My best compliments.
You did an amazing job of statistics.
But in the end I would like to ask you please, if you feel satisfied from the changes or not.
All the best.
Roberto.
Well, the answer is simple: I don’t know.
I first did this for myself to see how the new scores were related to the old ones. I could easily do this since I two times per year store the votes and scores given to my pages. As it happened I did this only two days before the change to V3 was made. This means that I have results for V2 and V3 based on identical votes. And as I did the investigation I thought maybe someone else would like to know.
/Hans
I agree with Roberto... so what is your conclusion?
I should note that page scores are independent of the object type. All other things being equal, a series of votes on a TR, article, or mountain page will give the same page score in both systems. What is different is the number of power points the score provides for an object. Also, as you demonstrate well here, the new system results in a larger range of scores (objects start lower... at 70%... but can often go higher into the upper 90s). Objects with few votes have lower scores (but more points than they would otherwise have with the same page score in v2).
Also the shape of the curve, and apparently the Mittellegi Ridge on the Eiger, is an asymptote.
And just as you theorized the kinks in the curves would be due to things like vote weights and the time impact in v2 (time has no factor in v3). The actual algorithms have no kinks.
Well, as I said: I don’t know – yet.
Thanks for the information. I see the asymptotic behavior. The chosen parameters do not seem to be optimized for images. The images on the first two pages have a score higher than 99.5 which shows them all as 100. I guess there is a simple fix for this problem.
/Hans
True... it is not optimized for the top images. Other objects were the main consideration here.
In the end it was decided having 60-70 images rounding up to essentially 100% (the actual calculation is more precise, but SP doesn't store that) was ok. There are over 600,000 images on SP and we didn't think it worth modifying things for 0.01% of images in case it had even the slightest negative impact anywhere eles. They seem to still be ranked somehow, though I am not sure how. If it becomes an issue, it can still be addressed.
EDIT... well it looks like creation date is the tiebreaker. Newest wins.
SPv2 this is obviously a 10/10. But those are much harder to come by now! Note in the vote dropdown a 9/10 is 'Wow! Amazing'. It is a fine article and I don't mean to do anything to disparage it.
To get a 10/10 it probably has to be more mountain related for me (like the article on glacier worms).
Where did you get that 3D Jungfrau region map? I had planned to do the trilogy last summer but got weathered off and only managed the Moench. I'd love one of those maps, though, to help keep me motivated to get back there.
I found the 3D model almost 25 years ago in the book store in Grindelwald. It was my first visit there also including my first visit to the summit of the Mönch. My guess is that it is hard to find one today.
Thanks for visiting,
/Hans
Chaberton - Feb 5, 2013 1:15 pm - Voted 10/10
Your Conclusions?My best compliments.
You did an amazing job of statistics.
But in the end I would like to ask you please, if you feel satisfied from the changes or not.
All the best.
Roberto.
hansw - Feb 5, 2013 1:39 pm - Hasn't voted
Re: Your Conclusions?Well, the answer is simple: I don’t know.
I first did this for myself to see how the new scores were related to the old ones. I could easily do this since I two times per year store the votes and scores given to my pages. As it happened I did this only two days before the change to V3 was made. This means that I have results for V2 and V3 based on identical votes. And as I did the investigation I thought maybe someone else would like to know.
/Hans
mrchad9 - Feb 5, 2013 1:34 pm - Voted 9/10
Yes?I agree with Roberto... so what is your conclusion?
I should note that page scores are independent of the object type. All other things being equal, a series of votes on a TR, article, or mountain page will give the same page score in both systems. What is different is the number of power points the score provides for an object. Also, as you demonstrate well here, the new system results in a larger range of scores (objects start lower... at 70%... but can often go higher into the upper 90s). Objects with few votes have lower scores (but more points than they would otherwise have with the same page score in v2).
Also the shape of the curve, and apparently the Mittellegi Ridge on the Eiger, is an asymptote.
And just as you theorized the kinks in the curves would be due to things like vote weights and the time impact in v2 (time has no factor in v3). The actual algorithms have no kinks.
hansw - Feb 5, 2013 2:09 pm - Hasn't voted
Re: Yes?Well, as I said: I don’t know – yet.
Thanks for the information. I see the asymptotic behavior. The chosen parameters do not seem to be optimized for images. The images on the first two pages have a score higher than 99.5 which shows them all as 100. I guess there is a simple fix for this problem.
/Hans
mrchad9 - Feb 5, 2013 2:15 pm - Voted 9/10
Re: Yes?True... it is not optimized for the top images. Other objects were the main consideration here.
In the end it was decided having 60-70 images rounding up to essentially 100% (the actual calculation is more precise, but SP doesn't store that) was ok. There are over 600,000 images on SP and we didn't think it worth modifying things for 0.01% of images in case it had even the slightest negative impact anywhere eles. They seem to still be ranked somehow, though I am not sure how. If it becomes an issue, it can still be addressed.
EDIT... well it looks like creation date is the tiebreaker. Newest wins.
hansw - Feb 5, 2013 2:28 pm - Hasn't voted
Re: Yes?I buy that. Back to business:
Since this page didn’t exist a few weeks ago I can not ask what your vote had been then. But I ask anyway.
/Hans
mrchad9 - Feb 5, 2013 6:10 pm - Voted 9/10
Re: Yes?SPv2 this is obviously a 10/10. But those are much harder to come by now! Note in the vote dropdown a 9/10 is 'Wow! Amazing'. It is a fine article and I don't mean to do anything to disparage it.
To get a 10/10 it probably has to be more mountain related for me (like the article on glacier worms).
pvnisher - Feb 10, 2013 9:52 pm - Hasn't voted
3D Jungfrau regionWhere did you get that 3D Jungfrau region map? I had planned to do the trilogy last summer but got weathered off and only managed the Moench. I'd love one of those maps, though, to help keep me motivated to get back there.
hansw - Feb 11, 2013 2:33 am - Hasn't voted
Re: 3D Jungfrau regionI found the 3D model almost 25 years ago in the book store in Grindelwald. It was my first visit there also including my first visit to the summit of the Mönch. My guess is that it is hard to find one today.
Thanks for visiting,
/Hans
TimB - Mar 20, 2013 8:54 pm - Voted 10/10
Re: 3D Jungfrau regionThat was going to be my question, also-where'd ya get the model?
:D
My favorite mountain in the entire world is the Eiger.
visentin - Mar 29, 2013 4:44 pm - Hasn't voted
howhow did you realize these curves ?
don't tell us you monitored every day the amount of votes and the score of each of your pages... this is sick !
hansw - Mar 30, 2013 5:52 am - Hasn't voted
Re: howPlease read my first comment.