I tried ranking my albums out of five stars - I think I've gotten it wrong. Thoughts?
TLDR/Warning: this might be a tedious read. But I'm curious if I could have gone about rating my albums better.
I tend to simply either favourite an album or not. The idea of giving albums and tracks marks out of five stars seems tedious, difficult to match to how I feel and just doesn't match how my head works. But my collection has grown over the decades and I've been bed bound a lot lately, so I'm trying to organise/categorise based on my feelings towards the albums rather than genre. I'm also hoping to rejuvenate my old interest in music (playing in a band and recording for a living took the shine off of casual listening for me). I thought it would be an interesting experiment to try out, so I rated songs from over 50 albums.
I came up with a rigid and hopefully balanced definition for each rating:
- 1 star - Dislike. I hope I never hear this song again (but I'll keep it purely because it's part of the album)
- 2 stars - Neutral. It doesn't annoy me, but it's too generic to be interesting
- 3 stars - Sometimes this song hits the spot.
- 4 stars - This song usually hits the spot.
- 5 stars - This song always hits the spot.
Then I rate the album out of five stars based on the average of the song ratings. The result is that no albums got 5 stars, a seven got 4 stars, the vast majority of albums are rated at 3 or 2 stars. Even among the 3-star albums, some I like much more than others depending on whether they contain mostly consistent 3-star songs or half 4-star songs and half 2-star songs.
I wonder if the lack of 5-star albums is because of the definitions I gave each of the 1-5-star possibilities. For example, I don't know if any song "always hits the spot". Or maybe it's just that I'm not as into my music as I used to be.
Anyway, I thought maybe people interested in music and data might have thoughts on going about this differently. It's worth asking before I do the next 1,000 albums :) Maybe you'd define each of the 5 stars differently. Any takers?
Edit: thanks to everyone for reading all this and commenting their thoughts. I have a system I'm happy with now, but always happy to continue to chat with fellow (and reluctant) pedants about this.
I don't do much music rating BUT i follow a similar rating system for my books/movies/shows/games.
The few things I've come across:
I think having 3 as Neutral made more sense. Its a "would watch it if it was on" vs a 2 being "would not want to watch again and would leave the room" while 1 is straight up garbage.
When looking at "combined" ratings, a 5 star is naturally going to be extremely unlikely as the most generous scenario that could be a 5 star is 1 5 and everything else is a 4 and you round up no matter what. You can of course tweak your logic if you think that doesn't reflect the reality you're trying to represent. For example if the max score is 5 * N, where N is number of songs, then if the album total is greater than (5 * N) / 2 it counts as a 5 could work (probably not that exact logic but you get the idea).
It's a hard balance to strike because on the one hand you can probably think of "must listen to" albums, but they likely are not literally all 5's. Probably work backwards from what you think should've been 5's and go from there? Personally I don't really care about "albums" because I just want to know what songs I like. Not what album tells a coherent experience/story or whatever.
Edit-
Another way to think of my scale is that 3 or higher is "would recommend it" where a 2 is "eh if its your thing maybe but probably not" and 1 is "i would not recommend unless youre trying to learn what bad media is like".
Ive also got a -1 for shit so bad its good but thats probably less common in music.
Re-centering around 3 stars could work, actually. I've basically set myself up where 5-stars is outside of the bell curve. I think I'll start from your 3rd point: I'll grade my favourite albums and see if the work with my existing criteria or if they expose a problem.
(some time later)
Okay, I've taken some absolute favourites of mine like Thriller by Michael Jackson (1982). 7 out of 9 songs get 5 stars and the remaining 2 get 4 stars, so it is possible. It might just be that the random 50 I started with were not representative enough.
