Decade charts scoring system - Suggestions

Robert Anton Wilson
Epic Proghead

Gender: Male
Age: 56
Location: Inside

LOL OK, I can do that. Works for me.

albummaster
Janitor

Gender: Male
Location: Spain
Site Admin

Over the past few weeks, option 3 has been tested in the background with the small pilot group and I'm ruling it out now as an option because the ranks are not consistent between the overall, decade and year charts.

Fergenaprido wrote:

revolver94 wrote:

Regardless of reason, maybe you could do this: have the decades charts only have up to 10 points be added per chart (for the #1 entry). This would create decimal places, but it allows for their to be a subtle difference created by the decades charts. (For year, it could go up to 1 point, since it is only 1 year).

Upon rethinking, I kind of like this idea, because then an album that isn't great enough to appear on any user's overall chart could still appear on the BEA overall chart, albeit towards the very bottom, if enough people thought it was significant in its year (or decade). Though I think the decade or year score should only count for the first appearance, and not be cumulative.

I'm now coming around to backing revolver94's idea (above) as I think it solves the problem of keeping all the charts in a consistent sequence whilst still allowing decade charts to contribute to the totals.

Fergenaprido wrote:

Bad appears at #10 in overall chart, #2 in decade chart, and #1 in year chart
Rumours appears at xx in overall chart, #4 in decade chart, and #2 in year chart
Revolver appears at xx in overall chart, xx in decade chart, and #5 in year chart

So,
Bad would get 91 points in overall chart, 99 points in decade chart, and 100 points in year chart
Rumours would get 9.7 points in overall chart, 97 points in decade chart, and 99 points in year chart
Revolver would get 0.96 points in overall chart, 9.6 points in decade chart and 96 points in year chart

So, a chart about charts Razz

First choice: album scores only affect the chart they are in (overall, decade, or year, no cross-scoring)
Second: the decimal system of revolver
Third: option 4 of joyofdivision
Fourth: option 3 of albummaster with Necharsian's check box suggestion

The first choice above is too computationally expensive. Each time a chart is updated, it could affect the scoring of other charts by the same user, so when you add a chart, the ranks of other albums by that user need to be checked/changed as well and the scores within those charts need to be updated if the album has been superseded. This could get extremely messy if other types of chart are added down the line. The advantage of revolver94's suggestion is that the calculation only needs to happen once (which was also a reason why I originally liked option 3).

So, what I am going to test next is to make decade charts contribute in a small way to the overall list. There have been 6 full decades since 1950, so let's say each decade chart contributes 20% to the total (allowing for rounding). For year charts, if a decade is worth 20% (and there are 10 years in a decade), let's say each year is worth 5% to allow year charts to mean something (when they are released).

Therefore, in a chart of 100 entries:

a top ranked album in an 'overall' chart would still contribute 100 points to the overall, decade and year; and the bottom ranked entry would be worth 1 point.
a top ranked entry in a 'decade' chart would contribute 20 points to the overall, decade, and year; and the bottom ranked entry would be worth 0.2 points.
a top ranked entry in a 'year' chart would contribute 5 points to the overall, decade, and year; and the bottom ranked entry would be worth 0.05 points.

This will keep rankings consistent between the different types of chart without distorting (very much) the overall rankings.

Taking Fergenaprido's example above:

Bad would get 91 points if it was added in an overall chart, 19.8 points from the decade chart, and 5 points from the year chart
Rumours would get 19.4 points for appearing in the decade chart, and 4.95 points for the year chart
Revolver would get 4.8 points for appearing in the year chart

The scoring is far simpler than the other options put forward - there would only be one 'overall' total which all charts would contribute to, and not separate scores for each type of chart (this is a good thing because if new types of chart are added in the future, things would get very confusing if there was a different score for each type of chart).

Last edited by albummaster on 11/25/2011 07:44; edited 1 time in total

Necharsian
Best Ever User

Gender: Male

Thanks, albummaster, for giving me all this undeserved credit, but I didn't say any of those things! Laughing

Regardless, I do like the idea, and I think having the overall chart match the decade charts is an important aspect. Is this new system being tested now?

albummaster
Janitor

Gender: Male
Location: Spain
Site Admin

Going to start testing the new system from today and see how things look. If anyone has any feedback once this is in place, let us know on this thread...

P.S. Apologies, should have credited Fergenaprido as well (now amended my post). Embarassed

Fergenaprido

Gender: Male
Location: GTA

albummaster wrote:

Going to start testing the new system from today and see how things look. If anyone has any feedback once this is in place, let us know on this thread...

P.S. Apologies, should have credited Fergenaprido as well (now amended my post). Embarassed

No worries Smile

A few questions, though.

The 20% per decade, will that change over time, as more full decades are added (yes, I'm probably thinking way too long-term for this)? Also, for incomplete decades ('30s, '40s, '10s), will the weighting be the same or is it different? I didn't quite understand that point.

As well, the system-wide BEA decade and year charts would then still be solely based on the overall ranking score, since all charts will contribute to that? I've noticed that when I view some charts (the decade ones I think), the overall score does not show up for each album. I'm guessing this method will rectify that?

