Acclaimed Music Forums

Come one, come all to our first poll to celebrate the success of the Unacclaimed game! whuntva has more than earned a rest from poll hosting duties after the most recent (and expanded) Unacclaimed IV. We have now explored 320 songs throughout the first 4 editions of this game, so whuntva had the great idea of running a poll in which we rank the songs.

As some (most?) of us may not have heard all of the tracks, or at least do not remember them all well, I feel like this poll should be left open for awhile to give everyone a chance to explore. As such, I propose that the poll remain open until April 30th, 2017.

To participate:

1. Download a copy of the spreadsheet containing links to all 320 songs. You can save the spreadsheet as an Excel file by clicking File -> Download as.

2. Fill in rankings for as many songs as you wish into column H.

3. Send a copy of your spreadsheet to Princepoll2016@gmail.com

The scoring system is yet to be worked out, but I'm open to consider suggestions.

Given many people have not participated in all tournaments, I think it would make sense to apply a formula which acknowledges this. I can think of two ways to proceed:

1. Applying a version of what Henrik does for the EOY lists: add a rank of (X + 30) to songs not included in a voter's list of top X Unacclaimed songs
2. Doing what has been done in decade polls where unranked songs are given a rank of (X + 1)/2

To use an example, let's say PosterX submits a list of 100 songs. The remaining 220 songs would be given rankings of 130 under scenario 1 and a ranking of 50.5 under scenario 2.

My preference is for option 1, but I think it's worth a discussion.

Does anyone have a Spotify playlist? And links?

That would be nice.

babydoll wrote:Does anyone have a Spotify playlist? And links?

BTW, you probably get this a lot, but I love Studio Ghibli.

babydoll wrote:Does anyone have a Spotify playlist? And links?

I'm working on one right now. It feels like this playlist will be on heavy rotation in the next few weeks and the YT links will take too much work

My surmise is that the song titled "No Love" is performed by Joan Armatrading, not John.

I plan to classify all the songs !
Can I just ask!

Will classifying songs in the 200th place, for example, will give more points to than not to classify it at all? Because I always proceed in the same way: first round on my spreadsheet: like/dislike on two columns and I don't want to help songs I really dislike just because I ranks them (even at the 320th rank).
Thanks.

Romain wrote:I plan to classify all the songs !
Can I just ask!

Will classifying songs in the 200th place, for example, will give more points to than not to classify it at all? Because I always proceed in the same way: first round on my spreadsheet: like/dislike on two columns and I don't want to help songs I really dislike just because I ranks them (even at the 320th rank).
Thanks.

I intend to rank them all - whichever formula Moonbeam uses. I prefer Moonbeam's formula option 1 for this poll. With that formula anything you don't rank is ranked lower than anything you do rank, but you would have to rank more that 290 songs to implicitly rank anything lower than 320th, and if we've ranked that many we might as well do the rest.

DaveC wrote:

Romain wrote:I plan to classify all the songs !
Can I just ask!

Will classifying songs in the 200th place, for example, will give more points to than not to classify it at all? Because I always proceed in the same way: first round on my spreadsheet: like/dislike on two columns and I don't want to help songs I really dislike just because I ranks them (even at the 320th rank).
Thanks.

I intend to rank them all - whichever formula Moonbeam uses. I prefer Moonbeam's formula option 1 for this poll. With that formula anything you don't rank is ranked lower than anything you do rank, but you would have to rank more that 290 songs to implicitly rank anything lower than 320th, and if we've ranked that many we might as well do the rest.

Thanks a lot. It's clear and I prefer option 1 too.

There is my labor of love...276 songs available on Spotify Europe.
A huge amount of listening to do. I will favor the rating out of 10 for each song and then use the allourideas.org website to separate the ties.




*Shudder* - Just listened to Anais Mitchell "Why We Build the Wall"- eerily, scaringly accurate song in the current political situation !

Hello, the song Cherry Tree is by "The National" or "Grand National" like on the spreadsheet?
Thanks.

