Listener Questions Special

[Cam Bert 8145]

15 hours ago, ol' eddy wrecks said:

If someone wants to go through and get the list of directors/movies of multiple directors on there, I'll modify this post to include those lists

Robert Altman - MASH, Nashville

Frank Capra - It's a Wonderful Life, Mr Smith Goes to Washington, It Happened One Night

Charlie Chaplin - Modern Times, City Lights, The Gold Rush

Francis Ford Coppola - The Godfather,  The Godfather Part II, Apocalypse Now

Michael Curtiz - Casablanca, Yankee Doodle Dandy

Victor Flemming - Gone With the Wind, The Wizard of Oz

John Ford - The Searchers, The Grapes of Wrath

Alfred Hitchcock - Vertigo, Psycho, Rear Window, North by Northwest

John Huston - The Maltese Falcon, The African Queen, The Treasure of the Sierra Madre

Elia Kazan - On the Waterfront, A Streetcar Named Desire

Stanley Kubrick - 2001, A Clockwork Orange, Dr. Strangelove

David Lean - Bridge of the River Kwai, Lawrence of Arabia

George Lucas - Star Wars, America Graffiti

Sidney Lumet - 12 Angry Men, Network

Mike Nichols - The Graduate, Who's Afraid of Virginia Woolf? 

Alan J. Pakula - Sophie's Choice, All the President's Men

Martin Scorsese - Taxi Driver, Goodfellas, Raging Bull

Stephen Spielberg - Jaws, E.T., Raiders of the Lost Ark, Schindler's List, Saving Private Ryan

George Stevens - Swing Time, Shane

Billy Wilder - The Apartment, Some Like It Hot, Double Indemnity, Sunset Boulevard

William Wyler - Ben-Hur, The Best Years of Our Lives

[ol' eddy wrecks 212]

58 minutes ago, Cam Bert said:

Robert Altman - MASH, Nashville

Frank Capra - It's a Wonderful Life, Mr Smith Goes to Washington, It Happened One Night

Charlie Chaplin - Modern Times, City Lights, The Gold Rush

Francis Ford Coppola - The Godfather,  The Godfather Part II, Apocalypse Now

Michael Curtiz - Casablanca, Yankee Doodle Dandy

Victor Flemming - Gone With the Wind, The Wizard of Oz

John Ford - The Searchers, The Grapes of Wrath

Alfred Hitchcock - Vertigo, Psycho, Rear Window, North by Northwest

John Huston - The Maltese Falcon, The African Queen, The Treasure of the Sierra Madre

Elia Kazan - On the Waterfront, A Streetcar Named Desire

Stanley Kubrick - 2001, A Clockwork Orange, Dr. Strangelove

David Lean - Bridge of the River Kwai, Lawrence of Arabia

George Lucas - Star Wars, America Graffiti

Sidney Lumet - 12 Angry Men, Network

Mike Nichols - The Graduate, Who's Afraid of Virginia Woolf? 

Alan J. Pakula - Sophie's Choice, All the President's Men

Martin Scorsese - Taxi Driver, Goodfellas, Raging Bull

Stephen Spielberg - Jaws, E.T., Raiders of the Lost Ark, Schindler's List, Saving Private Ryan

George Stevens - Swing Time, Shane

Billy Wilder - The Apartment, Some Like It Hot, Double Indemnity, Sunset Boulevard

William Wyler - Ben-Hur, The Best Years of Our Lives

Thanks, updated the post.  You missed Spartacus for Kubrick - though probably not a common choice for most people.

[ol' eddy wrecks 212]

I guess I'll go first

How did you two first meet? Origin story of show? What made you think of doing this podcast?

Changing this to how I first heard about this podcast, why I started listening to it - from listening to HDTGM. 

How much time do you spend researching the movie you're watching?

I put together that post of how a movie does in other lists for comparison.  When we hit movies that I'm not sure why it's considered good, I'll try googling something, but recently I've taken to the habit to waiting until after the episode is over before watching so I know what topics of conversation are going to come up.  That way I can keep an eye out for them.  Also, sometimes the episode does the legwork for answering, "why do people think this is a good movie."  The two that swayed me the most was Paul arguing for Shawshank made me realize it's really not out of the conversation compared to Cool Hand Luke, which I wouldn't have thought twice about being on the movie.  And the other was E.T. which I still feel only has one good scene, but I literally didn't get a sense of what the character/thematic arc was of the movie while watching it.  