So I basically need to think now how special I want 5 stars to be. Do I only want absolutely excellent albums to reach that score, or I do I want to re-center my scores around 3 stars (as previously mentioned) and that way elevate more albums to the 5-star level. Thanks for unjamming my brain on this. It can be a bit stressful! :)
I honestly gave up trying to rate my music because of this very problem. I stick to just “favouriting” the ones I like the most and leave it at that. It’s imperfect, but it’s definitely easier for my brain to wrangle! I have a smart playlist that surfaces anything favourited that I’ve not heard in a while (I think it’s set to six months, off the top of my head), and I get by pretty well like that. Though I do still find myself skipping tracks, that tends to be a mood-based response.
Which brings me to my main issue with rating (including my binary favourited/not favourited method) music and other media; I am a constantly evolving creature. What I am in the mood for shifts from day to day, week to week, and so on. It’s hard to build rankings that will be resilient to that kind of flux without constant tinkering and review. Which I would genuinely love to have the time for, but have not in a very long time!
Honestly, I agree so much. Music applications have had this 5-star rating facility since the early 2000s at least and I always thought it was for masochists! As you say, moods change and what might score highly now might not in a month or a year. It just seemed so pointless.
I'm lucky(ish) to have the time so I'm using it as an exercise to make the days pass more quickly and get more intimate with my admittedly gluttonous collection of albums gained over the decades. Maybe in a year I can do a retrospective here and say if it was worth the effort or not. Hey, maybe it would make a good retirement project for you :)
There are thousands of us doing this all the time over on rateyourmusic.com, my favorite and most-frequented website. Some of us have made the 5-star rating scale into an intricate science.
I don't rate songs as much as albums, though, and I don't think rating albums by mathematically combining song ratings is a great approach. For me, an album is a complete work on its own, and often more than simply the sum of its parts.
That's absolutely valid. And I think that's why I'm not entirely happy with my approach. But it was the only way I really know how to do it.
How do you decide how to rate albums: vibes? That's generally how I decide what to favourite since for me music is first and foremost about vibes.
It's tricky when some music I listen to is not because I love it, but because it's a good palette cleanser (such as certain jazz or classical). So I end up needing to rate those higher because they're important, but not necessarily music that moves me the way I wish it did.
Oh yeah, it's all vibes. When you've been thinking about music in this way for like 20 years, it gets pretty easy and I don't really have to think about it too much. Do I like it? It's 3 stars. Do I really like it? 3.5, and up from there. (I like to keep my 4s and up a bit on the rare side.) Do I feel nothing? 2-2.5. Dislike it? 1.5 and lower, depending on my vitriol.
But a rating number isn't everything. It is very important to have good palette cleansers for sure. I probably listen to my more middling-rated jazz albums more than my favorites, just because they're good but not so good they'll distract.
Yeah, music has various purposes like ambience - just like you said. I was good at vibes-based sorting/rating when I had a much smaller collection years ago (max 30 albums). But now instead of a small intimate collection I have a huge library with some stuff I don't even recognise haha. The plan now is to start deleting and get back to that, so maybe I can get back to simplicity and feeling again soon.
The lack of five-star albums is because as you average more things, by definition they tend to the mean. If you averaged your entire music library, would you be surprised for it to come out at a 3?
Now, of course, albums are not formed by taking a random sampling of songs with normally-distributed ratings. There probably is a signal here; do you generally prefer your 3.5-star albums to the 2.5-star ones? The challenge is that averaging the song ratings weakens this signal. Mathematically, there are a few approaches you could take:
…But the mathematical "fix the stats" approach is probably not what you want regardless. As you note, an album with all 3-star songs and one with half 2 and half 4 are quite different in subjective feel. (And also, people tell me that albums can be constructed as a cohesive more-than-the-sum-of-its-parts whole rather than just a collection of independent songs, though my listening habits don't allow me to observe that.) What you might want—and you kind of hint at this already—is to give albums their own rating, rather than trying to derive it from the component song ratings.
Thís is great. When I do my averages on the calculator, I do get varied remainders that just get rounded up. I could use those variances to renormalise, or pick a different function than mean.
One thing I could do is instead of representing something I don't ever want to hear again with 1 star, I could represent that with 0. I didn't want to do that this time because since my library is currently unrated, everything is currently 0. So by going no lower than 1 I can tell which tracks haven't been rated yet.
