I’ve noticed similar results, but in my case, that might be because the new hardcore tech-adjacent music isn’t well tagged.
for reference, my top genres according to YiM '23
16% PROGRESSIVE ROCK
9% PSYCHEDELIC ROCK
8% POP ROCK
8% ART ROCK
on the one hand, I do listen to a fair bit of rock, but I’m surprised to see no electronic genres here, no Vaporwave, no Future Bass, no Synth-Pop
Hi! For the top genres, we used recording tags, tags of all the artists present on the artist credit, and tags of the releases and the release groups on which the recordings is present.
We also removed “pop”, “rock”, “electronic”, “hip hop” from the final genres before generating the image because we felt that is what most people would get otherwise.
I’m assuming the top genres data also took into account playcounts too? like one song of Genre A played twice has the same (or at least similar) weight as two songs of Genre B played once?
Yes, both play counts from listens and tag counts from the entities in musicbrainz were considered in the weighting.
I sort of get what you wanted to do here but I feel this has unintentionally ended up making things a little misleading? At least in my case which seems similar to OPs. Electronic, for example, while vague and encompassing many sub-genres would’ve at least been a little more accurate for me. The 4 genres listed don’t even cover half of my listens.
My top two albums shown are both synthwave, and indeed the majority of my listens in 2023 were certainly synthwave/darksynth (I use Lastfm as well and can see there that those are my top genres). I listened to a fair amount of punk rock, alt rock, pop rock etc. too which is what’s given as my top genres here. Still though, it’s surprising given the number of listens by top ones had and the weighting you confirmed above. I’m pretty sure the artists I listened to most are tagged properly, but perhaps the releases weren’t - I’ll have to go through them at some point as it may well be tags that are the issue, or lack thereof I should say.
Also, may I ask the thought behind displaying data for this particular stat in such a fashion:
Is this a bug or intentional? I can barely read the first three. Firefox and Chrome latest stable both like this. I would’ve probably just stuck with the top album rather than two.
Same here. My list looks like this:
9% ALTERNATIVE ROCK
9% DOWNTEMPO
9% TRIP HOP
7% IDM
What I listened to mostly is electro, techno and acid house. Like UltimateRiff, I suspect the cause of that is the poor tagging of albums in those genres, whereas the occasional mainstream album I listen to is well tagged.
Most of what you mention is the result of a number of factors, but largely my fault 
I’ll lay it out:
We were only going to add one or two pre-decided features this YIM, but @lucifer and @mr_monkey are amazing and ended up having time to do a bit extra - though we missed some testing deadlines/goals. @UltimateRiff (if I remember correctly) had requested both a ‘summary’ tile, as well as the inclusion of genres somewhere. These seemed like really good additions, so I made a mockup, and it made it out with the go-live!
Before go-live we (quickly) tested the function. Every image had electronic, pop, and rock in the top five, regardless of the users listening habits. I had predicted this, because of the MB genre stats, and already recommended we filter these out, but the result was even more extreme than I expected. MB genre stats:
rock: 1,422,339
electronic: 1,122,807
pop: 668,125
jazz: 357,365
hip hop: 288,809
experimental: 261,259
ambient: 254,774
(MB presentation from the 2023 summit, slide 11)
The thinking behind filtering these out is that the secondary tags would probably still be related to these top tags, but be more interesting, and at least a bit more relevant to the user (based on the few preview pics I saw).
It turns out they often still just aren’t correct, whether this would be improved by adding those ones we filtered out back in I’m not sure. Potentially. We will have to revisit this next year. I suspect genres may simply not be in a state to be included on a ‘user summary’, and needs to be put into its own tile (so users can decide to only share them if they feel it is relevant), or have something more relevant picked out (not just top genres, something like ‘you had a spike in your listens to [genre] this year’, or something else that we find to give consistently relevant results).
I’m a very invested MB editor so for me one upside is that a few people are feeling encouraged to add or fix tags across MB, as a result. But obviously that doesn’t help us this year.
re. the text being behind the picture, that is entirely on me, I made the snap call to have the image on top (hiding the genres even more than now!), going for style over substance. That obviously doesn’t reflect the needs of our users. @mr_monkey pushed out some fixes after we received feedback, adding the ‘overlay’ effect, so it’s a bit more readable.
More learnings for next year - you can suggest and follow improvements on this ticket for YIM 2024, if you like:
I have added this feedback there as well.
Appreciate the thorough response to everyone’s concerns here, it’s insightful. Like I said, I definitely saw what your initial intentions were. The lack of decent (non-primary genre) tags is an obvious partial culprit here at least with regards to the raw genre stats.