P.S.: it's a pleasure to relisten all these songs.

It's Grand National (my nomination )

spiritualized wrote:There is my labor of love...276 songs available on Spotify Europe.
A huge amount of listening to do. I will favor the rating out of 10 for each song and then use the allourideas.org website to separate the ties.




*Shudder* - Just listened to Anais Mitchell "Why We Build the Wall"- eerily, scaringly accurate song in the current political situation !

Amazing effort! Thank you!

A few minor quibbles: the version of "Hills of Katmandu" by Tantra is not the correct one. It is both shorter and much slower than the original track. The same applies to "Lady Bug" by Bumblebee Unlimited.

Moonbeam wrote: A few minor quibbles: the version of "Hills of Katmandu" by Tantra is not the correct one. It is both shorter and much slower than the original track. The same applies to "Lady Bug" by Bumblebee Unlimited.

Wow, Tantra's track is 15-odd minutes long there is a longer version ?
Same for Lady Bug ? almost 10 minutes long ?

Sadly they are the only available songs on Spotify in Europe - are they available somewhere else ? if not, do you prefer to have them removed so as not to induce in error our listeners ?

spiritualized wrote:

Moonbeam wrote: A few minor quibbles: the version of "Hills of Katmandu" by Tantra is not the correct one. It is both shorter and much slower than the original track. The same applies to "Lady Bug" by Bumblebee Unlimited.

Wow, Tantra's track is 15-odd minutes long there is a longer version ?
Same for Lady Bug ? almost 10 minutes long ?

Sadly they are the only available songs on Spotify in Europe - are they available somewhere else ? if not, do you prefer to have them removed so as not to induce in error our listeners ?

I'd say remove Tantra in that case. I have a link to the full version of "Lady Bug" here: https://open.spotify.com/track/32NSwpfBIS3mPJ9wzQgVhy

Does that link work in Europe?

Henry wrote:My surmise is that the song titled "No Love" is performed by Joan Armatrading, not John.

Haha, indeed! Worse for me is that I know and kind of like Joan Armatrading.

Great idea for a game! Special thanks to whuntva, Moonbeam and spiritualized for your contributions. I've participated only of editions I and IV, but I'm looking forward to listening to all the 320 songs. There will surely be lots of good discoveries.

From my nominations, only "Ali Chukwuma - Ezi Okwu Bu Ndu" isn't available on the playlist, but it has a good quality link for it on Youtube. All the other tracks on the list correspond to the original, except for... "Ray Barreto - Cocinando". In fact, the song on the list, although very interesting, is a totally different one. The version of my nominated song can be found in Spotify on the albums "Ray Barreto Jazz" (1968), and "Que Viva La Musica" (which can be found as a reissue from 2006).

Moonbeam wrote:
I'd say remove Tantra in that case. I have a link to the full version of "Lady Bug" here: https://open.spotify.com/track/32NSwpfBIS3mPJ9wzQgVhy

Does that link work in Europe?

Removed Tantra and added your link, which works in Europe !

Toni wrote:The version of my nominated song can be found in Spotify on the albums "Ray Barreto Jazz" (1968), and "Que Viva La Musica" (which can be found as a reissue from 2006).

Done !

I have also removed the Bill Fay Spotify version of Time of the last persecution - horrible sound, ghastly guitar, no piano. The YT version is miles better.
And Bjork's Verandi - which has a very bizarre classical opera version on Spotify. Better stick with the YT version.


I guess I can't be accused of not listening to the tracks

OUf !
All the songs ranks...some stats

Average : 11.14733542 (U1: 11.34 / U2: 11.89 / U3: 10.77 / U4: 10.87)
Median: 11 (U1: 11.5 / U2: 12 / U3: 11 / U4: 11)
Highest : 19
Lowest : 1

Like: 229
Average: 12.71 (U1: 13.09 / U2: 12.54 / U3: 12.65 / U4: 12.66)

Dislike: 91
Average: 7.23 (U1: 7.76 / U2: 8.4 / U3: 7.39 / U4: 6.5)

spiritualized wrote: And Bjork's Verandi - which has a very bizarre classical opera version on Spotify. Better stick with the YT version.