So I think when I'm finished my first iteration I'll do that as well as taking your suggestions on board and spreading out the meaning of the values further. That way I won't have a load of albums with a rating of 3 that actually hold varying levels of appeal.
I don't have a lot to add, but thought I'd chime in with how I used to rate things. It's a bit different...
0 stars: Anywhere from "I hate this" to "this is fine but doesn't impress me." I might want to listen to it again and I might not, but it's nothing special.
1 star: This is great! If this were a student paper, I'd need to add a gold star to it.
2-5 stars: Similar thought process there. 2 stars is very, very good, whereas 5 stars is one of my favorites of all time and I'll come back to it for the rest of my life.
I don't really like the whole 5/5 rating system. It seems too negative. Why do I want to "punish" something with simply one or two stars? I would rather just ignore/delete it than waste my time on it. I'd rather save the stars for things that deserve them.
It's been a long time since I've rated music or movies or whatever though. Nowadays my "star" ratings are basically delete/block things I dislike, ignore or skip things I feel neutral about, add things to playlists that I like, and purchase/download things that I love.
That's a really interesting take on this. In theory I should be doing it this way since I'm not a collector - I don't want to keep any material I won't listen to and enjoy.
The only issue for me is that if I like an album overall but don't like some of the tracks, I can't bring myself to delete them. For some reason I have to preserve the album integrity. The exception there is if there are just two or three good tracks: in this case I'd copy them to my folder of singles and remove the album from the application itself.
I could use your system for movies, books or singles, though. Thanks for the suggestion. These answers are helping me see that the problem stems in part from unintentional accumulation of albums. I'll jettison many of these as I go and be far more conscious of what I keep from now on.
Agreed--I would never delete individual tracks from albums unless they were particularly offensive to my ears. Even then, I'd probably make an empty track or something as a placeholder to remind myself that it existed.
Just be careful if you start treating stars like this, as it might make you grouchy to have to use them in other ways! For example, if I'm rating a product on a shopping website, it annoys me to have to use stars at all, let alone the default of basically 5 stars if it's a reasonably good product that fits the description.
I use a simple 5 star system on rym, but I don't rate individual tracks only albums as a whole.
5 star albums are albums I love and that I've continued to listen to over many years. There really aren't many of those.
4 stars are albums I loved but doesn't quite mean as much to me or I haven't spent enough time with it to qualify as 5 star, but it has potential to be upgraded if it holds up over years.
3 stars are albums I find enjoyable but maybe aren't always in the mood for them. Above average, acceptable. I like it.
2 stars are not for me but it's not awful and I can see how someone could enjoy it.
1 star is what I'd consider a total waste of time, very few albums meet this criteria for me.
Personally. I find not using half-stars, and also not rating individual tracks makes it a lot easier to fit my music into the different categories.
That's a good system. I think if I'd been more intentional with my collection (I was a DJ for a few years and amassed all kinds of nonsense) I could go with a system like that based on personal connection to the music. I might try that after I've finished my first iteration.
I think this is where your system falls apart.
Unless an album only has 4- and 5-star tracks (and I can think of some that fit that for me, but not very many), then your albums will never be able to get that 5-star rating.
You should rate your albums independently of the track ratings. Obviously they'll still be related, but a 5-star album doesn't need to have only 5-star tracks. Some songs work well while listening through an album, but I never want to hear as singles, and some songs are great no matter the context. For example, I'd give Radiohead's OK Computer 5 stars, even though I'd probably give Fitter Happier 1 or 2 stars under your system.
Your album rating system could be nearly identical to your track rating system, but I think in that case, you'd never have any 1 or 2 star albums, since if you dislike them that much, why keep them around? That might be fine for your purposes though.
There's also just the fact that some albums can stand by themselves as a complete work, and some albums are just a collection of songs (as is increasingly the case since the advent of streaming). For me, a 5-star album is an album that I would want to listen to from beginning to end. If it's just a bunch of good songs (or some good and some bad songs) that I like listening to individually, the album probably doesn't get 5 stars from me, because the album isn't "good", even if the songs are. Hopefully that makes sense.