These movies are hard to find? Where do you go?

Amazon, filmstruck, netflix, (hulu if needed), iTunes, personal collection

If you could get everyone to watch one movie what would it be?

My tastes aren't everyone's tastes, so I don't know if I have an answer to this question.  Is there a movie that is important to see if you don't even enjoy it? Shoah? The Act of Killing?

2001 is my favorite movie, but it's not for everyone, and if you don't like it, it'll be a painful watch.  I think 8 1/2 is a fun movie, but I'm sure I'd meet people who'd hate it.

I guess for movies that I feel are underseen gems I'd like more people to see: Funeral Parade of Roses, The Trial, and I guess since we were talking about movies made by women, Girlfriends.

If you could appear in any movie on the list, which one would you?

I took this question to mean, as an actor (since Paul is an actor), but they answered it as being in that world.  I'm not a Raiders-person, but Nazis are a lot less scary in that world and they ended up dying.  I haven't seen the Wizard of Oz, but I guess I'll choose that one.  I wish Japanese movies were allowed so My Neighbor Totoro was on the list.  That way I could get a fantastic world in which I was not in danger of dying.

Will you ever do an episode comparing the 97 list?

I can't interpret this question to be relevant for us, other than, I'd like a Fargo episode.

Why haven't they updated it yet?

This question didn't seem relevant to Amy & Paul either. I do wonder if the Me-Too movement actually disrupted things enough to delay the list (e.g. did it potentially affect who should be receiving ballots to vote).

How many of the films are film capsule entries?

I don't know.  I am interested in the mentality of why some of these movies were voted for (but even then, who's voting for a movie isn't a monolithic constituency). 

If directors could only have one movie on the list, which one (for each director) would you choose?

• Robert Altman - MASH, Nashville 

My favorite Altman, McCabe & Mrs. Miller isn't on the list, and MASH has its problems.  I just think Nashville is the better film.

• Frank Capra - It's a Wonderful Life, Mr Smith Goes to Washington, It Happened One Night 

I haven't seen Mr. Smith Goes to Washington.  I'll go for the movie that starts off with the guy contemplating suicide.

• Charlie Chaplin - Modern Times, City Lights, The Gold Rush

I've seen one, maybe two of these and am unopinionated

• Francis Ford Coppola - The Godfather,  The Godfather Part II, Apocalypse Now

I just like it as a movie more

• Michael Curtiz - Casablanca, Yankee Doodle Dandy

I know I've seen YDD, but can't remember it.  Outside of this podcast, I'd assume I would never be revisiting it.  I also enjoy noir movies.

• Victor Flemming - Gone With the Wind, The Wizard of Oz

I haven't seen either!

• John Ford - The Searchers, The Grapes of Wrath

The Searchers.  The ending has the subvert the typical western narrative quality and scenes of the vistas as legitimately impressive.  It still suffers from tropes and racist depictions that were common for the era though.  I haven't seen GoW since high school.  We'll see if I change my mind.

• Alfred Hitchcock - Vertigo, Psycho, Rear Window, North by Northwest

If I'm going with the one I like best, Psycho.  Most representative, Vertigo, I guess.

• John Huston - The Maltese Falcon, The African Queen, The Treasure of the Sierra Madre

I think I've only seen The African Queen

• Elia Kazan - On the Waterfront, A Streetcar Named Desire

It's been a while.  I could have gone either way.

• Stanley Kubrick - 2001, A Clockwork Orange, Dr. Strangelove, Spartacus

Too many words.

• David Lean - Bridge of the River Kwai, Lawrence of Arabia

Not picking until I see BotRK and rewatch LoA.

• George Lucas - Star Wars, America Graffiti

One like best: American Graffiti.  Most representative/makes most sense to be on the list: Star Wars

• Sidney Lumet - 12 Angry Men, Network
• Mike Nichols - The Graduate, Who's Afraid of Virginia Woolf? 

I need to see Who's Afraid of Virginia Woolf?

• Alan J. Pakula - Sophie's Choice, All the President's Men

I need to see both of these

• Martin Scorsese - Taxi Driver, Goodfellas, Raging Bull

Taxi Driver for now.  I need to rewatch Raging Bull.  I never got the love for Goodfellas.

