

Interestingly, in your case, most adults who watch Miike don't also get the newest chick flick. But, lots of them have kids. And there are very few 6-year-old foreign film buffs. So, the only thing in common between people who watch the various things you do are that their kids watch Cody Banks.

There are some clever tricks to get around this (Amazon seems to recommend things based on random individual books you've read rather than the combination of books). But it's not easy.

Posted by: Dan G on May 23, 2005 6:46 PM

Besides what they guy above wrote, I would also bet that part of the recommendation weighing can be bought, meaning the makers of veggie tales or "Agent Code-High Banks" can pay to have it end up on everyone's recommendations. When they popped up on mine, I immediately said "no thanks" several times (they push a different religion than mine, plus vegetarianism which I think is unhealthy for children). Yet, they persist in my recommendations, too. So that smells like either a glitch or product placement, and the latter is most likely, imvho.

Posted by: Yahmdallah on May 23, 2005 6:52 PM

Hopefully the kids buying Agent C.B. don't get recommendations for Takashi Miike!

Posted by: Rob Asumendi on May 23, 2005 7:41 PM

Merle rocks. Cody sucks.

Enough said.

Posted by: Cowtown Pattie on May 23, 2005 10:25 PM

Hey, don't diss Anne of Green Gables!

Posted by: lindenen on May 23, 2005 10:33 PM

Let's investigate the possible connections.
Wasn't there some 1970's erotic/neurotica film titled "Behind the Green Door"? "Green Door"? "Green Gables"? Yeah, that's close.

Japanese Horror movies? "..A Veggie Tales Movie". That's close as well. If I remember some of those older Japanese horror movies, monsters and creatures tended to look like organic backyard gardens gone nuclear. Stalking stalks of celery. Vicious looking cumquats. I could see how Netlfix would connect those dots.

"Cody Banks"? That's too easy. Connecting those cowboy names like Merle and Radley to Cody seems (almost) logical.

What I'm surprised at is that Netflix missed on suggesting any John Waters movies based on the unique dvd's you'd already rented.

Posted by: DarkoV on May 24, 2005 10:42 AM

Another "Anne of Green Gables"-Miike connection might be that the
Japanese have a strange obsession with "Anne":

Check out this FAQ:

Posted by: J.W. Hastings on May 24, 2005 4:26 PM

I wonder if Amazon's recommendations are based more on how overstocked they are with a certain item. For example, I ordered Muellers Upgrading and Repairing PCs, 16th edition and the recommendation was to buy the 15th edition as well.

Posted by: ned on May 24, 2005 6:28 PM

Actually, for the media & entertainment buffs amongst us, there's a fascinating story behind Veggie Tales. Not so much for the entertainment value in religious anthropomorphic produce, but in the real-world showbiz background. A man has an idea, actually makes a success out of it -- then what? The company is facing a Grow Bigger or Wither on the Vine decision point, properties like this have a life cycle and Veggie Tales is peaking, but no new ideas are replacing it. Will borrowing a lot of money and producing a Veggie Tales feature film (i.e., Jonah) save the company? Not if the movie is two years too late in the product cycle... Phil Vischer ( has posted his reminiscences and it's a harrowing story of choking on success, well worth reading for its own sake.


Posted by: Dwight Decker on May 24, 2005 6:49 PM

The really funny thing would be if you rented "Agent Cody Banks" and liked it!'d really have to wonder about the magic of Netflix!

Posted by: annette on May 25, 2005 8:59 AM

Dan G gives a good explanation for the kid movie recomendations.

I think Amazon does this much better. Amazon's recommendation algorithm is based on a product level comparison rather than a customer level comparison. For each product they make a list of the related products. The related product lists are pretty good because Amazon has a lot of data at the product level. They then compile the related products lists for the products you purchased to produce your recommendations. If you liked merle haggard and Takashi Miike, maybe you will like Shion Sono and Willie Nelson. This is a pretty simple algorithm.

It also completely avoids the "Beatles Problem". Amazon doesn't check for what other products customers who like both merle haggard and Takashi Miike also like. Who cares? There just aren't enough customers who like both merle haggard and Takashi Miike to get valid data.

Here is an uncharacteristically bad Malcolm Gladwell article from 1999 on collaborative filtering. Gladwell dismisses the Amazon algorithm in favor of the powerful collaborative filtering algorithms that are just around the corner. It is 2005 and they are still just around the corner.

I especially like the quote:

>"What we'd like to see is nice little clusters," Goldberg says. "But, when you look at the results, what you see is something like a cloud with sort of bunches, and nothing that is nicely defined. It's kind of like looking into the night sky. It's very hard to identify the constellations."

Which is to say collaborative filtering doesn't work very well, but Gladwell doesn't pick up on this.

Netflix is a special case. They don't want to just recommend the films that they think you will like. They want to recommend the films that they think you will like and that they have a lot of copies of just sitting around.

They make it hard to rent big blockbuster movies when they first come out. They don't recommend such movies; you have to search for them by name. Netflix wants to keep the number of new movies people rent to a minimum so that they can keep the inventory of each new release as low as possible.

Relatively obscure films cause a similar problem. Criterion collection DVDs don't grow on trees. The wait for some obscure movies can be very long. I suspect Netflix doesn't have very many copies of the more obscure films. I also suspect people who like obscure movies tend to rent a lot of movies and are thus not particularly desirable customers .

Posted by: Joe O on June 2, 2005 4:54 AM

Post a comment

Email Address:



Remember your info?