Best Ever Albums

Re: Chart Formula Suggestions

CA Dreamin — Thu, 30 Jul 2020 16:56:29 GMT

Author: CA Dreamin

Posted: 07/30/2020 20:56

I just checked my overall chart and found my #1 album gets 99 points instead of 100; my #2 album gets 98, and so on. And I just updated it last month, so recently, and it's already started to depreciate. BEA, ya'll gotta stay on top of your overall list. Like dihansse said, reconfirm your list periodically even if you don't change any rankings.

Re: Chart Formula Suggestions

albummaster — Thu, 30 Jul 2020 05:37:45 GMT

Author: albummaster

Posted: 07/30/2020 09:37

Yes, it's the date the chart was last updated (but only for overall charts). Year/decade charts are not affected.

Re: Chart Formula Suggestions

dihansse — Thu, 30 Jul 2020 04:44:56 GMT

Author: dihansse

Posted: 07/30/2020 08:44

So the trigger for that depreciation is the last date the chart was changed (or reconfirmed)? So it makes sense to regularly reconfirm a chart (even if it doesn't change) to keep it relevant... (doesn't really apply to me because I change them all the time 😉 )

Re: Chart Formula Suggestions

albummaster — Thu, 30 Jul 2020 04:38:01 GMT

Author: albummaster

Posted: 07/30/2020 08:38

Scores depreciate daily over ten years instead of using a cliff-edge (from start of 2020).

Re: Chart Formula Suggestions

Jameth — Wed, 29 Jul 2020 17:24:55 GMT

Author: Jameth

Posted: 07/29/2020 21:24

AM,

Do scores now depreciate continuously or stepwise (every month/year/etc.)?

Re: Chart Formula Suggestions

albummaster — Sat, 4 Jan 2020 05:55:42 GMT

Author: albummaster

Posted: 01/04/2020 10:55

dihansse wrote:

The new implementation of decaying overall chart scores over time as announced by albummaster recently seems to have a fairly big effect on scores although I can't really capture the real effects yet. I suppose not for the worst.
Maybe I'll try to see the effects on the overall top 2500 chart I recently finished listening to (I think on time otherwise I'd have to do the whole thing again 🧱 🤐 😉 🤣 )

Hi, sorry for that 😄 (and for anyone else here that may have been using the list) you can use the 'View chart at point in time' toolbar (on the overall chart page) to view previous versions of the chart before the algorithm change.

Re: Chart Formula Suggestions

dihansse — Fri, 3 Jan 2020 14:55:57 GMT

Author: dihansse

Posted: 01/03/2020 19:55

The new implementation of decaying overall chart scores over time as announced by albummaster recently seems to have a fairly big effect on scores although I can't really capture the real effects yet. I suppose not for the worst.
Maybe I'll try to see the effects on the overall top 2500 chart I recently finished listening to (I think on time otherwise I'd have to do the whole thing again 🧱 🤐 😉 🤣 )
At first view about 100 new albums have entered the overall top 2500 chart of which of course also some 2019 albums and many albums of the naughts. But also older surprise ex Come Fly With Me, a 1958 Frank Sinatra album rising from rank 2587 (ie outside the top 2500) to 2162; or The Atomic Mr. Basie also from 1958 rising from 2515 to 2148.

Re: Chart Formula Suggestions

PurpleHazel — Tue, 4 Sep 2018 05:35:25 GMT

Author: PurpleHazel

Posted: 09/04/2018 09:35

baystateoftheart wrote:

I propose that once a member chart goes five years without being updated, it should no longer apply toward the overall rankings.

Playing devil's advocate a little bit, what if a member doesn't update a yearly, decade or overall chart for five years because they're happy with it, but have revised most of their yearly and decade charts in the last five years and are active, contributing members? I could see decaying it after five years, but to discount someone's yearly chart from the previous century completely because they haven't had a chance to revisit it and are active members seems a tad harsh.