• Stephen Spielberg - Jaws, E.T., Raiders of the Lost Ark, Schindler's List, Saving Private Ryan

Favorite: Jaws.  Best representative: E.T.  Haven't seen SPR, but Jaws is the only Spielberg movie on the list I like.

• George Stevens - Swing Time, Shane

Haven't seen either.

• Billy Wilder - The Apartment, Some Like It Hot, Double Indemnity, Sunset Boulevard

Loved Sunset.  Really like The Apartment.  Double Indemnity next level down.  I enjoyed SLiH, but don't feel the love for it.

• William Wyler - Ben-Hur, The Best Years of Our Lives

Haven't seen either

How do you think we should be viewing this list? eg Historical, Innovations, fun, etc

As a list of movies that people, who are members of the AFI, when asked to select 100 movies from a curated list of 400 that they think are the greatest American movies of all time, these were the most common selections.  I am very literal on this.  The result, a list that includes enough classic-era (1950s and before) Hollywood movies that I haven't seen that I should probably watch those to fill in some gap of knowledge and viewing.

Which movie are you most looking forward to and why?

Do the Right Thing.  Spike Lee is an embarrassing blind-spot for me, so it'll give me a chance to correct that at least a little.

There's a lack of female directors and movies from directors of people of color. A lot have been made recently, but they wouldn't have been eligible for the 2007 list because they've come out since then. Are there films that would be eligible for the list you are doing now that you would put on?

We talked about this last week.  I guess Killer of Sheep was the answer that I've seen.  I need to see Wanda.

A comedy isn't really represented in the AFI top 100. What are some of your favorite comedies? (Not necessarily for the AFI top 100)

Again, a topic we talked about recently. Arguably, my favorite comedy, Dr Strangelove, is on there.  After that, I might say, Monty Python's Life of Brian (but that isn't considered American, is it?).  I like dark comedies, and it's crossing my mind, Brazil (again, not American, right?), Barton Fink and The Man Who Wasn't There are ones I really enjoy.  Looking at other people's answers, I keep forgetting Christopher Guest on these type of questions (This is Spinal Tap and Best in Show).

Related, what animated or documentary films would you want on the list?

For Errol Morris, I feel like people would gravitate towards The Thin Blue Line.  I haven't seen Hoop Dreams, but people talk about it a lot.  One of my documentary choices, Shoah, isn't American and still wouldn't be eligible for this list.

When thinking of American animated films, I draw a blank on well known ones that I love.  Trying to think of one, The Plague Dogs, comes to mind, it's not iconic, well-seen (I guess Watership Down would be the more appropriate film for those categories and even then), or particularly something I'd think of as a general best-of list.  If we could include international movies, it could turn into a Studio Ghibli-discussion, but we aren't.  It'd be funny if the animated version of The Hobbit was on the list right next to The Fellowship of the Ring.

Edited (again!): Alice in Wonderland seems like an answer I could live with and appreciate the oddness of its world.

What HDTGM movie would you place on this list?

I think Paul's answer is the correct one.

Will you just own up and admit that you think Scorsese is tremendously overrated?

Since I like Taxi Driver and Raging Bull, I wouldn't think so, but everything of his outside of that that I've seen, I've been less enthralled by (namely Goodfellas).  I guess I'll point out, as others have, he did a lot more than toxic masculinity movies with Robert DeNiro.  He just did that genre amazingly well and it's stuck in our minds.  I still need to see Silence and that's one that I might really like.  Might.

Would you agree that 1939 was the greatest year for films? If not, which year?

I haven't given this one that much thought and this Q&A is longer than I had originally hoped when started.  I'll come back to this later.

What are your favorite books about movies?

I haven't read any.  I'm curious to read Hitchcock-Truffaut and there is a book that's a compilation of interviews with Kubrick I'd like to read.

Amy and Paul what is the best theater food snack food or drink?

I don't know about best, but I hate the smell of popcorn, so I'll say that's the worst.  I usually just need soda/caffeine while watching a movie and not much else.

100 years from now, someone on a ghost tour says you are haunting a location. What location would you want them to say you're haunting?

Probably a hotel.  I'd hate to get lonely for eternity.  If it weren't for that, a university building could be cool.  That way, as things change, you could still just sit there and keep learning.  Though if you keep getting an intro math/science course year after year, the repetition of the same material/knowledge would get old fast.

ETA: Updated animated and comedies answers because re-reading my post, I realized, I stopped writing out them out mid-thought.

