SummitPost.org

Bob Sihler wrote:...Personally, I don't put much stock in scores except for the very low ones. After all, Mount Rainier is going to get way more hits and votes than an excellent page for some obscure desert peak. In other words, I don't think a high page score necessarily means the page is better than many others, so I wouldn't be upset to see actual scores go.

As there is no doubt that scores do not do justice to the quality of pages, I wonder if it would be possible to display – beside the page score – the actual average grade, e.g.
http://www.summitpost.org/ironton/836476 Page Score: 78.87% Average Grade: 9.9
http://www.summitpost.org/chamber-of-the-basilisk-technical-route/836000 Page Score: 72.64% Average Grade: 9.67
http://www.summitpost.org/traverse/835810 Page Score: 72.09% Average Grade: 7.75

The above is a good example of why the new scoring system needs improvement.

The page below isn't that good of a page (it has pretty photos, but the information is really thin; maybe it isn't a really bad page, but it isn't really good either):

http://www.summitpost.org/traverse/835810

Yet, no matter what the next person votes (even if meant to be constructive), the score won't change hardly at all.

Out of 12,267 pages, it seems impossible to get a score between 20-69%.

Voting is supposed to be for quality control (not ego, to make someone feel good or whatever), but now if a page needs improvement, a vote is almost meaningless. It was already bad enough that many people would vote 10/10 on poor pages on the old system, but now it is worse because now even a constructive criticism vote is almost meaningless.

What I hear you saying is that votes of five or below need to really hit a pages score negatively?

Btw if a couple folks wanted to tank the score as you suggest... Just give it am even lower score.

But is the score inappropriate? That's the bottom line, and I'm not so sure it isn't. Most pages have scores of 80%+ when they have decent information, and here we have one stuck at 72%. It reflects it is a weaker page does it not? You said yourself if it not really bad, so why should it have a really bad score? The score is already lower than most everything else.

Most pages have scores of 80%+ when they have decent information, and here we have one stuck at 72%.

A lot of good pages have scores around 72%. The one below is but one example:

http://www.summitpost.org/west-ridge/834587

There are plenty of good submissions in the 70's% range.

You said yourself if it not really bad, so why should it have a really bad score?

It shouldn't, but it shouldn't have a really good one either. Why is it impossible to receive a score between 20-69%?

Scott wrote:Why is it impossible to receive a score between 20-69%?

It isn't impossible at all. But it's not going to happen when a bunch of folks give a page an 8.

That's the root of this issue. No system is going to know when you are unfairly downvoting a page unfairly vs when a group of people give 8s without much constructive thought. Sounds like your issue is more with the other voters than the system.

Sounds like your issue is more with the other voters than the system.

Much of it, yes. But, a page with a bunch of “bad” votes is rated the same as a page with 3 or 4 four good ones. I do not think it's right and is counter productive to quality control.

Is it really feasible to expect enough votes within a month to consider the page score stable?

No, but there are hundreds (perhaps thousands) of good route pages on this site that have only a few votes. The one above was just a random example.

It is possible to increase the rate at which votes under 5 decrease the score. Is that what you're suggesting?

Yes; that would be a good start.

Scott wrote:

Sounds like your issue is more with the other voters than the system.

Much of it, yes. But, a page with a bunch of “bad” votes is rated the same as a page with 3 or 4 four good ones. I do not think it's right and is counter productive to quality control.

But this page doesn't HAVE any bad votes Scott! What is the basis for your complaint?

Do you have a solution? I you have a proposal for an addition to the algorithm that doesn't present more problems than it causes I'll be the first to back it up.

Montana Matt wrote:

Scott wrote:Much of it, yes. But, a page with a bunch of “bad” votes is rated the same as a page with 3 or 4 four good ones. I do not think it's right and is counter productive to quality control.

It is possible to increase the rate at which votes under 5 decrease the score. Is that what you're suggesting?

This page has a 6, 5, 8, 7, 8, and a 6. Scott's complaint is that everyone didn't give it a 5 or less. The score looks appropriate for the votes it got to me.

Scott wrote:

It is possible to increase the rate at which votes under 5 decrease the score. Is that what you're suggesting?

Yes; that would be a good start.

So you want a page with six votes to have a crap score if one person votes a 5?

Bad idea.

But this page doesn't HAVE any bad votes Scott! What is the basis for your complaint?

The "complaint" is that no matter what the next person votes, the page score will not be affected hardly at all (unless a lot of people immediately gain up on the person and start voting down, which will probably cause hurt feelings).

This page has a 6, 5, 8, 7, 8, and a 6. Scott's complaint is that everyone didn't give it a 5 or less. The score looks appropriate for the votes it got to me.

The complaint is not that everyone didn’t give it five stars, but that a 6, 5, 7, 8, and a 6 is equal to 3 10’s. To me, it doesn't seem right.

So make a proposal.

Simply changing the weight of scores under 5 doesn't chane this, and present other problems.

What's is your recommendation?

What's is your recommendation?

Once a page receives a certain amount of votes, score should level off, rather than increase. It should be easier to have a page score between 20-69%. The scoring system needs to be more linear. Scores are too uniform even though votes are not and have a big spread. With the new algorithm, scores are grouped all together rather than spread out between 0-100%.

Everyone could come up with their own algorithm and none are perfect. Still, the old one was still better IMHO than the new one, despite being flawed.

I could try to come up with one, but everyone won't like that one either. One that allows a spread between 0-100% is best and one that levels off after a certain amount of votes are reached.

LOL! Linear! You complained a couple weeks ago because you thought votes didn't increase the score enough, and now you wan the score to be both linear AND level off? That is not possible.

So once a page has 5 votes you want future votes to be meaningless?

Propose an algorithm Scott. Otherwise you have so solution here, just complaints. Based on what you've outlined so far I do not think you will find something that meets your requirements (unless it gives your vote more weight than everyone else's).

You complained a couple weeks ago because you thought votes didn't increase the score enough

Really? Show me? It was the other way around.

===========================================================================================================================================================

Here is what I wrote:

If you take a nearly blank page (this is a random example), here is how the scoring changes when I vote:

http://www.summitpost.org/medetsiz-tepe/780188

Current score 69.51%.

The score changes to the following when I vote:

1/10: 15.02%
2/10: 68.61%
3/10: 68.88%
4/10: 69.14%
5/10: 69.39%
6/10: 69.65%
7/10: 69.90%
8/10: 70.16%
9/10: 70.40%
10/10: 70.65%

Is this intentional because the 10/10 score system is going away?

It appears the only way one could really change the score is to vote 1/10.

So once a page has 5 votes you want future votes to be meaningless?

No, only more or less level off eventually if all the votes are 10/10 and exactly the same. Too much emphasis is on the number of votes rather than the average rating of the vote.

I could try and come up with an algorithm, but give me a while.

Scott you just want your vote to count and not everyone else's.

Propose an algorithm (third time I suggested this). Or just keep whining. Up to you.