Re: Chart Formula Suggestions

albummaster — Mon, 3 Sep 2018 10:49:17 GMT

Author: albummaster

Posted: 09/03/2018 14:49

baystateoftheart wrote:

Ok I'm really confused. How can To Pimp A Butterfly fall from #60 in the overall to #74 in a custom chart with a five year expiration date when it was released less than five years ago? 100% of its points are from the last five years, so going down in rank should be impossible. Something must be wrong here.

sounds like a bug in the custom chart feature which might also explain Puncture's findings. I need to check into (EDIT: excluding recognised charts seems to yield better results - I don't think the year filter is being applied to recognised charts.).

Re: Chart Formula Suggestions

baystateoftheart — Mon, 3 Sep 2018 10:22:33 GMT

Author: baystateoftheart

Posted: 09/03/2018 14:22

This is what the chart looks like with a five-year expiration date for all charts: https://www.besteveralbums.com/makechar...=RankScore

Two new albums in the top 10, five new albums in the top 50, and ten new albums in the top 100 are not insignificant changes, and at the same time, are not too drastic of an overhaul. There was no way to make only overall charts expire in the custom chart.

If we are going with AM's decay idea, instead of a hard expiration date, what would be a good half-life? I'm thinking one or two years.

EDIT: Ok I'm really confused. How can To Pimp A Butterfly fall from #60 in the overall to #74 in a custom chart with a five year expiration date when it was released less than five years ago? 100% of its points are from the last five years, so going down in rank should be impossible. Something must be wrong here.

Re: Chart Formula Suggestions

albummaster — Mon, 3 Sep 2018 09:17:52 GMT

Author: albummaster

Posted: 09/03/2018 13:17

Puncture Repair wrote:

I still believe in the spirit behind the idea though. Whether it will actually change much, maybe in 20 years.

Definitely coming around to the idea of changing something but the site has been around since 2005, so I think a slow evolution is more favourable than a drastic shift which might seem like a knee-jerk reaction. Also, need to test the impact of an algorithm change before implementing anything live to see how much it affects things (& maybe run the rankings side by side for a while).

Re: Chart Formula Suggestions

Anonymous — Mon, 3 Sep 2018 03:54:06 GMT

Author: Anonymous

Posted: 09/03/2018 07:54

baystateoftheart wrote:

I agree that everyone's opinion is equally valid. However, I think it's important that the chart reflects the people who are actually involved in the site. A lot of people put up a list of 40 and never contribute to the site after that. Adding weighting would give them an incentive to join the community.

I've been advocating this for the last 5 years or so.

https://www.besteveralbums.com/phpBB2/v...mp;start=0

As a comprise you're able to only include member charts updated in the last specified period

https://www.besteveralbums.com/phpBB2/viewtopic.php?t=11600

Turns out, even if you include only charts updated in the last 12 months, from members above level 4, it looks like, well...

https://www.besteveralbums.com/makechar...amp;page=1

When I realized this I kind of just gave up, lol.

I still believe in the spirit behind the idea though. Whether it will actually change much, maybe in 20 years.

Re: Chart Formula Suggestions

albummaster — Thu, 30 Aug 2018 06:00:10 GMT

Author: albummaster

Posted: 08/30/2018 10:00

Decade charts updated while a decade is still in progress only get a proportion of the available points e.g. a 2010s chart created today only achieves approx 80% of the maximum since the decade is approx 80% complete. Once the decade is over, that same chart will still only achieve 80% of the maximum since it was not created at the end of the decade. In addition, the points of that chart would never be decayed since the chart covers a discrete period in time (unlike a 'greatest ever' chart).

Re: Chart Formula Suggestions

baystateoftheart — Wed, 29 Aug 2018 16:47:17 GMT

Author: baystateoftheart

Posted: 08/29/2018 20:47

albummaster wrote:

Decade charts already work in the same way but omitted to mention :oops:.

Wait so do all decade charts expire after ten years? Or only the ones that were updated partway through a decade and never revised after?