[ol' eddy wrecks 212]

I would advise not answering all of those questions in one sitting.  That took long enough that it started to become unenjoyable.

[Cameron H. 23786]

On 11/8/2018 at 10:25 PM, Cameron H. said:

That’s cool. I didn’t really want to pollute our hive mind  

I agree. I was a bit resistant to the idea at first, but when Paul brought up E.T. as being quintessential Spielberg, it turned me around. It might not be the “best” Spielberg movie ever, but it is the best in terms of getting across his reoccurring themes. And I feel like most artists, even when working across genres, have certain themes that appeal to them, so maybe we should just present one movie as the Platonic Ideal of what the director has to offer. (i.e. “We know Spielberg is a great director with many great films, but when we think of Spielberg, what movie best sums him up?”)

I mean, I am kind of annoyed that Godfather 1 and 2 are both on the list. I love both of them, but it’s like, “Yeah, we get it...”

Okay, I’ve thought about this some more and I would like to re-adjust my stance on this (again). I now disagree with Paul (and my previous post) because by only allowing one movie per director that suggests that the director, and no one else, is responsible for a movie’s quality - and that’s bullshit. I mean, it would be one thing if every director worked with the same cast, crew, and scriptwriter on every film, but since that’s not the case, disallowing a film because it shares the same director ignores the contributions of everyone else who worked on that film. For example, even if all the other elements were the same, I doubt Raiders would be on the list if it starred Gary Busey or Rear Window if it were written by E.L. James.

[bleary 333]

1 hour ago, Cameron H. said:

Okay, I’ve thought about this some more and I would like to re-adjust my stance on this (again). I now disagree with Paul (and my previous post) because by only allowing one movie per director that suggests that the director, and no one else, is responsible for a movie’s quality - and that’s bullshit. I mean, it would be one thing if every director worked with the same cast, crew, and scriptwriter on every film, but since that’s not the case, disallowing a film because it shares the same director ignores the contributions of everyone else who worked on that film. For example, even if all the other elements were the same, I doubt Raiders would be on the list if it starred Gary Busey or Rear Window if it were written by E.L. James.

Definitely agree.  I think there's a happy medium though, because some items on the list do feel unnecessary.  As you mentioned, I don't think Godfather 2 needs to be there.  As I mentioned earlier, it seems like a waste of a spot or two to have City Lights and The Gold Rush and Modern Times all representing Chaplin.  Similarly, do we really need both of Duck Soup and A Night At the Opera?  Do we need both The African Queen and Treasure of the Sierra Madre, for that matter?

But none of Spielberg's films on the list feel superfluous.  Some of them might not belong on the list for quality reasons, but I certainly wouldn't say he's retreading the same ground, at least in these five films.

[AlmostAGhost 2718]

How did you two first meet? Origin story of show? What made you think of doing this podcast?

How Did This Get Made and just a general Earwolf fan. It does match my level of film fan-ness pretty much to a tee though: I know most of these, but haven't seen them (or if I have, it's been years). It came along just as I was getting back into movies again. 

How much time do you spend researching the movie you're watching?

I might glance through Wikipedia for some background or whatever, but that's about it.

These movies are hard to find? Where do you go?

iTunes, streaming apps I belong to, torrents

If you could get everyone to watch one movie what would it be?

Rushmore

If you could appear in any movie on the list, which one would you?

Lord of the Rings

Will you ever do an episode comparing the 97 list?

I'll trump that: I hope Paul & Amy expand this podcast for 300 more weeks, and go through the whole 400-movie ballot.

Why haven't they updated it yet?

Because putting out a list with 98 white male directors would be pretty appalling in 2018.

How many of the films are film capsule entries?

All of them?  Isn't that the point?

If directors could only have one movie on the list, which one (for each director) would you choose?

[can't answer this one: I am waiting to judge movies along with the episodes or more current viewings]

How do you think we should be viewing this list? eg Historical, Innovations, fun, etc

Whatever. Every movie has its own valid reasons.

Which movie are you most looking forward to and why?

Snow White. I've been interested in, and just started, going through Disney movies because it's a side of culture I'm not overly familiar with, and its the first one, but I'm waiting for Unspooled to hit on it. Which sort of bugs me -- I tend to be a nerdy sort of completist and prefer to do things in order if I can. I finally gave up waiting and just skipped Snow White, but it bugs me that I did.

