Folks,

With regard to the Star Voting System.  I realized an irregularity when Eldave1 noticed something odd in the voting, namely large jumps in scores when voting.  I have investigated and found out why.  

Poor Math...

In the human world scores of 1, 3, 5  and 5, 3, 1 would average to  3

1+3+5=9 ; 9/3 = 3

5+3+1=9; 9/3 = 3

However, in the discussion board, the star system averages things differently.  
In the case of votes in order of 1, 3 and 5

1+3 = 4 ; 4/2 = 2 ; 2 + 5 = 7 ; 7/2 = 3.5 final score

In the case of votes in order of 5, 3 and 1

5+3 = 8 ; 8/2 = 4 ; 4 + 1 = 5 ; 5/2 = 2.5 final score

In the case of votes in order of 1, 5 and 3

1+5 = 6 ; 6/2 = 3 ; 3 + 3 = 6 ; 6/2 = 3 final score

So, the order of voting did matter.

What are we going to do about it?

Nothing.  I am working under the assumption that, in the giant scheme of things, everything kinda balances out.

I know this is a free challenge, however, we do strive to make it as professional and transparent as possible since the participants take the challenges seriously.

I apologize for this and thank you in advance for your understanding.



- Don  

Wow, that's interesting. Does this mean that the overall score at the second to last vote was averaged with the last vote?

That is, if a script had 19 5s, with a an obviously overall score of 5, and the last vote was a 1, the script would show as a 3?

That's quite unfortunate. But, what can you do?

As you said, we're all doing this for fun, so, no complaints from me.

Quoted from PKCardinal Wow, that's interesting. Does this mean that the overall score at the second to last vote was averaged with the last vote? That is, if a script had 19 5s, with a an obviously overall score of 5, and the last vote was a 1, the script would show as a 3? That's quite unfortunate. But, what can you do? As you said, we're all doing this for fun, so, no complaints from me.

Unfortunately, yes.  And vice versa - 19 1s, and the last vote a 5 would show a 3.

- Don

I've no idea if that issue makes a material difference to anyone's scores when the example is a lot more than 3 votes, but if you know the scores for each script then a real average would be simple enough, though a little time-consuming.

I work for a bank and have been known to add up every now and again so happy to do it.

Any thought to just posting the script pairings and letting us all vote a top 4?

(I don't want to complicate things for you, as this has been a ton of work for you both, and I appreciate it greatly.)

Don,

Do we have the actual scores saved up, or just the averages? If we have the actuals, I can do them tonight and have the results of this round up then begin round 4 tomorrow and just move things back one day.

Math hurts when I look at it.

Quoted from AnthonyCawood I've no idea if that issue makes a material difference to anyone's scores when the example is a lot more than 3 votes, but if you know the scores for each script then a real average would be simple enough, though a little time-consuming. I work for a bank and have been known to add up every now and again so happy to do it.

Anthony,

Unfortunately,  I don't know the votes for each script.  I don't know why it was coded this way but what happens is that a .rate file is created with the first vote, e.g.,

1VOTER

When there is a second vote, that file is opened up and the new vote - 5-  is added to the current vote and divided by 2 and the voter is appended and the file save so the file looks like this:

3VOTERVOTER2

I didn't imagine to check before it started.

- Don

Interesting. This thread shows one vote for five stars. Let's see what happens when I give it a one star....

Hold on...I'll get my 13 year old daughter to figure out this new math ;)

Quoted from Mr. Blonde Don, Do we have the actual scores saved up, or just the averages? If we have the actuals, I can do them tonight and have the results of this round up then begin round 4 tomorrow and just move things back one day.

All we have are the 'averages' (which we know are not averages). The discussion board does the math and adds it to a file with the name of the voter.  So, I know who voted for each script, but I don't know what those votes were.

- Don

I wouldn't mind submitting my scores the way we normally do. I have them down. I imagine most of us could probably recollect their scores for the scripts.

Just tested it on this thread. That's crazy.

Not something you'd have any reason to check Don... you'd assume normal averages would be used, the way it has been coded is bizarre to say the least!

Can you create a poll with all the scripts listed and PM the poll to all those who voted? I'll gladly vote again, and I'm pretty sure I remember what I gave each script.

Here, in real time is the contents of the 1523044509.rate file:

2
StevnClark
JEStaats
PKCardinal

If voter 4 gives this a 5 star rating, then it would look like this:

3.5
StevnClark
JEStaats
PKCardinal
VOTER

- Don

Quoted from khamanna I wouldn't mind submitting my scores the way we normally do. I have them down. I imagine most of us could probably recollect their scores for the scripts.

I don't have mine written down, but I'd be happy to revote.

Doing the math for this type of calculation is simple. All that's needed are numbers to work off of. =)

Quoted from Mr. Blonde Doing the math for this type of calculation is simple. All that's needed are numbers to work off of. =)

We can do that.  We also know that it isn't a direct reproduction of the original votes as folks have started to out themselves, folks will change their minds, we may have folks who didn't vote before vote now and folks who voted who don't vote again.

- Don

I'd be happy to revote if needed... could also return to head to head format, assign pairs and vote on the trio of scripts each has.

It would be very hard to vote for trios if you ask me. I'd just throw a dice most of the time.

I have sent a proposal to Sean.  Stand by.

- Don

Quoted from Don I have sent a proposal to Sean.  Stand by. - Don

Each writer must write three new 1 page scripts that can be made into a Bollywood musical. Entries due in 30 minutes.