For what it’s worth I absolutely do want to see genre stats in YIM, but I don’t have a good solution for “fixing” how it’s been calculated this year. Forcing or even encouraging users to tag releases when added for example seems too drastic a measure. And your decision to filter out those main genres looks to be sound, despite my earlier grievances.
As for the presentation of the data, it’s purely a design issue, one that could still be fixed now if you wanted to. Remove the second top album art, move the primary one there instead. Might not look as fancy but at least everything would be readable.
And from your ticket
Since we can’t fix our MB genre stats on a whim
Why not, though? I’m assuming you generate all these stats in a huge batch job then save them to the database at runtime, such that they are cached for that point in time. Obviously, running this for the entire userbase is presumably a very resource intensive task. But how come I, as an individual user, can’t force refresh my own YIM stats on a whim as you say?
You could even limit this to a one time refresh, or once per every X days or something. So for those of us who may be inclined to go through our listens and meticulously tag artists/releases/recordings could in fact get an improved YIM page should we desire. (which I do, and am potentially willing to spend time tagging if I knew it’d be worth it)
Thank you for all the detailed feedback!
The issue is not a lack of want, but a lack of time. YIM takes almost a month of work from a huge chunk of MeB’s (very small) team, which is already too much. So non-critical fixes get put on the roadmap for next year.
With that in mind, I’m not sure we should make the genres more prominent this year anyway, considering people aren’t finding them very relevant.
I meant that we can’t fix all the genre tags in MusicBrainz, not that we couldn’t refresh the LB stats for people 
It would be really cool to allow people to update tags/update MB and refresh their YIM as they go! Please do add ideas like this to the ticket.
On that note, this year @lucifer is planning to recalculate YIM for people who import listens (e.g. from last.fm or libre.fm) after the cut-off, in a couple of weeks. I’m not sure if this includes updating already generated YIM’s, which always takes a considerable amount of processing power and server space. I’ll let lucifer comment on whether that’s possible - potentially just recalculating genres would be interesting.
The data will be recalculated for all users upon re-run and any updates in tags on MB will also be reflected automatically.
I guess this is another sign that genres are a rather silly and not very useful concept.
I’d rather choose another subject to display in the summary, but I know you can’t please everyone.
My top “genres” are all rock-related, but my top artists looks like this:
a tip for those of us with local libraries, there is a Picard plugin (thread here) for submitting genre tags (and other tags like moods, styles, keywords, and more, if desired) made by @Victini. just look for Submit Folksonomy Tags in the list of plugins~ (it does require loading everything you wanna submit and manually running the plugin from the context menu, but I think it’s quite easy myself~)
I understand the reasoning here, but I want to be sure that genres are still visible on the YiM 2024 (even if they’re not so front-and-center). if they get left off, I could see them becoming forgotten by many users and not getting the egg for their chicken
I’ve actually had this same idea myself, and almost made a ticket for it (but haven’t yet…)
another thought about genres, I think part of the issue is oftentimes you’ll have users tagging genres as well as subgenres on the same entity (for example, tagging rock, soft rock, and pop rock on Rumors by Fleetwood Mac*). fairly frequently the supergenre has more tags than the subgenre. I wonder if it might make sense to filter out the supergenres when this happens, perhaps that might improve the genre part of YiM? might be worth testing at least~
That is my thinking.
If we find top genres continue to not reflect users that accurately these are the options I’m thinking of:
Putting my two cents in here; the only issue I have is that for a pleb like me I genuinely use the “big genres” to organize my library; that means Arena Rock, Hard Rock, Classic Rock etc. all falls under just Rock. Due to the way I setup my smart playlists I am happy to have these sort of tracks all fall into these groupings.
I am trying to add individual genre styles here and there if I feel I can accurately identify (I have to admit there are a lot of genres that I am not that familiar with identifiying) but for the most part pop/rock/jazz/hip-hop/blues music is just that for me.
Maybe a way to toggle “big genre” on off maybe?
PSA: Add and correct LB/MB genres (and fix your missing listen links etc) before the 22nd of Jan to get a refreshed Year in Music!
I haven’t checked with @lucifer if he has time to do this, but possibly we could do the refresh without removing those top level genres this time? The top listens for everyone is likely to be ‘rock, electronic…’, but that may be preferred. Thoughts?
We have now re-run all the YIM stats 
I’m keen to know if anyone managed to change their YIM summaries by updating the genres for a decent number of their releases!
I tagged as many artists as I could think of. My K-Pop genre went from 68% to 69% and dance-pop went from 4% to 5% (Though I didn’t tag that, so someone else must have) 
I need to do more tagging for next year!
And we will improve how we calculate and show the genres statistics !
I think we could also add non-genre tags next year too? with filtering, of course. I’d love to see my percentage for the various fandom musics I listen to~