There's a lack of female directors and movies from directors of people of color. A lot have been made recently, but they wouldn't have been eligible for the 2007 list because they've come out since then. Are there films that would be eligible for the list you are doing now that you would put on?

I don't know. Moonlight, maybe. I echo the call for more fun lighter fare: a rom-com or two, some comedies, Clueless. Those aren't necessarily recent choices though. I haven't really thought about it. This podcast is not at all about what isn't on the list to me. It's about these 100 movies, period. Would I put like 7 Coen Brothers movies on there if I could?  Sure. But that's irrelevant. 

A comedy isn't really represented in the AFI top 100. What are some of your favorite comedies? (Not necessarily for the AFI top 100)

Rushmore, Top Secret!, Spinal Tap, The Big Lebowski, Pee-Wee's Big Adventure, UHF, Dirty Rotten Scoundrels.

Related, what animated or documentary films would you want on the list?

Fantasia is the only one I've seen recently I'd hold close to that high in regard (but that's a relatively small sample). Fantastic Mr. Fox is an all-time fave too, but I don't know if it's close to this list really.

What HDTGM movie would you place on this list?

none, or, Teen Witch

Will you just own up and admit that you think Scorsese is tremendously overrated?

Sure! Though I think Taxi Driver is great, and so is Age Of Innocence and Alice Doesn't Live Here Anymore. His 'gangster' stuff, for sure, is.

Would you agree that 1939 was the greatest year for films? If not, which year?

I just watched Top Secret! again yesterday, and I'm with Paul: it's gotta be 1984.

What are your favorite books about movies?

do not care about Hollywood enough in that way. I like critics and evaluating culture, but don't particularly care about filmmaking.

Amy and Paul what is the best theater food snack food or drink?

Diet Coke and Reese's Pieces are my go-tos

100 years from now, someone on a ghost tour says you are haunting a location. What location would you want them to say you're haunting?

Paisley Park because I'd be hanging out with Prince and Vanity in heaven

[ol' eddy wrecks 212]

On 11/9/2018 at 7:30 PM, bleary said:

You've got 7 more films to rank!  The method I used requires a value for each of the 25 films (since it's essentially just computing the Euclidean distance in a 25-dimensional vector space), but I could get numbers by plugging in 19 for each of the missing films.  But that wouldn't be very consistent with the rest of the scores.

I think you mean square of the euclidean distance. 

Though I'm trying to interpret it in my brain what the dimensions would be in a way that could be extractable. If we all actually rated the movies, you could have worked as the difference of ratings as each dimension of distance (and then divided for the average for incomplete lists). If each dimension is the distance in the movie rankings... then that would mean the value of each dimension would also be dependent upon the other dimensions. Abstractly speaking I am finding that thing a bit difficult to conceptualize at the moment. Oh well.

[bleary 333]

3 minutes ago, ol' eddy wrecks said:

I think you mean square of the euclidean distance. 

Sure, and this is what I swept under the rug when I used the word "essentially," since I didn't feel the need to mention that the monotonicity of the square root function ensures that order will be preserved between the quantity calculated and the actual Euclidean distance, and that the only downside of not taking the square root is that the triangle inequality is not necessarily satisfied, so that we don't have a proper metric space under the quantity calculated, but that's not a big deal for the purpose it was used for.

9 minutes ago, ol' eddy wrecks said:

Though I'm trying to interpret it in my brain what the dimensions would be in a way that could be extractable.

The way I did it, the dimensions are the 25 films themselves, and a person's ranking is the component in that direction.  (So there doesn't need to be a 1st dimension, 2nd dimension, etc., but rather the Titanic dimension, the Sixth Sense dimension, etc.)  I alphabetized just to standardize, but one could easily just take the films in the order the episodes were released, in which case the vector <1, 2, 3, 4, ..., 25> would be a list in which Citizen Kane is ranked #1, Ben-Hur is ranked #2, and every subsequent episode's film took the next place on the list.  Under this system, there's nothing good to put in for an unranked film that would give a consistent result.  If I put in 0 for all unranked films, that would score them as being high on the list rather than not on the list at all.  Putting in a high number for unranked films is fairer, but still won't be comparable to other scores.

Granted, my background is in analysis rather than statistics, which is why I took an analytical approach rather than perhaps a statistical approach.  Someone more skilled in statistics than me could probably tell me a better way for computing similarity scores on lists.

