Star Vote Problems 9.6 - 9.7

[not2serious]

There are problems with the star voting by an anonymous user. See the attached images to document the problem.

Image 1 - Shows the story with the five voted as created by the user posting story.

Image 2 - Shows the hoover over the 4th star by an anonymous users as the vote.

Image 3 - Shows the results of the anonymous vote of the 4th star. This shows two votes with an average score of 3.25.


1. This only occurs when there is an anonymous user who votes.

2. Immediately after clicking on the 4th star the image is refreshed to the voted image of 3 stars.

3. After uploading the 9.7 vote.php file, the condition was not corrected.

4. The beatniak fix in SVN 1168 did not correct the problem.

5. Every time an anonymous user votes, the vote is recorded as one or more stars less that the highlighted intended vote.


Browser: IE6 , none others testes
Pligg Version: 9.6, 9.6 with 9.7 vote.php and 9.6 with SVN 1168.

Please advise if there are any corrective measures available.

Thank you.

[beatniak]

Originally Posted by not2serious

4. The beatniak fix in SVN 1168 did not correct the problem.

SVN 1168 just made the vote.php work again for voting method = 2. Nothing else.


But why the 3 stars after an anonymous vote???
one word: karma

Having karma turned on does some strange stuff with the voting method 2:

First you have to know this:
1 star = link_rating 2
5 stars = link_rating 10

normal voting 2 formula: star_rating = link_rating/2

voting 2 with karma: star_rating = (link_rating/2) * (karma/10)
//don't know about this formula for sure, but it's something like this

Anonymous has lowest karma of all. So when anonymous (let's say karma =2) votes 4 stars (link_rating = 8):
star_rating = (8/2) * (2/10) = 4 * 0.2 = 0.8

So why the 3 stars after an anonymous vote?
5 stars from the 1st vote + 1 star from anon vote = 6 stars / 2 votes = 3 stars

[tbones]

That´s the reason why I don´t use karma and bury with voting method 2 *g*

I you find the time, could you try to find a fix for the live page (v9.5 Live view: votes miscalculated) with the above mentioned simplifications?

Edit: If karma makes no sense for five-star-installations where can I comment out the display. especially the row at top users?

[beatniak]

- Fixed the voting 2 values for the live pages (see thread v9.5 Live view: votes miscalculated). Added this code to the SVN in revision 1201.

- Comment out the karma display? Dude... don't be lazy and do some code digging.

[tbones]

Originally Posted by beatniak

Comment out the karma display? Dude... don't be lazy and do some code digging.

I´ll try my very best!

[not2serious]

Beatniak,

Thank you, for your comprehensive response to my post. Based on that response, it becomes clear that the problem is not karma, but is the formula that is being used to count the contribution of the anonymous users vote. If properly calculated, karma is a very good feature and a great marketing tool. Maybe, I could influence you to help with the following revision to the star voting calculation.


Pligg Community,

Essentially, the five star vote is a review, appraisal or valuation (I have a company that performs these services). The use of karma in the calculation, of the five star voting method, is to create a weighted average (Weighted mean - Wikipedia, the free encyclopedia). This MUST be accomplished by giving different weights for the user vote, NOT by changing their vote (the current way the star rating is calculated with karma active).

Some of the principals desired by the Pligg community are:

1. The anonymous user vote is less reliable, and should be assigned less weight, than a logged in registered user.

2. Anonymous votes can artificially skew the vote total. Especially, when given equal weight or using the current karma method.

3. The logged in register user, should have more clout than an anonymous user, and must be given greater weight.

4. There should be a reward for being a registered user. That reward being a greater affect on a story success or failure.


The formula used is not correct.

Anonymous has lowest karma of all. So when anonymous (let's say karma =2) votes 4 stars (link_rating = 8):
star_rating = (8/2) * (2/10) = 4 * 0.2 = 0.8

This example clearly demonstrates how an anonymous vote can significantly and adversely affect the vote total. If the intention is to give the anonymous voter 20% of the weight of the register user with karma=10, this formula fails miserably.


Simple average is not equitable.

So why the 3 stars after an anonymous vote?
5 stars from the 1st vote + 1 star from anon vote = 6 stars / 2 votes = 3 stars

This simple averaging of the votes, still provides too much power to the anonymous voter. And, there is no incentive to be a register user. This formula is correct, but I am sure that it is not the way a knowledgeable developer would want the star votes to work when they allow anonymous users to vote.


How to FIX the weighted average voting and use karma.

Here is a quick way to accomplish this:

First you have to know this:
1 star = link_rating 2
5 stars = link_rating 10

Delete this calculation function. It is not needed to perform a weighted average. I can see the attempt is to work from a percentage. However, this is not the way to create a weight average.

To demonstrate; lets use the following data of users, votes and karma:

Anonymous 1 = 1 stars, karma = 2
Anonymous 2 = 4 stars, karma = 2
Anonymous 3 = 4 stars, karma = 2
Anonymous 4 = 1 stars, karma = 2
Beatniak = 5 stars, karma = 10
Not2Serious = 4 stars, karma = 8.75

The karma for the anonymous user has been coded as 2 and is consistent for all anonymous users on the site. In this case the karma for the registered users is tracked and maintained by the site. The karma acts as the weighting factor base on the user’s activity. The higher the karma, the greater the influence, of the vote average.


Let’s see how the calculation should be made.

User Stars * karma = weighted vote
Anonymous 1 = 1 * 2 = 2
Anonymous 2 = 4 * 2 = 8
Anonymous 3 = 4 * 2 = 8
Anonymous 4 = 1 * 2 = 2
Beatniak = 5 * 10 = 50
Not2Serious = 4 * 8.75 = 35
Sum 26.75 113

Weighted average calculation: weighted vote / karma (SUM) = weighted average
113 / 26.75 = 4.22 stars



Advantages of using a Weighted Average with karma.

1. The karma will function properly with the star ratings.

2. The votes are given fair weight which is based on the user participations (the intended use of karma).

3. The weight on the anonymous user vote could be adjusted by the admin.

4. The weighted average method does not penalize the post by the activities of the malicious anonymous user.

5. At the same time, it does increase the vote count by the anonymous user voted and encourages site activity.



OK, I want to use the weighted average method.

Nothing would please me more that to offer the Pligg Community the code to accomplish this proposed calculation revision. However, I have a big problem, I am not a coder. I have been studying various sources of material appropriate to learning more to accomplish the task. It will be some time before I am ready to attempt this task. But, I know there is some great talent on this forum, so maybe someone here would be willing to get this calculation method into proper operation. I would be more than willing to help with formulas and test. It would be great not to have to comment out the karma calculation (besides, I don’t know where to find it in the code).


Karma, is good, lets use it.

If someone would explain (like the way beatniak did the vote calculations) the karma calculations, I would be happy to see if a revision in the weighting method may be appropriate.

[tbones]

Originally Posted by tbones

I´ll try my very best!

I was successful: It´s yget/user_center.tpl and yget/topusers_center.tpl. I just wonder whether there is now so much space with the last column while the name column doesn´t fit for long names.

[doct]

Hello everybody!
I'd like to know if anyone has solved this: a weighted average 5-star voting method would be very useful to me...