Changes to Voting System

Suggestions and comments about SummitPost's features, policies, and procedures. Post bugs here.
User Avatar
mrchad9

 
Posts: 4545
Joined: Thu Jul 02, 2009 12:01 am
Thanked: 1338 times in 911 posts

Re: Changes to Voting System

by mrchad9 » Mon Jan 14, 2013 7:16 pm

If Matt says we can use two arguements (use one formula for 6-10 and another for 1-5 and add the results together) then this is very easy to do to handle low votes. Otherwise I need to think on it a while. Could probably come up with something that looks cleaner and hopefully accomplishes the same result (and perhaps better in the long run).

User Avatar
Bob Burd
Forum Moderator
 
Posts: 4271
Joined: Sun Jul 22, 2001 10:42 pm
Thanked: 572 times in 296 posts

Re: Changes to Voting System

by Bob Burd » Mon Jan 14, 2013 8:54 pm

The 5.5 constant is what makes votes 6 and higher add to the score, 5 and below subtract from the score. The 5.55555555 give it the rate of falloff per vote, but seems arbitrary. One or two significant digits should suffice.

User Avatar
mrchad9

 
Posts: 4545
Joined: Thu Jul 02, 2009 12:01 am
Thanked: 1338 times in 911 posts

Re: Changes to Voting System

by mrchad9 » Mon Jan 14, 2013 9:08 pm

Oh hell... sorry for the error Matt. I messed up the order of operations when aggregating some of the constants in the formula from what I worked out in Excel versus what I gave to you. The correct constant is 112.5, not 5.555555 (112.5 is 25*4.5, whereas I gave you 25/4.5).

So the correct formula is

Page score = 70 + 30 * [1-1/e^((x-5.5)*y/112.5)]

That should give the exact results of my examples.

To explain the constants:

5.5 in the formula is used to define a neutral vote (as Bob correctly explained). Thus a ten gets treated as a 4.5, a nine as a 3.5, a six as a 0.5, and a one as a -4.5 for example. This is why an eight counts about half as much to the score as a ten (actually 55% as much since 2.5/4.5 = 55%). A six counts 11% as much (0.5/4.5). And any vote under 5.5 is negative and starts lowering the score.

The other constant, 112.5 being the correct one (not 5.555555 anymore) is what affects how quickly scores increase. 112.5 is 25 * 4.5 and has the effect of increasing a page score by 4% of the remaining amount from the previous score towards the eventual goal of 100 for each 10 vote received (25 is 4% of 100 and 4.5 is the value of a 10). To increase scores faster... like 5% for each 10, use 20 * 4.5 = 90. For smaller changes with each vote the constant is increased (225 will increase scores 2% for each 10 and takes twice as many votes as the current formula). In picking a constant I tried to balance between the score not changing too rapidly if a bunch of 10s were given, but also changing enough to be noticed if only a few 7s and 8s were received.

This 112.5 constant can basically be anything depending on how much you want to change the score, but the above shows how to determine the value if you already have a desire in terms of what magnitude a single vote should result in.
Last edited by mrchad9 on Mon Jan 14, 2013 9:11 pm, edited 1 time in total.

User Avatar
Sarah Simon

 
Posts: 937
Joined: Tue Jan 02, 2007 12:01 am
Thanked: 240 times in 108 posts

Re: Changes to Voting System

by Sarah Simon » Mon Jan 14, 2013 9:11 pm

Y'all: I'm a geek...and this conversation is making my head hurt. I think I'll go back to work to relax for a while... ;)
Go climb a mountain

User Avatar
mrchad9

 
Posts: 4545
Joined: Thu Jul 02, 2009 12:01 am
Thanked: 1338 times in 911 posts

Re: Changes to Voting System

by mrchad9 » Mon Jan 14, 2013 9:16 pm

You didn't enjoy reviewing Euler's formula in calculus class?!?!?!

Sarahs head hurts.jpg
Sarah's head hurts
Sarahs head hurts.jpg (15.08 KiB) Viewed 1627 times

The following user would like to thank mrchad9 for this post
Sarah Simon

User Avatar
lcarreau

 
Posts: 4226
Joined: Thu Sep 06, 2007 10:27 pm
Thanked: 1898 times in 1415 posts

Re: Changes to Voting System

by lcarreau » Tue Jan 15, 2013 1:42 am

So ... my question was completely buried by constants and algorithms ... way back on Page 4 ... :D

lcarreau wrote:So ..... when's all this supposed to go into effect :?:
"Turkey Vultures always vomit when they get nervous."

