The rock/paper/scissors paradox: ranking A>B and B>C but C>A ?

Introduce yourself to the community or chat with other users about whatever is on your mind
livelove
Posts: 759
67 Ratings
Your TCI: na
Joined: Thu Sep 17, 2015 9:36 pm

Re: voting: A>B and B>C but C>A ?

Post by livelove »

CMonster wrote:An option might be less grainularity in your rankings
agreed. But then you more easily ran into the issue of having 2 movies, of which you clearly prefer one to the other, but have to give them the same ranking, because you don't have enough granularity. How do you deal with that?

livelove
Posts: 759
67 Ratings
Your TCI: na
Joined: Thu Sep 17, 2015 9:36 pm

Re: voting: A>B and B>C but C>A ?

Post by livelove »

Anomaly1 wrote:Yeah things like this can happen. It's an inevitable outcome of trying to quantize emotional response - there's no 1:1 correspondence between integers and feelings, we have to wing it a bit sometimes.
I think I have to agree.

Anomaly1 wrote:After a few mass re-rankings early on I found it too much of a hassle and stopped letting it bother me.
I wish I could. But it makes me uneasy when I perfectly know that at least 1 of the 3 rankings (for movies A, B and C) must be wrong and I just have not figured out yet, which one. I know there is an error that needs to be corrected. There is a part of me, which likes to find the error and correct it.

livelove
Posts: 759
67 Ratings
Your TCI: na
Joined: Thu Sep 17, 2015 9:36 pm

Re: voting: A>B and B>C but C>A ?

Post by livelove »

Luna6ix wrote:If you score based upon a variety of factors and don't have a weight system in place that truly represents your preferences of those factors, then that'll happen. It happens to me too, I'm sure, but I don't usually compare more than two movies with the same score when looking at my total rankings.
I agree. The problem must be a lack of "weight system". Some kind of gauge or reference value. What is your "weight system"? How do you approach this issue?

livelove
Posts: 759
67 Ratings
Your TCI: na
Joined: Thu Sep 17, 2015 9:36 pm

Re: voting: A>B and B>C but C>A ?

Post by livelove »

Hawkins wrote:Sure, plenty of movies work like rock/paper/scissors. 2001 > Star Wars > Leprechaun 4 > 2001
Hm, I a bit torn. Maybe it's like you say.
OR this occurs only if one the movies is over/underrated.
When I see something like 2001 > Star Wars > Leprechaun 4 > 2001 occuring in my movies, I am intrigued to find out, which one of those 3 movies is over/underrated, similar to what CosmicMonkey said:

CosmicMonkey wrote:Yeah, this happens to me all the time. I used to spend hours fiddling around with my rankings until it "made sense"


CosmicMonkey wrote:watch the far superior Iron Fist and realize that that was in fact, the greatest achievement the genre would ever see
Could you please post the IMDB link. I am unsure which Iron Fist movie you are referencing.

CosmicMonkey
Posts: 594
1271 Ratings
Your TCI: na
Joined: Sun Apr 05, 2009 6:52 pm

Re: voting: A>B and B>C but C>A ?

Post by CosmicMonkey »

livelove wrote:
CosmicMonkey wrote:watch the far superior Iron Fist and realize that that was in fact, the greatest achievement the genre would ever see
Could you please post the IMDB link. I am unsure which Iron Fist movie you are referencing.



I was jokingly referring to the Iron Fist Netflix series, which is actually one of the worst comic book adaptations that I have ever seen. Apparently season 2 is a significant improvement, but if it getting to it means I have to force myself to sit through the entire first season, I'll gladly pass.

livelove
Posts: 759
67 Ratings
Your TCI: na
Joined: Thu Sep 17, 2015 9:36 pm

Re: rock/paper/scissors: voting A>B and B>C but C>A ?

Post by livelove »

lol, thanks for clearing that up.
no wonder I got no meaningful results.
I hope you are happy that you made me search like a dumbo :D :D

spacefloh
Posts: 25
2565 Ratings
Your TCI: na
Joined: Sun Nov 04, 2007 6:33 pm

Re: voting: A>B and B>C but C>A ?

Post by spacefloh »

First I'll answer the second question:

livelove wrote:
spacefloh wrote:Also i want to mention, as you probably assume, that movies are watched from differnt angles. Is it cool, spectacular, overwelming, is it rewatchable, is it a "classic one" for you? All that matter - probably also depending from one own's situation.
However, my solution was to build a "raster" of referenceable, prototype-movies that I will remember for a long time. And rate new ones based on that raster. Thus, the feeling of rating correctly (within slight errors) will stay within that raster.
Thank you. Could you elaborate a bit on the last part? What prototype-movies did you pick?
And if your opinion on those changes, doesn't that then break your system?

