Toward a Quantification of Love FOR ANIMU Part 1: The Re-watchability Ratio

We know his love for Animu is OVER 9000

We know his love for Animu is OVER 9000

In this series of posts, We Remember Love would like to propose quantitative measures for rating anime in form similar to financial ratios used for evaluating the performance of stocks.

These ratios are not to be taken as absolute measures of quality, but rather may be taken as subjectively determined measures of one’s personal enjoyment of an anime.

The “Rewatchability Ratio”

A “Rewatchable Moment” is defined as an extremely enjoyable scene within an anime episode such that the viewer will watch said scene an unusually large number of times.

Ex 1. Mechafetish watches a reasonably enjoyable scene approximately 5 times in one year. A Rewatchable Moment on the other hand is one that he will watch more than 100 times a year.

Ex 2. ghostlightning watches a reasonably enjoyable scene approximately 2 times in one year. A Rewatchable Moment on the other hand is one that he will watch more than 2 times a year.

The “Rewatchability Ratio” is defined as the number of Rewatchable Moments in an anime divided by the total number of episodes thereof. It basically shows how many Rewatchable Moments an anime was able to deliver relative to its entire length. A higher Rewatchability Ratio is usually better than a lower one because it indicates that an anime was able to deliver more enjoyable moments for a given number of episodes. The Rewatchability Ratio is comparable between anime for a single viewer, but is not comparable for a single anime between several viewers.

Its form can be given as follows:

Rewatchability Ratio= (# of Rewatchable Moments)/(# of Anime Episodes)


Sample Computations

The following are some sample Rewatchability Ratios from yours truly (SPOILERS AHEAD):

 

Tengen Toppa Gurren Lagann

1. Episode 1 (The Introduction)

gurren-lagann-01-large-012

2. Episode 1 (Breaking Through to the Surface)

gurren-lagann-01-large-283gurren-lagann-01-large-314

3. Episode 8 (Goodbye My Friends)

gurren-lagann-08-large-28gurren-lagann-08-large-37gurren-lagann-08-large-39

4. Episode 15 (Vs. Lord Genome)

gurren-lagann-15-large-21gurren-lagann-15-large-28gurren-lagann-15-large-30

5. Episode 27 (The End)

gurren-lagann-27-large-21gurren-lagann-27-large-29gurren-lagann-27-large-32

Computation:

# of Rewatchable Moments = 5; # of Episodes = 27

Rewatchability Ratio = 5/27 = 0.185

 

 

 

Macross Frontier

1. Episode 6 (Leaving for Battle)

macross-frontier-06-large-31macross-frontier-06-large-321

2. Episode 15 (The Sing Off)

macross-frontier-15-large-27macross-frontier-15-large-28macross-frontier-15-large-29

3. Episode 25 (The End)

macross-frontier-25-large-30macross-frontier-25-large-32macross-frontier-25-large-34

Computation:

# of Rewatchable Moments = 3; # of Episodes = 25

Rewatchability Ratio = 3/25 = 0.12

Applications

Though, as stated above, the applications of the rewatchability ratio do not lie in terms of its comparability across users (the ratio does not make it possible to control for taste between viewers), the ratio itself does reveal a lot in terms of the utility/enjoyment generation process of a particular user across anime.

As such, by comparing Rewatchability Ratios between anime, insights may be gleaned as to the personal preferences of the user with regard to the directorial styles/scene construction of anime. This can be done by modifying the components of the formula with parameters that narrow down the total number of re-watchable moments. For example:

Rewatchability Ratio for Mecha Battles =

Rewatchable Scenes with Mecha Battles/Total Episodes with Mecha Battles

Now, both TTGL and Macross Frontier have mecha battles in abundance, however, despite my personal preference for real robots and Macross as a whole, my re-watchability ratio for the mecha battles in TTGL is clearly higher than that for mecha battles in Macross Frontier. Does this mean that, as far as my preferences go, the mecha battles in TTGL are better directed? Perhaps. This then allows me to dissect the nuances that are present in one anime that but are absent in another which have shifted my preferences in one direction. (This is an example only and such experiments should be performed with larger sample sizes of anime for more conclusive results).