User Avatar
Josh Lewis

 
Posts: 3412
Joined: Thu Jan 06, 2011 11:12 pm
Thanked: 1108 times in 677 posts

Re: Changes to Voting System

by Josh Lewis » Tue Jan 15, 2013 1:50 am

Either later this week or sometime next week. :) The theme for the front page is mostly done, but has a few additions that it needs.

The following user would like to thank Josh Lewis for this post
lcarreau

User Avatar
chugach mtn boy

 
Posts: 942
Joined: Sat Jun 10, 2006 7:54 pm
Thanked: 224 times in 129 posts

Re: Changes to Voting System

by chugach mtn boy » Tue Jan 15, 2013 4:16 am

A simpler version of what Chad is proposing was discussed in 2011. Whether it works better or worse, I dunno.

chugach mtn boy wrote:
Alpinist wrote:Hooboy. Are we really opening this can of worms again? The voting system has been discussed ad nauseum in the past. I was a big proponent of changing it when SP2 was first launched but it may be too late now since all of the current votes are based on a binary scale (10 or nothing).

The formula that should have been implemented is based on a weighted average 10 point scale.

((V1*P1)+(V2*P2)+(V3*P3)) / (P1+P2+P3) = S

V1, V2, V3 = Votes from 3 different people
P1, P2, P3 = Power ranking of the 3 people
S = the object's Score

In this example, only 3 people voted. Simply extend the formula for any number of people.

This would yield the results that people are expecting to see. Scores would be based on a 10 point grading scale similar to the way schools grade. If an object had a score of 88 and you voted 9, the score would always increase slightly based on your Power Ranking. Similarly, if you voted 8 on that object, the Score would always decrease. There are no quirky exceptions. The calculation would always work as expected.

The weighted average method gives members with a higher Power Ranking a greater ability to influence the Score.

This is a nice starting point, but it would mean that most objects on SP would be rated 100. If one person voted 10/10 and nobody else voted, that object would go straight to 100, same as an item with 96 votes of 10, and ahead of an item with 95 10 votes and one 9 vote.

The one thing that is good about the current system, and that ought to be preserved when it is reformed, is the concept of a "par" score set well below 100. Right now, par is 85, and you need some 10 votes to progress above that; you need a lot of 10 votes to get into the 90s.

You could keep par at 85 and create a built-in gravity that pulls items toward par unless gravity is overcome by votes. Using Alpinist's example, it would work mathematically like this (the example also tweaks one other aspect of Alpinist's formula so as to produce scores on a scale of 100, which I believe was his intent):

((10*V1*P1)+(10*V2*P2)+(10*V3*P3)+85000) / (P1+P2+P3 + 1000) = S

V1, V2, V3 = Votes from 3 different people
P1, P2, P3 = Power ranking of the 3 people
S = the object's Score

With the above example, if 3 people with power of 100 each vote, respectively, 10, 9, and 10, you end up with a score of 87.69. If they all vote 10, the score is 88.46. If they vote 10,9, 10 but one of the "10" voters is Dow and has power of 1000, the score goes to 92.72 (even though one of the 3 votes was a "9").

You can increase the "par gravity" by increasing the automatic add-on to the numerator and denominator. The system would probably work best with somewhat stronger gravity than in the above example--maybe I'd go with +850,000 in the numerator and +10,000 in the denominator. But 9 would still be a positive vote in such a system.

If you want 8's and 7's to be (marginally) positive votes, par needs to be lowered to less than 70. This is easily done. For example, the automatic add-on in the numerator could be 600,000 and the automatic add-on in the denominator 10,000, and, mission accomplished: 7-10 would be positive votes to varying degrees, at least initially; 6 would be neutral; and 1-5 would be negative. Note, however, that a 7 vote on an item that had already progressed to a high score (such as 90) would have a marginally downward effect, although much less than it currently has.
[Edited for clarity]

User Avatar
mrchad9

 
Posts: 4545
Joined: Thu Jul 02, 2009 12:01 am
Thanked: 1338 times in 911 posts

Re: Changes to Voting System

by mrchad9 » Tue Jan 15, 2013 4:37 am

That formula does not accomplish what I propose. The goal of the formula I proposed was to make 6-10 ALWAYS positive votes. I believe that any other method will gravitate towards everyone using 10/10, or at least some negative connotations being associated with lower than 10 votes. If 6 is a positive vote 100% of the time, and every time you vote it you immediately see page scores increase, then there should be less resistance to most people using them.

The following user would like to thank mrchad9 for this post
chugach mtn boy, MarkDidier, Matt Lemke, Sarah Simon

User Avatar
Sarah Simon

 
Posts: 937
Joined: Tue Jan 02, 2007 12:01 am
Thanked: 240 times in 108 posts

Re: Changes to Voting System

by Sarah Simon » Tue Jan 15, 2013 5:33 am

mrchad9 wrote:You didn't enjoy reviewing Euler's formula in calculus class?!?!?!