Anyway, I didn't think anyone would care about the details, so I didn't get into it before.  But I'm glad to talk about it since you seem interested!

[ol' eddy wrecks 212]

2 hours ago, bleary said:

Sure, and this is what I swept under the rug when I used the word "essentially," since I didn't feel the need to mention that the monotonicity of the square root function ensures that order will be preserved between the quantity calculated and the actual Euclidean distance, and that the only downside of not taking the square root is that the triangle inequality is not necessarily satisfied, so that we don't have a proper metric space under the quantity calculated, but that's not a big deal for the purpose it was used for.

And I said, ""!

Also, my mind just glazed over when you said, "essentially."

Quote

 

The way I did it, the dimensions are the 25 films themselves, and a person's ranking is the component in that direction.  (So there doesn't need to be a 1st dimension, 2nd dimension, etc., but rather the Titanic dimension, the Sixth Sense dimension, etc.)  I alphabetized just to standardize, but one could easily just take the films in the order the episodes were released, in which case the vector <1, 2, 3, 4, ..., 25> would be a list in which Citizen Kane is ranked #1, Ben-Hur is ranked #2, and every subsequent episode's film took the next place on the list.  Under this system, there's nothing good to put in for an unranked film that would give a consistent result.  If I put in 0 for all unranked films, that would score them as being high on the list rather than not on the list at all.  Putting in a high number for unranked films is fairer, but still won't be comparable to other scores.

Granted, my background is in analysis rather than statistics, which is why I took an analytical approach rather than perhaps a statistical approach.  Someone more skilled in statistics than me could probably tell me a better way for computing similarity scores on lists.

Anyway, I didn't think anyone would care about the details, so I didn't get into it before.  But I'm glad to talk about it since you seem interested!

 

Unfortunately I haven't taken a stat (non-calc based) since high school.  I think I figured out what I was having trouble with.  Trying to conceptualize it as a vector space gives it weird properties that make common things we do with vector spaces (in this case, projections)... wonky.

So to back up, when I said what are the dimensions, that was more focused on what the values were (the, use the ratings as our dimensions, still had each movie serving as what each dimension was, just the value in each dimension was the star rating for the movie) and it was getting conflated with the assigned metric/inner product.  But to get to the wonky example, if we flip the values around, make the top rated movie of 25, value of 24, and the lowest rated movie, value of 0 (we could also do 25 and 1), while it shouldn't change anyone's computed squared-distance, it's just easier for me to think about for thought experiment purposes because of associating higher values with better movies for these things.  Let's hypothetical ranker Kevin loves Citizen Kane and ranking it number 1, hates Titanic (always ranking it last), and let's say those were the first two movies reviewed.  Also let's say there's a hypothetical ranker, Nivek, who's the reverse - loves Titanic, hates Citizen Kane, but for every other movie they have the same opinion and rank.

So after week three, Kevin's vector is <2, 0, 1> and Nivek's vector is <0, 2, 1>.  After 25 weeks, Kevin's vector is <24, 0, a, b, c...> and Nivek's vector is <0, 24, a, b, c...>.

The squared-distance between them at week 3 is 8.  At week 25, it is 2*24^2, even though all of the new dimensions are contributing 0 to this increased distance.

Similarly, let's say users Ke and Ni come around, but they've both only seen the first three movies.  And their vectors are, respectively, <2, 0, 1> and <0, 2, 1>.  We want to project Kevin and Nivek's vectors down to 3-dimensional space from 25 dimensional space.  Instead of projecting <24, 0, a, b, c...> and <0, 24, a, b, c....> down to <24, 0, a> and <0, 24, a>, it seems like we should project them down to <2, 0, 1> and <0, 2, 1>.  And maybe that's the best way to compare for subsets, though simply taking the average squared-distance by dimension would be misleading, as above demonstrates, the squared distances increases at a rate of n^2 (where n = number of dimensions).

That's what I mean by, that is wonky in terms of what I normally expect for a vector space/inner product space.  And I focused on a projection because, well, trying to compare those of us with fewer rankings, it should be useful, even if it's incomplete data, to calculate what the distance is when projecting your rating vector point down into a lower dimensional space.  Granted, I say that as someone who's only ever completed an undergraduate degree (which I guess I should say, linear projection, because as I type that out, an inverse stereographic projection would be one way to go from a subset of 3 dimensions to 2 dimensions which doesn't uphold the 'typical' linear projection behavior; though my gut reaction the inverse of a stereographic projection isn't an isometry, so talking about distances might not be the best idea.  Not that I've seen talk about stereographic projections in a linear algebra class).