Issues with the ratio

The Rewatchability Ratio represents the beginning of several attempts we shall undertake to subject our biases and preferences to statistical scrutiny. As such, it is not without possible issues notable of which are the following:

1. Recent Viewing Bias – Rewatchability is a time varying characteristic and tends to be greater upon the initial viewing of an anime, and less as the initial viewing recedes from the viewers memory. As such, the optimal time period for testing rewatchability would be one year after initial viewing, controlling for additional viewings caused by extraneous circumstances (viewing with friends, ‘compromise’ viewings for clubs, etc).

2. Unit of Measure – The use of # of Rewatchable Moments vs. # of Episodes may not be statistically sound because the length of time (minutes/seconds) scenes and episodes take are not equal. As such, we may want to take a look at further refinements to the model in the future.

In any case, it could be a fun and revelatory exercise. At least we think so.

Exercises

1. Consider your 2 favorite anime, and compute for the rewatchability ratio.

2. Does the result surprise you? Will this influence their standing among your favorites (if you rank them)?

3. Do you tend to watch whole series, certain episodes, or certain moments only?

This entry was posted in analysis, Toward a Quantification of Love and tagged , , , . Bookmark the permalink.

12 Responses to Toward a Quantification of Love FOR ANIMU Part 1: The Re-watchability Ratio

  1. schneider says:

    1. GaoGaiGar TV: 7 watchable moments would be a good estimate. 7/49 = 0.14. Eureka Seven: 10, from what I can remember. 10/50 = 0.2.

    2. Not really. I know some of my favorites have less rewatchable moments than others (GGG again, tons of monster-of-the-week episodes), but I still like them. Another principle of mine is not rank my favorite shows, so sorry. :P

    3. It really depends. Sometimes I don’t like the show at all but watch its good moments. Or I’m pressed for time and I just watch the episodes I like. That said, if I could rewatch a show from beginning to end, I would.

  2. I have exactly 2 rewatchable moments. The first is the Kurenai musical, and the second is the climax of RahXephon episode 19. There are other scenes I may have watched once outside of the anime itself, but those are the only two I’ve done so several times.

  3. mechafetish says:

    @ schenider

    Never saw GGG TV (Only the OVA), but I have a similarly high rewatchability ratio for Eureka 7.

    The rewatchability ratio definitely doesn’t cover all aspects of an anime, but we will slowly be adding more and more ratios to evaluate other aspects of the anime we love.

    Lastly, I tend to rewatch moments themselves A LOT (about 100 or more times in a given year for stuff I really like), but my behavior is already a bit extreme.

    @ 21stcenturydigitalboy

    The rewatchability ratio attempts to capture an unusual (for the person) love for a particular moment. As such, if watching a moment more than once is already unusual for you, then those moments that you do watch 2 times or more are rewatchable moments. As an exercise, try to fill in the following:

    “21stcenturydigitalboy watches a reasonably enjoyable scene approximately __ times in one year. A Rewatchable Moment on the other hand is one that he will watch more than __ times a year.”

    Hehe. I loved RaXephon ep. 19 as well. But in my case, it doesn’t count as a rewatchable moment because I get really depressed when I watch it (so I don’t).

  4. Phäzys says:

    1. Code Geass and Death Note at least 1/5th for either of them.
    2. Not really, i’d expect the rewatchability to be much higher haha.
    3. i would like to watch the whole series again, although Death Note loses its tension once its been watched once.

  5. lelangir says:

    Hmm, what about entire series that are rewatchable? You would then need an operative definition of “moment”. For instance, LoGH: 110 episodes, but how many moments precisely? Perhaps for this reason we need to devise to paradigms of rewatchability, the first on the level of particularities (using your formula, which works well for specific moments for otherwise mediocre series); the second on the level of generalities.

    This second paradigm would be more effective for series like (for me, that is) Lucky Star and LoGH. LoGH because it has basically infinite rewatch value (100%), and Lucky Star because I always go back and watch 10 minutes or so of some random episode. ATM I can’t think of a clever formula for creating a generalized rating system for this though…

  6. ghostlightning says:

    @ lelangir

    LOLZ. When I got home yesterday I watched 3 random Lucky Star episodes. I agree that Lucky Star has mugen rewatchability = as soon as I finished it (for the first time) I rewatched the whole thing, and another time with the waifu.

  7. mechafetish says:

    @ lelangir

    That’s some really great input!