When the hell is Björk going to release a damn B-sides compilation!?

Romain wrote:OUf !
All the songs ranks...some stats

Lowest : 1

Go on... without divulging too much, who got 1 out of 20 ??????

spiritualized wrote:

Romain wrote:OUf !
All the songs ranks...some stats

Lowest : 1

Go on... without divulging too much, who got 1 out of 20 ??????

Something with a fish tail

Romain wrote:
Something with a fish tail

I'm stumped !

Think about Copenhague

Got it I haven't listened to it yet.... but when I compiled the list, I briefly thought : What the Hell ?

If possible spiritualized could you replace the badly recorded live bootleg of On Reflection with the original version from 'Freehand'.

DaveC wrote:If possible spiritualized could you replace the badly recorded live bootleg of On Reflection with the original version from 'Freehand'.

OK, I couldn't find Freehand, but found a compilation, can you check it's OK ?

List sent !

spiritualized wrote:

DaveC wrote:If possible spiritualized could you replace the badly recorded live bootleg of On Reflection with the original version from 'Freehand'.

OK, I couldn't find Freehand, but found a compilation, can you check it's OK ?

Perfect. Thank you.

Romain wrote:List sent !

Phowar,I've listened to 120 songs so far still some to go, let alone ranking them

spiritualized wrote:

Romain wrote:List sent !

Phowar,I've listened to 120 songs so far still some to go, let alone ranking them

Likewise. I've given 135 an initial rank. Long way to go.

How did 'Fade to Grey' make it into this poll? Surely this was never unacclaimed.

Submitting vote now.

sorry it took so long. I have been very busy.

I'd like to participate, but I'm concerned that if I don't submit a full list, the songs I don't include will be unfairly penalised.

I only participated in Unacclaimed 3 (so I've got all of these rated), and I'd try to get through as many more as I can. But I doubt I'll get to all the rest while giving each a fair listen (i'd want to listen to each at least a few times).

Of the options proposed, I think option 2 would be better - but it still looks like it could unfairly penalise songs not voted for. Can you please clarify the scoring?

Sent mine. I'm all about Option 1 ... I've listened to everything, and only ranked what I consider to be very good. Everything else I felt the same, but felt force-ranking others' favorites would be wrong. This should just be the cream of the crop of what we feel was unacclaimed (though I worry a little that older polls will be impacted, as a lot of folks here now were not around then).

Just sent my rankings. I ranked them all. Still think scoring option 1 is fairest.

Update: I've been so busy with work (and the Prince poll), that I feel I won't be ready to post any results for awhile. How about we set a new deadline of May 31st?

I've also done a fair amount of thinking about the scoring system, and my preference now is for an Option 3!

I feel like most of the systems we use impose an implicit "preference distribution". The assigned point gap between everyone's 1st and 2nd listed items will be bigger than the gap between their 2nd and 3rd listed items, but what if that's not true? What if someone has an obvious top 2 for a particular list (say, an obvious top 2 songs from 1989), but the gap to the third item is bigger? Most of our scoring systems don't allow for that (I can only think of Pazz n' Jop style point systems which do).

Moreover, because for this poll we are all scoring a pre-identified set of 320 very specific songs, some of which we may love while others we may not, rather than providing a list of our X favorite songs, I think the variation in each participant's preference distribution will be even starker than it is when we list our favorite X albums/songs of all time or for a particular year/decade/genre/artist.

With that in mind, I currently have a strong preference for lists of RATINGS on a 0-10 scale, with 5 being neutral. There are no other rules - you can rate as many songs a 10 that you like, or as many a 0 as you like, and you can get as precise as your hearts desire: a rating of 8.973 is more than welcome if it seems right to you.