I believe that year charts don't build up to full points until you update them after the year is over, but perhaps they should also have an expiration date if they're not updated after the year is over?

You make good points about the weighting, AM. Upon further reflection, I'm inclined to agree. Looking forward to the implementation of your decaying points idea 😁

Re: Chart Formula Suggestions

albummaster — Wed, 29 Aug 2018 05:29:02 GMT

Author: albummaster

Posted: 08/29/2018 09:29

Decade charts already work in the same way but omitted to mention :oops:. I think there are other ways of attracting people to the community (e.g. by modernising & improving the site, adding new features etc). I just feel adding more weight at the top of the tree is not overly democratic and there are plenty of people here for whatever valid reason don't feel the need (or don't have the time) to chase points and their voice shouldn't be deemed inferior.

Re: Chart Formula Suggestions

baystateoftheart — Tue, 28 Aug 2018 14:17:51 GMT

Author: baystateoftheart

Posted: 08/28/2018 18:17

albummaster wrote:

Currently, greatest album charts updated over ten years ago do not contribute to the rankings (year and decade charts are not affected since they cover discrete time periods and are not affected by the passage of time (no new albums can be released once a period is over).

A thought I've had for quite a while is to decay the points of a chart over the X year period so that a chart is worth more points when it is newly created to when it is a few years old and possibly missing some newer releases (the decay does not necessarily have to be linear). This might be a reasonable way of removing the effect that you describe. I'm personally not in favour of weighting the points towards higher level members because I see everybody's opinion as equally valid and it shouldn't really be based on how many points you've received.

To your point about year and decade charts, what about year and decade charts that were updated partway through the year or decade in question and then never updated after its conclusion? I would argue they should have an expiration date too.

Decaying the points of a chart over time is a great idea!

I agree that everyone's opinion is equally valid. However, I think it's important that the chart reflects the people who are actually involved in the site. A lot of people put up a list of 40 and never contribute to the site after that. Adding weighting would give them an incentive to join the community.

Re: Chart Formula Suggestions

albummaster — Tue, 28 Aug 2018 05:24:34 GMT

Author: albummaster

Posted: 08/28/2018 09:24

Currently, greatest album charts updated over ten years ago do not contribute to the rankings (year and decade charts are not affected since they cover discrete time periods and are not affected by the passage of time (no new albums can be released once a period is over).

A thought I've had for quite a while is to decay the points of a chart over the X year period so that a chart is worth more points when it is newly created to when it is a few years old and possibly missing some newer releases (the decay does not necessarily have to be linear). This might be a reasonable way of removing the effect that you describe. I'm personally not in favour of weighting the points towards higher level members because I see everybody's opinion as equally valid and it shouldn't really be based on how many points you've received.

Re: Chart Formula Suggestions

Luigii — Mon, 27 Aug 2018 22:44:18 GMT

Author: Luigii

Posted: 08/28/2018 02:44

Just caught this. And yes, this is cool. Would give an incentive to new people and maybe old users can come back from this.

Chart Formula Suggestions

baystateoftheart — Mon, 27 Aug 2018 16:57:01 GMT

Author: baystateoftheart

Posted: 08/27/2018 20:57

A lot of members who haven't been active in a long time are still having a big effect on the charts. As a result, the charts are slow to adapt to changes in member taste. I propose that once a member chart goes five years without being updated, it should no longer apply toward the overall rankings.

I also propose that users at higher levels should have more weight to their charts, implemented by a multiplier to point totals. For instance:

Level 1 = x1
Level 2 = x1.1
Level 3 = x1.2
Level 4 = x1.3
Level 5 = x1.4
Level 6 = x1.5
Level 7 = x1.6
Level 8 = x1.7

These proposed changes would be for the default chart. If people wanted to see a chart without these effects, they could use the "Customize Overall Chart" feature.

Interested to hear everyone else's thoughts on these suggestions.