 

[bleary 333]

6 minutes ago, ol' eddy wrecks said:

And I said, ""!

Also, my mind just glazed over when you said, "essentially."

Unfortunately I haven't taken a stat (non-calc based) since high school.  I think I figured out what I was having trouble with.  Trying to conceptualize it as a vector space gives it weird properties that make common things we do with vector spaces (in this case, projections)... wonky.

So to back up, when I said what are the dimensions, that was more focused on what the values were (the, use the ratings as our dimensions, still had each movie serving as what each dimension was, just the value in each dimension was the star rating for the movie) and it was getting conflated with the assigned metric/inner product.  But to get to the wonky example, if we flip the values around, make the top rated movie of 25, value of 24, and the lowest rated movie, value of 0 (we could also do 25 and 1), while it shouldn't change anyone's computed squared-distance, it's just easier for me to think about for thought experiment purposes because of associating higher values with better movies for these things.  Let's hypothetical ranker Kevin loves Citizen Kane and ranking it number 1, hates Titanic (always ranking it last), and let's say those were the first two movies reviewed.  Also let's say there's a hypothetical ranker, Nivek, who's the reverse - loves Titanic, hates Citizen Kane, but for every other movie they have the same opinion and rank.

So after week three, Kevin's vector is <2, 0, 1> and Nivek's vector is <0, 2, 1>.  After 25 weeks, Kevin's vector is <24, 0, a, b, c...> and Nivek's vector is <0, 24, a, b, c...>.

The squared-distance between them at week 3 is 8.  At week 25, it is 2*24^2, even though all of the new dimensions are contributing 0 to this increased distance.

Similarly, let's say users Ke and Ni come around, but they've both only seen the first three movies.  And their vectors are, respectively, <2, 0, 1> and <0, 2, 1>.  We want to project Kevin and Nivek's vectors down to 3-dimensional space from 25 dimensional space.  Instead of projecting <24, 0, a, b, c...> and <0, 24, a, b, c....> down to <24, 0, a> and <0, 24, a>, it seems like we should project them down to <2, 0, 1> and <0, 2, 1>.  And maybe that's the best way to compare for subsets, though simply taking the average squared-distance by dimension would be misleading, as above demonstrates, the squared distances increases at a rate of n^2 (where n = number of dimensions).

That's what I mean by, that is wonky in terms of what I normally expect for a vector space/inner product space.  And I focused on a projection because, well, trying to compare those of us with fewer rankings, it should be useful, even if it's incomplete data, to calculate what the distance is when projecting your rating vector point down into a lower dimensional space.  Granted, I say that as someone who's only ever completed an undergraduate degree (which I guess I should say, linear projection, because as I type that out, an inverse stereographic projection would be one way to go from a subset of 3 dimensions to 2 dimensions which doesn't uphold the 'typical' linear projection behavior; though my gut reaction the inverse of a stereographic projection isn't an isometry, so talking about distances might not be the best idea.  Not that I've seen talk about stereographic projections in a linear algebra class).

 

Ah, I get what you're saying, instead of viewing an incomplete ranking in the 25-dimensional space, just take the projections of everyone else's rankings down to the lower dimensional space.  Yeah, that would work well to give comparisons to other lists for that incomplete list.  However, the scores between two 25-dimensional vectors aren't well-comparable to the scores between lower dimensional vectors, since eliminating films from consideration will decrease the squared-distance.  More importantly to me, the spreadsheet I whipped up in 5 minutes doesn't easily give me the capability to consider those projections. 

[tomspanks 9039]

@bleary all that and you still can’t use my incomplete list? Shenanigans!

[grudlian. 4725]

10 hours ago, tomspanks said:

@bleary all that and you still can’t use my incomplete list? Shenanigans!

This also confirmed everyone here is way smarter than I am. 

[ol' eddy wrecks 212]

19 hours ago, tomspanks said:

@bleary all that and you still can’t use my incomplete list? Shenanigans!

Well, it was all getting to the root challenge of why it's hard to use partial lists.