The points for each song would be something like: (Average rating - 5)*ln(# votes). Hence, songs with a higher number of ratings have a higher multiplier, but the multiplier tapers off pretty rapidly such that the average rating becomes more important rather quickly.

This means that unrated songs are treated... as unrated songs - they aren't assigned any score/rank if you don't rate them.

This gives participants the control over how to distribute their ratings, and it doesn't entail a nagging feeling that unrated songs will be either unfairly punished or unfairly promoted, depending on how much you like the songs you have rated. As I have not received many lists yet, I do wonder whether the sheer size of the set of songs and concerns about how unranked songs will be treated could be a deterrent to participation.

Would those who have already sent lists be open to sending a list of such ratings instead?

Moonbeam wrote:Update: I've been so busy with work (and the Prince poll), that I feel I won't be ready to post any results for awhile. How about we set a new deadline of May 31st?

I've also done a fair amount of thinking about the scoring system, and my preference now is for an Option 3!

I feel like most of the systems we use impose an implicit "preference distribution". The assigned point gap between everyone's 1st and 2nd listed items will be bigger than the gap between their 2nd and 3rd listed items, but what if that's not true? What if someone has an obvious top 2 for a particular list (say, an obvious top 2 songs from 1989), but the gap to the third item is bigger? Most of our scoring systems don't allow for that (I can only think of Pazz n' Jop style point systems which do).

Moreover, because for this poll we are all scoring a pre-identified set of 320 very specific songs, some of which we may love while others we may not, rather than providing a list of our X favorite songs, I think the variation in each participant's preference distribution will be even starker than it is when we list our favorite X albums/songs of all time or for a particular year/decade/genre/artist.

With that in mind, I currently have a strong preference for lists of RATINGS on a 0-10 scale, with 5 being neutral. There are no other rules - you can rate as many songs a 10 that you like, or as many a 0 as you like, and you can get as precise as your hearts desire: a rating of 8.973 is more than welcome if it seems right to you.

The points for each song would be something like: (Average rating - 5)*ln(# votes). Hence, songs with a higher number of ratings have a higher multiplier, but the multiplier tapers off pretty rapidly such that the average rating becomes more important rather quickly.

This means that unrated songs are treated... as unrated songs - they aren't assigned any score/rank if you don't rate them.

This gives participants the control over how to distribute their ratings, and it doesn't entail a nagging feeling that unrated songs will be either unfairly punished or unfairly promoted, depending on how much you like the songs you have rated. As I have not received many lists yet, I do wonder whether the sheer size of the set of songs and concerns about how unranked songs will be treated could be a deterrent to participation.

Would those who have already sent lists be open to sending a list of such ratings instead?

The May 31st deadline is great news! (I only have about 50 songs left to rank and had planned on completing them this weekend. However, I also need to finish my 2002 list, so I was wondering how I was going to manage all this. Thanks for the extension!)

The one thing that always concerns me about going by ratings instead of rankings is that I think rating typically rewards those with a consensus as being "good" while penalizing more polarizing songs (those that are considered masterpieces by some, while others absolutely hate them). The best parallel I can come up with is the difference between the critics lists of the greatest films of the year when compared to the percentage at Rotten Tomatoes. (I think this became obvious to me one year when "Toy Story 2" was at about 100% on Rotten Tomatoes, but didn't even rank in the top 20 of a critical consensus list.) Since my absolute favorite songs are likely to be at the very top or very bottom of many lists (especially modern classical pieces, which may not appeal at all to some voters), I worry that these will be hurt while songs that everyone gives around an 8 or so will end up being those at the top.

With that said, I'll be happy with whatever you decide. Since I nearly have all the songs ranked already, I'll just add another column for my 1-10 ratings so you'll have everything you need for whichever option is chosen. I just really appreciate you hosting this!

jamieW wrote:

Moonbeam wrote:Update: I've been so busy with work (and the Prince poll), that I feel I won't be ready to post any results for awhile. How about we set a new deadline of May 31st?