    First, some clarifications:

    To avoid the problem you cited, we have strictly defined rewatchable moments to be moments you have already rewatched, not just want to or are willing to rewatch.

    I’m on board with idea of “rewatch value” which can be infinite for a show, but the limited time we have on this earth forces us human beings to rank order what we in fact rewatch as opposed to what we only want to rewatch.

    You have a limited amount of time to allocate between certain anime scenes to be rewatched (all these scenes are drawn from a pool of “rewatchable anime” as you talked about in your comment). Each anime scene generates a certain amount of utility for you which amount is known to yourself and unique to yourself. We assume that you are a rational utility maximizer.

    It stands to reason that you will efficiently allocate your time between anime moments which generate maximum utility for yourself (this type of utility is separate and distinct from utility generated by a sad scene for example, which you would not rewatch).

    However, we cannot quantify in fact the utility generated for us per scene. As such we fall back on observable behavior to determine what scenes generate maximum utility for us. Remember, this is a rank ordering mechanism.

    Ex.

    For myself, LOGH and Lucky Star both have infinite rewatch value. However, by myself, I am unable to rank order them properly in terms of rewatchability criteria. As such, I examine my previous years behavior to see which scenes from either anime, I have in fact chosen to rewatch a greater number of times than my average rewatched moments for any anime series (for most people this is only 1 – 3 times). Then, using the formula above, I can determine which series generated more utility for me in terms of rewatchability. This does not mean that the one that comes out on top (ex. Lucky Star) is better, but only that I am more likely to rewatch a moment of Lucky Star than a moment of LOGH.

    That said, the model itself is definitely applicable to total series but suffers from several complications that make any revised model more difficult to work with (i.e. distortion due to differences in total number of eps., difficulty of accurately recalling the necessary data such as total number of all full series ever watched including rewatches, etc.).

    Lastly, with regard to the behaviour you mentioned (watching random eps of Lucky Star), it seems that this behavior crosses a certain threshold which renders the model inapplicable. At this point, the series clearly will have a greater rewatchability ratio than any other series wherein you have a rewatchability ratio less than total number of scenes over total number of eps. The difficulty now lies in the comparability of such “randomly rewatched” series with each other because (unless you keep track) the number of random watches of a series is non-observable.

    Please rest assured however that we shall continue to refine this and future models.

    Thanks again for your input!

  8. Turambar says:

    I wonder if there is a way to modify the model to incorporate the degree of rewatchability of particular moments. Taking Macross Frontier as an example for me, the battle in the first and last episode are both moments of rewatchability. However, the latter holds much more of it for me. Of course, this involves another level of examination in how to measure the exact rewatchability of a moment, whether it be simply how many times you have rewatched it or something else.

  9. mechafetish says:

    @ Turambar

    The reason the model is so simple is I felt it was important to balance accuracy with usability. However, one way we can incorporate your suggestion is to come up with multiple thresholds for rewatchability and then assign a weight to each one (greater weights for more rewatchable moments).

    Ex.

    Mechafetish watches a reasonably enjoyable scene approximately 5 times in one year.

    A Standard Rewatchable Moment (weight 1) on the other hand is one that he will watch more than 20 times a year.

    An Above Average Rewatchable Moment (weight 1.5) on the other hand is one that he will watch more than 50 times a year.

    A Superlatively Rewatchable Moment (weight 2) on the other hand is one that he will watch more than 100 times a year.

    Thus:

    Rewatchability Ratio = [(# of Standard Rewatchable Moments) + 1.5(# of Above Average Rewatchable Moments)+ 2(# of Superlatively Rewatchable Moments)]/(# of Anime Episodes)

    However, the more accurate the model gets, the more difficult it becomes to remember the kind of info you need. For me for example, I would have trouble distinguishing between moments I only watched 20 times versus moments I watched 50 times simply because I don’t keep track of that kind of data. This I felt, was one of the most compelling reasons to stick to the single threshold model.

  10. Pingback: Toward a Quantification of Love FOR ANIMU Part 1a: The Value of Rewatchability through the Lens of Eternal Recurrence « We Remember Love

  11. Pingback: The Recency Bias and How it Affects Anime Appreciation « We Remember Love

  12. Pingback: Still Enjoying Endless Eight (Yeah I went there) « We Remember Love

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s