My guess is all 7 go to the next round and only 2 may survive it! That would be some challenge.

I had 4 spare entries from last round... I think I could shoe-horn some Bollywood in there ;-)

Quoted from khamanna My guess is all 7 go to the next round and only 2 may survive it! That would be some challenge.

I think this is the best option. But only if we can all call ourselves Semi-finalists in the next round ;)

I'd prefer to be called the Midwest Regional Semi-Champion, Co-Divisional Finalist Eastern Wing Leader.

But, that's just me.

How about survey monkey? We’ll have to revote again but rereading one page won’t take long for us to remember what we gave last time.  I say reread, it’ll be more like a quick scan.

Or excell, I know the function to use if you need it.

Quoted from Gary in Houston Each writer must write three new 1 page scripts that can be made into a Bollywood musical. Entries due in 30 minutes.

And those from Austrailia are automatically out because they are sleeping and don't know the news. You snooze you lose, haha!
(or loose, whatever)

Just saw this thread before I was about to out myself assuming the voting was over - that was a close one.

If I understand it, the only thing that matters mathematically is the average of the last votes.

e.g., a script with nineteen fives and a last rating of 1 would show a rating of 3 rather than the true rating of 4.8.

A script with nineteen 1s and a last rating of 5 would show a rating of 3 rather than the true rating of 1.2.

A script with 17 ratings of 5 and the last three ratings of 1 would have a rating of 1.5 rather than the true rating of 4.4  

That is a problem. Have no idea what a rational solution would be since the votes are not recorded anywhere. The only thing I can think of is to list the scripts in a poll like they are done for normal OWCs and ask folks to re-vote.

Regardless of what happens - an enjoyable exercise anyway.

I listed each vote, so it'd be easy for me to vote again.  I also volunteer to help add up the scores.  I have no stake.

Dan

That voting system is really... broken. The board keeps track of how many votes there were, so even if it doesn't record individual votes it would be able to weight the prior vote average properly.

Unfortunately the information we'd need to reconstruct everything was destroyed. I don't have a problem with re-voting, or whatever the solution ends up being.

Don, by any chance are old revisions of the file recoverable from the server's filesystem? That could reconstruct the voting as it happened, allowing someone to build a real average.

Quoted from khamanna And those from Austrailia are automatically out because they are sleeping and don't know the news. You snooze you lose, haha! (or loose, whatever)

'Snooze you lose' - yes, I'm still half asleep here and don't know what the heck is going on here? Did everyone win?

Quoted from LC Did everyone win?

The eight who were eliminated in Round 2 will be required to write three 1-page scripts for the sadistic amusement of those who advanced to Round 3, then all of the Round 3 contestants advance to Round 4. ;)

Quoted from LC 'Snooze you lose' - yes, I'm still half asleep here and don't know what the heck is going on here? Did everyone win?

The news is pending the grinding of mathematical machinations.

Go back to sleep, there's nothing one can do.

PS - 6:00 PM here in Los Angeles - ready for dinner and a movie with the misses.  

Quoted from eldave1 The news is pending the grinding of mathematical machinations. Go back to sleep, there's nothing one can do.PS - 6:00 PM here in Los Angeles - ready for dinner and a movie with the misses.

Thank you for the cool and calm elucidation, Dave.
Nope, I slept in and this is what happens. Coming up to lunchtime now.
Speaking of movies: A Quiet Place looks like the next go to!
P.S. I think that'd be the missus, unless your other half is a bit slow on the uptake.  ;D Just kidding. Have a goodin.

Math issues huh? I thought Sean told me my OWC with the logline score, script score and readers choice was complicated!  ;D

Need a beer, Sean?  ;)

Quoted from LC Thank you for the cool and calm elucidation, Dave. Nope, I slept in and this is what happens. Coming up to lunchtime now. Speaking of movies: A Quiet Place looks like the next go to! P.S. I think that'd be the missus, unless your other half is a bit slow on the uptake.  ;D Just kidding. Have a goodin.

Thanks - me and the MISSUS are going to watch 3 billboards. Popcorn is already popping. Review to follow.

This is far beyond my pay grade, but is there a way to reverse engineer the score?  You know how many voter to give you the score leading up to the last score.  So couldn't you figure the average to get that score times the people who voted to figure out the score?

Then, you do that each time until you have all the votes.

Or, just put it the way it is for this round and fix it for next round.  It's possible that the pairings will show a clear winner regardless of the issue.  

Or scrap this round entirely and come up with a new round.

This is the first time this has been done, so it's good to get the kinks out.  

I have a few other ideas, but I'll wait to list them.

Dan

Dan, unfortunately there is a serious bug in how the forum software records votes. It’s supposed to take the number of previous voters into account when averaging in a new vote, but it always treats the existing average as if it was from one voter. The information to reconstruct the real average has been destroyed unless the server OS happens to maintain backups/versions of each file as it is overwritten.

If that huge pile of backups exists, then yes someone can work forward and reconstruct the votes based on the new average stored each time. This would reveal to that person exactly how each person voted.

I'm going to close this thread out.  Thanks to everyone for the suggestions and offers to help and understanding of the situation.  

The decision regarding how to deal with the voting irregularity has been outlined here:

http://www.simplyscripts.net/cgi-bin/Blah/Blah.pl?m-1523072283

- Don