I've also done a fair amount of thinking about the scoring system, and my preference now is for an Option 3!

I feel like most of the systems we use impose an implicit "preference distribution". The assigned point gap between everyone's 1st and 2nd listed items will be bigger than the gap between their 2nd and 3rd listed items, but what if that's not true? What if someone has an obvious top 2 for a particular list (say, an obvious top 2 songs from 1989), but the gap to the third item is bigger? Most of our scoring systems don't allow for that (I can only think of Pazz n' Jop style point systems which do).

Moreover, because for this poll we are all scoring a pre-identified set of 320 very specific songs, some of which we may love while others we may not, rather than providing a list of our X favorite songs, I think the variation in each participant's preference distribution will be even starker than it is when we list our favorite X albums/songs of all time or for a particular year/decade/genre/artist.

With that in mind, I currently have a strong preference for lists of RATINGS on a 0-10 scale, with 5 being neutral. There are no other rules - you can rate as many songs a 10 that you like, or as many a 0 as you like, and you can get as precise as your hearts desire: a rating of 8.973 is more than welcome if it seems right to you.

The points for each song would be something like: (Average rating - 5)*ln(# votes). Hence, songs with a higher number of ratings have a higher multiplier, but the multiplier tapers off pretty rapidly such that the average rating becomes more important rather quickly.

This means that unrated songs are treated... as unrated songs - they aren't assigned any score/rank if you don't rate them.

This gives participants the control over how to distribute their ratings, and it doesn't entail a nagging feeling that unrated songs will be either unfairly punished or unfairly promoted, depending on how much you like the songs you have rated. As I have not received many lists yet, I do wonder whether the sheer size of the set of songs and concerns about how unranked songs will be treated could be a deterrent to participation.

Would those who have already sent lists be open to sending a list of such ratings instead?

The May 31st deadline is great news! (I only have about 50 songs left to rank and had planned on completing them this weekend. However, I also need to finish my 2002 list, so I was wondering how I was going to manage all this. Thanks for the extension!)

The one thing that always concerns me about going by ratings instead of rankings is that I think rating typically rewards those with a consensus as being "good" while penalizing more polarizing songs (those that are considered masterpieces by some, while others absolutely hate them). The best parallel I can come up with is the difference between the critics lists of the greatest films of the year when compared to the percentage at Rotten Tomatoes. (I think this became obvious to me one year when "Toy Story 2" was at about 100% on Rotten Tomatoes, but didn't even rank in the top 20 of a critical consensus list.) Since my absolute favorite songs are likely to be at the very top or very bottom of many lists (especially modern classical pieces, which may not appeal at all to some voters), I worry that these will be hurt while songs that everyone gives around an 8 or so will end up being those at the top.

With that said, I'll be happy with whatever you decide. Since I nearly have all the songs ranked already, I'll just add another column for my 1-10 ratings so you'll have everything you need for whichever option is chosen. I just really appreciate you hosting this!

A similar formula which I use for ranking artists by song ratings sounds like it could assuage this somewhat:

(Weighted average - 5)*ln(# of votes + a),

where a is a small number and weighted average is determine like this:

a.ratings = ratings - 5
s.ratings = sign(a.ratings)
r.sum = sum(s.ratings*a.ratings^2)/length(ratings)
wt.av = sign(r.sum)*sqrt(abs(r.sum)) + 5

Let's consider song A, with ratings of 10 and 3, and song B, with ratings of 8 and 8.

Song A:

a.ratings = {5, -2}
s.ratings = {1, -1}
r.sum = (1*(5)^2 + (-1)*(-2)^2)/2 = (25 - 4)/2 = 21/2 = 10.5
wt.av = (1)*sqrt(10.5) + 5 = 8.24

Song B:

a.ratings = {3, 3}
s.ratings = {1, 1}
r.sum = (1*(3)^2 + (1)*(3)^2)/2 = (9 + 9)/2 = 18/2 = 9
wt.av = (1)*sqrt(9) + 5 = 8

So Song A would have a higher weighted average - that score of 10 really boosts it.