[ol' eddy wrecks 212]

20 hours ago, bleary said:

Ah, I get what you're saying, instead of viewing an incomplete ranking in the 25-dimensional space, just take the projections of everyone else's rankings down to the lower dimensional space.  Yeah, that would work well to give comparisons to other lists for that incomplete list.  However, the scores between two 25-dimensional vectors aren't well-comparable to the scores between lower dimensional vectors, since eliminating films from consideration will decrease the squared-distance.  More importantly to me, the spreadsheet I whipped up in 5 minutes doesn't easily give me the capability to consider those projections. 

So, thinking about it, maybe the solution would be scale/stretch out everybody's current rankings out to a scale of 100 (and there will just be numeric gaps). That way as more movies get filled in, one can also more closely compare if tastes start converging or diverging as they get to the 50th and 75th episode. And then take the square root and divide by number of entries. I should double-check the formula for standard deviation or variance to see how stats handles something vaguely similar to this.

I'll admit this is all overthinking what's basically trying to come up with a heuristic on just seeing who has similar/different tastes to who, that ignores the details of why we like what we do.

And it'd take more than five minutes, but if your spreadsheet is in Google sheets, you can always write custom functions in javascript. That is, if you deem it worth the effort, which I'd get that unless one got an itch of curiosity, it probably isn't. 

[GammaDev 14]

On 11/10/2018 at 3:22 PM, AlmostAGhost said:

 

Would you agree that 1939 was the greatest year for films? If not, which year?

I just watched Top Secret! again yesterday, and I'm with Paul: it's gotta be 1984.

Sorry to disagree with Paul about 1984 (and we won't even talk about 1999) but he was so close yet 1982 blows away 1984 in terms of movie quality and historical significance.  1984 has Star Trek III?  Well 1982 has the best Star Trek- The Wrath of Khan.  1984 has cult classics like The Last Starfighter and Cloak and Dagger?  Well 1982 has Tron, Blade Runner, and The Dark Crystal- films that were studied and mimicked for decades and which (even though they flopped) earned sequels decades later.  Amy's favorite teen comedy, Fast Times at Ridgemont High came out in 1982.  Also, John Carpenter's best movie (and perhaps the best horror film of all time) The Thing also came out in 1982- and it happens to be the only JC film that he can watch and enjoy without picking apart its flaws.  Eddie Murphy had his breakout film role in 48 Hrs in 1982.  Arnold Schwarzenegger went from bodybuilder to major action star with Conan the Barbarian.  Stallone proved he too could be an action hero, kicked of his biggest franchise, and that he could do more than just get hit in the face in First Blood.  Jason finally got his signature hockey mask look in Friday the 13th Part III.  Animation also saw the historic Secret of NIMH which gave Disney a well-needed kick in the pants and probably inspired animators into saving feature animation from extinction.  The best PG horror film, Poltergeist and the best anthology horror movie, Creepshow.  Meryl  Streep starred in her most iconic role (and gave the world an expression for an impossible situation) in Sophie's Choice.  Dustin Hoffman and Julie Andrews both got Oscar nominations for cross-dressing in Tootsie and Victor/Victoria.  The historic movies in 1982 just go on and on but let's not forget the all-time box office champ for over a decade and Paul's personal choice for Spielberg's best movie- E.T. was the box office juggernaut in 1982.  So next to all that, 1984 doesn't seem nearly as significant, does it?

[AlmostAGhost 2718]

Hmm this is a fruitless debate, but we can have it.  I find it interesting you mention Schwarzenegger and Eddie Murphy's 1982 films, which are both cool, but their 1984 films are both better and more iconic (The Terminator and Beverly Hills Cop).  Comedies were much better in 1984 (Ghostbusters, Top Secret, Spinal Tap, to name just 3). 1984 also had Karate Kid and Gremlins which are still culturally relevant.  Also, Purple Rain and Stop Making Sense!  For horror, I think 1984's Nightmare On Elm Street > 1982's The Thing.

[ol' eddy wrecks 212]

So I still don't really have an answer for the year question, but ordering TSPDT top 1000 by year and just scanning to see if any years stood out to me, in terms of famous movies, 1960...

• Breathless
• Psycho
• La Dolce Vita
• L'Avventura

There's others to help the argument for 1960 (The Apartment, Spartacus - which I'm not the biggest fan of, Peeping Tom, Shoot the Piano Player), but just seeing those four together in the same year just seemed kind of crazy.  Though now that I'm looking closer, 1959 also had a slew pretty great movies...