That makes a lot of sense, and I like the idea of defining a five-star album as one that I can listen to from beginning to end without skipping anything.
I did find - eventually - a few five star results for albums. I thought it was never going to happen. This was based purely on the calculations. But one problem I still have is that the rating of 3 represents a relatively wide range of personal feelings towards those albums. So I think when you say that things fall apart by my averaging the song ratings to rate the albums, you're correct.
I'm starting to think I'll do a few passes in this case. For news stuff I'm listening to blind, I'll use the calculated method. And when I start to bond with the album a bit more (sounds cheesy but hopefully you know what I mean) I can tweak it either with a better formula as whbboyd mentioned, or on more personal criteria that yourself and others suggested.
I think the difficulty is that so many of the albums are unfamiliar to me. Half of them are albums I bought because I really liked them, but half are random ones I ended up with while DJing. I haven't got a feel for them yet and I might be rushing it. Thanks for the comment!
Could you use the mode by default instead of the mean? Maybe revert to using mean for any bimodal or multimodal albums?
Thank you for re-introducing me to the mode. I haven't seen that since I was in school many years ago. It does appear to represent the albums better:
That seems to give more meaning to the ratings of the tracks in an album. The other good thing about this is that it's easy to do without changing how I rated the tracks in the first place.
If a balance between the two exists, that could be even better. Maybe something that uses the mean and the mode? but I think considering I'm looking to represent the dominant rating rather than just an average, this is an improvement. Thank you!
Edit: I've just encountered a situation where the mode does the opposite of what I wanted, so I'll have to abandon it. But it was a very interesting experiment. Basically, raising the rating of a song on an album lowered the rating of the album itself. Here's the data:
First one isn’t a mode of 4; it’s bimodal at 3 and 4. The second one is a mode of 3.
The ratings of songs on albums isn’t likely to be normally distributed so a mode would probably be a better reflection of what the “average” is than a mean or median. If there isn’t a single mode, I think swapping to mean and rounding is the simplest solution. That would make both examples a 3 fwiw.
Good luck however you decide to do it!
My music listening is centered around albums, so I mainly rate albums rather than individual songs. Occasionally, I keep track of my favorite/standout tracks to add to playlists, but other than that, I don't have a system for rating songs.
For albums, I calculate ratings based on the ratio of liked songs:
I believe it's more meaningful to determine my favorite albums by feel rather than by calculation, so I only assign 5 stars to albums that I would like to listen to regularly, regardless of the calculated rating. I don't always get the chance to come back to these albums, but knowing where they are fulfills the purpose of my rating system, which is to remind myself which albums are worth revisiting.
Actually, this is really interesting. For my system that would make it easier because I'd either like or not like a song (a binary) and rate the album based on that. I think you're right that there's something deeper that I'm not capturing using my current system.
I'll continue as I am for the first iteration and use that to identify albums to delete (or archive). Then as I listen more recreationally next time around I'll give myself a bit more freedom to go by feeling. I'm avoiding it this time because I don't want to allow more music through that might be challenging at the moment (and might rank lowly) and see if a few more listenings might change my attitude. Thanks for the ideas!
I use some key words to make things easy to rate in my mind. I find it really easy to ask myself “is this a good album, or a great album?”
The scale is originally based off an out of 10 hence the .5s I don’t really rate anything below 3, because 0-2 is just the same words but for how much you dislike it.
I’m going to go with something like that in future. I’ll use no star for anything I dislike instead of one star.
This is an interesting problem! Would you get more useful album ratings if you always rounded up? For example if an album's got a calculated average of 4.1 then you would round that up to 5. Is that a more accurate rating? Because to get a rating of 4.1 it means it must have at least one 5 star song on it, and either the rest are all 4 star songs, or for every 3 star song there's a 5 star song balancing it out, and that sounds like a pretty good album? And that's the worst case, an album with a calculated rating of 4.4 is going to have quite a few 5 star songs!