Sarahs head hurts.jpg


:ugeek: Dude, how fast was your propeller-beanie spinning today thrashing through all these formulae? :ugeek:
Go climb a mountain

User Avatar
Matt Lemke

 
Posts: 734
Joined: Fri Jan 01, 2010 8:34 am
Thanked: 163 times in 102 posts

Re: Changes to Voting System

by Matt Lemke » Tue Jan 15, 2013 6:01 am

Thanks Chad for doing that...glad to know we have some math geeks here! I really like your algorithm and I think it will work well.

I assume there won't be a like/dislike system then? I am really excited to see the changes. :)

The following user would like to thank Matt Lemke for this post
mrchad9

User Avatar
mrchad9

 
Posts: 4545
Joined: Thu Jul 02, 2009 12:01 am
Thanked: 1338 times in 911 posts

Re: Changes to Voting System

by mrchad9 » Tue Jan 15, 2013 6:13 am

Matt Lemke wrote:I assume there won't be a like/dislike system then?

Not sure. But based on Matt's comments in another thread, I was inclined to believe dropping in a new formula was less work for Matt than changing to a new system... even if two votes seems simpler than a 10 vote system on the surface.

It looks promising to me! Which is great news considering how much this voting system has bothered so many of us for so long. I'd nearly lost hope. I hope more members start voting in the future now that every page isn't automatically between 87 and 90%.

While I don't pay too much attention to page scores and rankings due to the current setup, I put a lot of work into pages and I admit it pleases me when 40+ people make the small effort to acknowledge it by voting on one.

User Avatar
chugach mtn boy

 
Posts: 942
Joined: Sat Jun 10, 2006 7:54 pm
Thanked: 224 times in 129 posts

Re: Changes to Voting System

by chugach mtn boy » Tue Jan 15, 2013 6:39 am

Chad has devised something quite clever here. I also like his add-on idea, one page up, of making very low votes especially powerful--if two or three people really ding a page, it's usually because it has some colossal flaw that knowledgeable people can spot, and this should outweigh the the mindless 10/10s.

I would also favor incorporating a power points element so that major contributors have more say than casual drop-ins.

The following user would like to thank chugach mtn boy for this post
mrchad9

User Avatar
Noondueler

 
Posts: 220
Joined: Tue Aug 21, 2007 6:55 am
Thanked: 71 times in 28 posts

Re: Changes to Voting System

by Noondueler » Tue Jan 15, 2013 7:16 am

All I can say is with the current voting system it seems absurd that one 9/10 vote can have a major impact on a pic or page with, say, 20 10/10 votes! :o
It's like if you get a less than 10/10 vote the voter appears to be deliberately undermining all the other votes. :twisted:
And one 1/10 vote....forget about it! :lol:
Last edited by Noondueler on Tue Jan 15, 2013 7:42 am, edited 1 time in total.

The following user would like to thank Noondueler for this post
Alberto Rampini, Marcsoltan, mrchad9, Silvia Mazzani

User Avatar
mrchad9

 
Posts: 4545
Joined: Thu Jul 02, 2009 12:01 am
Thanked: 1338 times in 911 posts

Re: Changes to Voting System

by mrchad9 » Tue Jan 15, 2013 7:27 am

Tomorrow I'll see if I can add some element to the formula so that the negative side of the votes is handled more powerfully like Bob and Chris requested. I hope I can do it elegantly by adding to the formula, rather than having two seperate formulas for positive and negative votes. We'll see.

My tendency is to only have the result change when the net result is negative (in other words when the simple average of the votes is less than 5.5). Also this means that a 1 in most cases would only cancel out a 10. This would be easier, and I think also limits the need for outlier low votes to need to be dropped (though it may be appropriate to still drop in extreme examples, I won't try to weigh in on that). However, if I can think of a way to increase there power without having to only deal with the net result I will. In reality I don't think this should happen too much... with an active set of content elves pages that are this bad shouldn't last very long unless turned into custom objects.

Matt two questions... Is it an issue if the formula goes negative? (or in this case can we simply have the score show as zero and give no power points). Also do you have a clearer picture on the extra server load if there is one formula for 6-10 and another for 1-5? This makes the solution easier and more flexible in terms of reaching the proposed treatment for poor votes... but since it is applicable to a low percentage of pages I don't want it to be too costly.

PreviousNext

Return to Site Feedback

 


  • Related topics
    Replies
    Views
    Last post

Who is online

Users browsing this forum: No registered users and 0 guests