It might even introduce imperceptible errors. E.g. you pick "Requiem for a dream" as a gauge for 8 (out of 10). Then you rate another movie 9 because you slightly prefer it by comparison. Now, 5 years later, you like "Requiem for a dream" less than you did 5 years ago, although you might not notice. Let's say upon rewatching you would rank it around 6 or so – while it is still an 8 in your system. If you now compare a movie to it and rank it 9 by comparison, you might actually like it less than the other movie ranked at 9, because your reference movie has lost some appeal to you. (if you get what I'm trying to say)

The prototype movies is a mix of movies that I watched and where I'm sure that their rating does not change over time for me. You can see them on my profile (in ger language), so I recommend a ger->eng translation if you need it. And since my opinion does not really change for those, no break of system [example comparison, i fix the north and southpole as prototype reference points, and i am sure that everything else on earth is somewhere inbetween].

Sorry, I probably missed the thing you wanting to say.
But let me explain an answer from my understanding to that question:

first, I don't (or extremely rare will) rerank movies due to:
livelove wrote:
spacefloh wrote:Happens. In my experience. I would assume that this might depend on the situation one is watching the specific movie (and personal time,...). As someone grows experience (even from movie to movie), opinions can change. So i think that not changing the rating is a valid option, because it's correct at the time you give it.
Here is a thread dedicated to the question of whether or not we should change our ratings.
In my opinion, rewatching a movie is comparable to 'you watching a different movie' or 'being another person that watches the same movie'.

second: in information theory there is sth. called 'Nyquist-Shannon sampling rate theorem' (https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem).
Roughly speaken, when I transfer the theorem to the movie rating setting, it would say that you need twice the percision of ratings (for your samples [i.e. movies] of the personal movie rating function [your taste]) to make sure that the ordering of the movies does not have an interference (overlapping or unordered situation [A>B and B>C but C>A]).
When I did see that problem, my solution was the following (refering to my "first"): making a prototype list of the movies where I rated on a fine granularity (in the early phase it was a 1-99 scale, and each single option was fine [except the mult. of 10]). Compare the new movies to the list, and after that within the bucket [i.e., bucket of 50 to 59 rated movies]. So I made sure, compared to the sampling rate of the bucket granularity [sample rate of 10] I 'was save'. Nevertheless I accepted that based on the bucket granularity there might be an interference of 10 rating points (meaning: on an evenly distributed moviebase a 5 rank movie [i.e. 55 rated] could be counted as a 4 or even a 6 true taste rate). And to solve that: I decided that I can live with such imprecision. Therefore, at least my taste for every two movies with a rating distance of 20 will be unambiguous :)

I hope that makes it clear, and if you write a science paper for movie rating don't forget to ask for adjustment details for my sampling, the real name and to cite me ;)

livelove
Posts: 759
67 Ratings
Your TCI: na
Joined: Thu Sep 17, 2015 9:36 pm

Re: voting: A>B and B>C but C>A ?

Post by livelove »

@spacefloh:
very interesting. Thanks. A lot to digest for me.
You don't know how much I'd wish to be able to talk to you face2face ...
Last edited by livelove on Sun Sep 30, 2018 12:19 am, edited 1 time in total.

fra paolo
Posts: 18
984 Ratings
Your TCI: na
Joined: Sun Aug 19, 2012 6:24 pm

Re: rock/paper/scissors: voting A>B and B>C but C>A ?

Post by fra paolo »

This isn't a problem for me at the extreme ends of my rankings, but it does happen in the middle. Thus, I don't worry about it, because I think of the middle of my rankings as a constant work in progress. Like those buildings in Italy that are always 'in restauro'.

Luna6ix
Posts: 501
4609 Ratings
Your TCI: na
Joined: Sat May 17, 2008 3:26 pm

Re: voting: A>B and B>C but C>A ?

Post by Luna6ix »

livelove wrote:
Luna6ix wrote:If you score based upon a variety of factors and don't have a weight system in place that truly represents your preferences of those factors, then that'll happen. It happens to me too, I'm sure, but I don't usually compare more than two movies with the same score when looking at my total rankings.
I agree. The problem must be a lack of "weight system". Some kind of gauge or reference value. What is your "weight system"? How do you approach this issue?


I don't use one. Too much work. I'd need to rerank a couple thousand films and most of them, I'd have a poor assessment of now. Like a number of the other old timer users have said, I just accept that it'll happen. Maybe tweak a little bit, but perfection is either unattainable or too laboreous to be worth it.

Post Reply