The drawback is you'd never have a 1 star rated album, but if every song on an album is only 1 or 2 stars, which I think it what you'd need to really justify a true 1 star album rating, the solution might be to delete that album!
Reconsidering the way I do rounding is an interesting idea! t would certainly help the current situation where it appears I have no really good music :)
You're basically making me rethink how special I want 5-stars to be: a rarity or more inclusive of less-than-perfect albums. I've given it some thought and tried rating my favourites (as suggested by Eji1700) and decided that having fewer 5-star albums probably works best for me. The reason for that is I would prefer a smaller collection any way (I'm not a collector as such. I actively listen to what I have). So your suggestion of deleting albums that rank lowest would actually help me with decision fatigue. And by keeping the 5-star mark a rarity, I can achieve a leaner collection.
Thanks for poking my brain on this! I read both answers and in combination came up with this resolution which I'm now content with. Phew!
I generally don't rate stuff on 5 star scales (or any scale, for that matter). The closest I get is the three tier system on netflix. If I were to use a 5 star scale, music or otherwise, star ratings wouldn't have concrete definitions: ratings would be a vibe thing for me, a non-specific combination of things like how much the song resonates with me, how much a song makes me feel or want to move, how happy I am when it comes on and how consistent that is, etc. a three star in one genre might be quite different to a three star in another because I gravitate towards the genre with the higher ratings more.
That's relatable. That's why this is so difficult I think. I've decided to be consistent for reasons I'm not fully sure of. But the result aren't what I expected.
A three-tier system might have simplified things if the apps supported it by default. I might not need to rate individual songs, but rather albums. But I'm enjoying the admin aspect of it, oddly enough.
if you use foobar2000, you can use foo_skip to mark those shitty songs and skip them. For me, I often recut a record so it flows how I want without all of the extra bullshit (especially skits and long intros.)
That’s a seriously nice touch! I’m using a subsonic server and it doesn’t support that feature, but I might ping the dev about it. Thanks for the heads up. That would suit me perfectly.
i havent used it, but my guess there is probably a smart playlist thing to bypass zero starred items.
but yeah, everything should have a skip thing. it also has an option to skip tracks less than x seconds.
I love it! I do my album rating different: 10/10 system, where albums are scored based on the proportion of bangers/filler on the album. So even if there are 2 amazing songs, but 8 snoozers, that's a 2/10. Like you I have very few 10/10 albums but they absolutely exist. Examples include:
Parcels | Tieduprightnow
Mapache | Mapache/From Liberty Street
Whitney | Light Upon the Lake
Bueno Vista Social Club| Bueno Vista Social Club
It was interesting because it wasn't the albums I was expecting. Most of those ended up being 8-9/10. Those include:
Daft Punk | Discovery
Vampire Weekend | Vampire Weekend
Mac Demarco | Salad Days
Lauryn Hill | The Miseducation of Ms Lauryn Hill
Stan Getz and Joao Gilberto | Getz/Gilberto
Amy Winehouse | Frank & Back to Black
The Clash | London Calling
All still absolute banger albums, just not "perfect" ones as it were. Also, obviously all of this is subjective.
In hindsight, it could have been a lot easier if I’d done what you did and just gave a binary rating for the songs and then scored the albums based on that.
Thanks for sharing the insights. I know what you mean about being surprised by results. That’s giving me motivation to continue and it’s changing my listening. Parcels | Tieduprightnow is high up on my results too, with Shadows by Remy van Kesteren in top position as of now.
I'd love to do a rating swap if you'd be interested. It's always so cool to see what albums speak to people or even which musicians are consistent. I'll just have to reformat my excel sheet as it currently uses color as a stand in for a number system - green 10, yellow 9, etc,... :).
That sounds like fun. I’m happy to send you on what I have. I’m really just getting started, though. I haven’t rated the most important stuff yet. I’m sure there’s a way to query a subsonic-compatible server ands retrieve a list of ratings!