For year charts, would they take effect right away, or only after the year is done? I.e. will 2011 year charts start contributing now, or would their points only come into effect January 1, 2012?

Lastly, currently any chart that has not been updated in the last 10 years does not contribute to the overall points system. Will this also be the way for decade and year charts?

As for other suggestions, I like the one someone (revolver?) made about including decade or year published charts, though I like having the option to not include them in the overall chart.

For future additional charts, such as genres, artists, or countries, I don't think they should contribute to points in the overall system. It's very clear when an album is released, and can't usually be argued if its 2010 or 2011, but genres (and sometimes countries, it seems, given the Crowded House debate Wink

) are highly subjective, and artist charts would disproportionately affect popular artists.

I think my only concern would be that more recent decades and years would be disproportionately influencing the overall chart, but I really don't see any way around that, and I think that will happen regardless of the system that is used, especially at the beginning. Over time I suppose this influence would largely peter out.

All in all, I think I like this suggestion, and it mostly makes sense to me so far. Smile

RFNAPLES
Level 8

Gender: Male
Age: 75
Location: Durham, NC, USA

KISSMIF, don't include decade, year, genre, etc charts in rankings of the overall chart. Not even sure they have to be ranked at all, if so rank them only among themselves, eg all decade chart ranks, all year chart ranks, etc.
_________________
Top 100 Greatest Music Albums by RFNAPLES
Bubbling Under The Top 100 Greatest Mus...y RFNAPLES

RFNAPLES
Level 8

Gender: Male
Age: 75
Location: Durham, NC, USA

Suppose younger members not familiar with earlier generations only post 2000s charts. Suppose further that there are few older generation charts for various reasons. Yet suppose that these same members list Revolver high on their overall chart. Would the proposed generation chart schemes artificially lower the ranking of Revolver on the overall chart?
_________________
Top 100 Greatest Music Albums by RFNAPLES
Bubbling Under The Top 100 Greatest Mus...y RFNAPLES

albummaster
Janitor

Gender: Male
Location: Spain
Site Admin

Fergenaprido wrote:

The 20% per decade, will that change over time, as more full decades are added (yes, I'm probably thinking way too long-term for this)? Also, for incomplete decades ('30s, '40s, '10s), will the weighting be the same or is it different? I didn't quite understand that point.

Every decade (at the moment) will count the same i.e. all decade charts will contribute a maximum of 20 points to the overall chart. Decade charts can only be created for 1950s onwards. 2010s is the only decade that might be contentious, so we'll just wait and see how this plays out.

Fergenaprido wrote:

As well, the system-wide BEA decade and year charts would then still be solely based on the overall ranking score, since all charts will contribute to that? I've noticed that when I view some charts (the decade ones I think), the overall score does not show up for each album. I'm guessing this method will rectify that?

I think this issue is now fixed.

Fergenaprido wrote:

For year charts, would they take effect right away, or only after the year is done? I.e. will 2011 year charts start contributing now, or would their points only come into effect January 1, 2012?

If they existed, a 2011 year chart would take effect right now. I think most people would like to create a list for the current year, and having those count would be useful to a lot of people.

Fergenaprido wrote:

Lastly, currently any chart that has not been updated in the last 10 years does not contribute to the overall points system. Will this also be the way for decade and year charts?

Hadn't thought about that until you mentioned it, not sure at the moment but there's quite a few years left to think about it Wink

Fergenaprido wrote:

As for other suggestions, I like the one someone (revolver?) made about including decade or year published charts, though I like having the option to not include them in the overall chart.

Once things settle down a bit, external decade charts could be added to the custom chart function.

Fergenaprido wrote:

For future additional charts, such as genres, artists, or countries, I don't think they should contribute to points in the overall system. It's very clear when an album is released, and can't usually be argued if its 2010 or 2011, but genres (and sometimes countries, it seems, given the Crowded House debate Wink

) are highly subjective, and artist charts would disproportionately affect popular artists.

Yes, point noted. I think we can address this when/if these types of chart are allowed to be created.

Fergenaprido wrote:

I think my only concern would be that more recent decades and years would be disproportionately influencing the overall chart, but I really don't see any way around that, and I think that will happen regardless of the system that is used, especially at the beginning. Over time I suppose this influence would largely peter out.

I think we can all keep an eye on this to make sure the D20/Y5 ratio is working and it can be tweaked if necessary.

albummaster
Janitor

Gender: Male
Location: Spain
Site Admin

RFNAPLES wrote:

Suppose younger members not familiar with earlier generations only post 2000s charts. Suppose further that there are few older generation charts for various reasons. Yet suppose that these same members list Revolver high on their overall chart. Would the proposed generation chart schemes artificially lower the ranking of Revolver on the overall chart?

Let's see how this plays out. I'm going to keep an eye on the ratios to see how many charts from each decade are added. If it gets too out of kilter, the effect could always be smoothed out using weightings. We won't know this though until there are a statistically meaningful number of decade charts in the system.

Fergenaprido

Gender: Male
Location: GTA

Okay then, it all sounds good to me. Smile

I think, no matter what, there will always be some potential for skewing the system, and some tweaks to be made, but by and large I believe you've thought of the most plausible things that can happen and have taken them into account.

Kudos.