I concede that a different scoring system would be preferable. Although I'm a bit unsure as I rank by preference and don't normally attempt to score albums or songs. But I'm sure I could give it a go (and would gain pleasure from giving the three songs at the bottom of my list a zero - don't worry Jamie they are not modern classical).

Moonbeam wrote:
A similar formula which I use for ranking artists by song ratings sounds like it could assuage this somewhat:

(Weighted average - 5)*ln(# of votes + a),

where a is a small number and weighted average

I trust your judgment, Moonbeam. I've been amazed at your statistical prowess, from the '80's poll (where you provided the voter correlations, one of my favorite features of any poll) to that incredible album ranker you designed (which has helped me immensely in the past). Whichever formula you use, I know it will produce great results.

DaveC wrote:I concede that a different scoring system would be preferable. Although I'm a bit unsure as I rank by preference and don't normally attempt to score albums or songs. But I'm sure I could give it a go (and would gain pleasure from giving the three songs at the bottom of my list a zero - don't worry Jamie they are not modern classical).

LOL, that's a relief, Dave! I tend to be pretty stingy with my very best ratings and very generous with my "low" ratings. In fact, since I have rated all the songs I've ranked so far, I've only given one song below a 4 (and that's a 3). I tend to reserve a 1 for something that offends me (which isn't easy to do), such as "Blurred Lines."

Sounds good Moonbeam. And I'm glad for the extended deadline - hopefully I'll get through them all.

Just submitted mine, with rankings and ratings, so either option should work. I also had the same concern as jamieW, but it sounds like formula handles that pretty well.

So, I have a lot of trepidation with this new methodology, for a couple main reasons.

1. Even though I've listened to all of these over the years, I did not rank in the first go-round what I did not initially find memorable, in addition to those few songs I flat-out did not like. In this new method I would feel obliged to intently relisten to a bunch of songs I didn't care for a whole lot when I went through them initially, which is really not something I care to spend time on. (Yes, that's kind of selfish.)

2. (Related to the above, and more important) ... I have a lot of discomfort in giving low ratings to songs that I know others on this forum love, in particular those that are hidden, unacclaimed classics, particularly because I believe these songs have personal meaning to those who nominated them. It just doesn't feel right to me.

This is why I think the initial weighting method is best ... just have people pick the cream of the unacclaimed crop, and see what rises to the top. To me the biggest risk of doing it this way is that the more recent polls will be over-valued due to the changing board membership over time, but that still outweighs my general lack of excitement over individually rating everything, based on the reasons I gave. On the other hand, I would of course love to give 10s to all my faves, but even that seems like a bit of a cop-out.

Moonbeam, by the way, I'm not criticizing your methodology and certainly not the statistical soundness ... I trust you with that implicitly. I just think the unacclaimed stuff is a different animal. If we were giving ratings to, say, the Top 500 AM songs, knowing that most of us have heard them and have familiarity, I would be much less reluctant.

I'll need a bit of time to collate results, so keep submitting lists if you've got them!

So, I have felt bad about letting this project slide. To put it shortly, my personal life got really hard shortly after this, and I became quite depressed. Things are looking up again, so I'm happy to revisit this.

However, there are still a couple of issues. One is my fault: of the lists I received, some had ranked the songs, and others had provided ratings on a 0-10 scale. Some provided both, as well. The other issue is that there are only 10 ballots.

With this in mind, I'm happy to consider posting results based on ranks (I can create ranks based on the ratings as needed), though it seems a little lacklustre with only 10 ballots. I'll consider any feedback over the next couple days. At worst, I'll post results based on the 10 lists I have.

Moonbeam wrote:So, I have felt bad about letting this project slide. To put it shortly, my personal life got really hard shortly after this, and I became quite depressed. Things are looking up again, so I'm happy to revisit this.

I hope the light will become brighter and brighter for you.

Take your time and do what you want for these lists, it will be perfect.