This is a revised poll question, since someone pointed out that the first poll was a bad comparison of quality vs. popularity. The purpose of the poll was to get opinions on what is more meaningful, counting opinions that are registered explicitly via voting, or an implicit measure made by counting the number of times a node has been viewed.
I think that time is also a factor... Maybe "rate of votes"? This is why I like the idea of the Radioactivity Module...
A post could have got 1000 votes in its 1 year life and another have 10 in its one day life. In their respective lives, the 10-in-a-day post effectively has 3650-in-a-year... So which is more popular?
Are you counting anonymous voting? Anything anonymous (view count or voting) is going to be open to gaming, so isn't necessarily "meaningful".
04jibbajabba
04/03/08 @
11:14
Nicholas Thompson: Yes agreed.
alf: Agreed on that note as well. An ideal situation is a closed ecology where votes or page views can't be duplicated. Wondering how often node views is gamed as well?
Anything that requires a user action provides more information about popularity. Still, what kind of popularity are you measuring?
As I write this there are already 4 comments and 6 votes on the poll. That means that 6 people gave enough of a damn about this article to *do* something. Its an additional metric that can be used to calculate popularity.
In an +1 / -1 scheme even a score of zero or negative 1000 tells you a lot about popularity.
Imagine a page providing support information for gadget X. At the bottom of the page is a simple question. "Did you find this article useful?" with a +1/-1 voting tool.
Logs show that the page has been viewed 150,000 times and the vote 'score' is -3,000. AND that the content had been voted on 9,000 times.
There is no question that the page is popular. Very popular (obviously gadget X has a common flaw people are trying to find a solution for) based on raw pageview numbers. And 9,000 people took time to vote - caring enough about the content to voice an opinion.
But what about the -3,000 'score'? The "Page" is popular, but the "Content" isn't. AND your "Product" (gadget X) is losing popularity FAST. Looks like a time for a re-write before you make your customers any more angry than they might already be.
Advertisement
06jibbajabba
04/03/08 @
14:10
Andre: Your example describes a very specific type of situation where the number of views might definitely mean something is popular, but the quality of the content could be very low.
Here's the backstory. On this konigi blog I started, I post blog entries of examples of good design and good interfaces. On the index pages in the RSS feeds, you only see thumbnails of screenshots, so you have to actually view the entry to see what the screenshot is for. I had anonymous +1 voting, but in this situation I did think that the numbers were being gamed.
I also was experimenting with 5 star rating of another node type, interfaces. I felt that was working pretty well, actually, and views did seem to correspond with star rating to some degree. But I have a problem with star rating, based on usability testing sessions I've done in the past with that interface. Unless people understand the system to give them something back (a la Netflix's star rating affecting recommendations), then many people tend to simply be 5 star/1 star people, rarely giving anything otherwise. Again, this is on systems where there's little incentive to rate.
So that's the story. I don't know if rating/voting was doing much for Konigi, so I opted to kill rather than analyze. But after the fact, I'm just curious about what people's opinions and perceptions are on the issue.
Comments
04/03/08 @ 08:06
This is a revised poll question, since someone pointed out that the first poll was a bad comparison of quality vs. popularity. The purpose of the poll was to get opinions on what is more meaningful, counting opinions that are registered explicitly via voting, or an implicit measure made by counting the number of times a node has been viewed.
04/03/08 @ 09:37
I think that time is also a factor... Maybe "rate of votes"? This is why I like the idea of the Radioactivity Module...
A post could have got 1000 votes in its 1 year life and another have 10 in its one day life. In their respective lives, the 10-in-a-day post effectively has 3650-in-a-year... So which is more popular?
04/03/08 @ 10:13
Are you counting anonymous voting? Anything anonymous (view count or voting) is going to be open to gaming, so isn't necessarily "meaningful".
04/03/08 @ 11:14
Nicholas Thompson: Yes agreed.
alf: Agreed on that note as well. An ideal situation is a closed ecology where votes or page views can't be duplicated. Wondering how often node views is gamed as well?
04/03/08 @ 12:15
Anything that requires a user action provides more information about popularity. Still, what kind of popularity are you measuring?
As I write this there are already 4 comments and 6 votes on the poll. That means that 6 people gave enough of a damn about this article to *do* something. Its an additional metric that can be used to calculate popularity.
In an +1 / -1 scheme even a score of zero or negative 1000 tells you a lot about popularity.
Imagine a page providing support information for gadget X. At the bottom of the page is a simple question. "Did you find this article useful?" with a +1/-1 voting tool.
Logs show that the page has been viewed 150,000 times and the vote 'score' is -3,000. AND that the content had been voted on 9,000 times.
There is no question that the page is popular. Very popular (obviously gadget X has a common flaw people are trying to find a solution for) based on raw pageview numbers. And 9,000 people took time to vote - caring enough about the content to voice an opinion.
But what about the -3,000 'score'? The "Page" is popular, but the "Content" isn't. AND your "Product" (gadget X) is losing popularity FAST. Looks like a time for a re-write before you make your customers any more angry than they might already be.
04/03/08 @ 14:10
Andre: Your example describes a very specific type of situation where the number of views might definitely mean something is popular, but the quality of the content could be very low.
Here's the backstory. On this konigi blog I started, I post blog entries of examples of good design and good interfaces. On the index pages in the RSS feeds, you only see thumbnails of screenshots, so you have to actually view the entry to see what the screenshot is for. I had anonymous +1 voting, but in this situation I did think that the numbers were being gamed.
I also was experimenting with 5 star rating of another node type, interfaces. I felt that was working pretty well, actually, and views did seem to correspond with star rating to some degree. But I have a problem with star rating, based on usability testing sessions I've done in the past with that interface. Unless people understand the system to give them something back (a la Netflix's star rating affecting recommendations), then many people tend to simply be 5 star/1 star people, rarely giving anything otherwise. Again, this is on systems where there's little incentive to rate.
So that's the story. I don't know if rating/voting was doing much for Konigi, so I opted to kill rather than analyze. But after the fact, I'm just curious about what people's opinions and perceptions are on the issue.
04/03/08 @ 22:30
hmm... both. Combine them.
04/04/08 @ 00:16
Well, yeah, that's how I was doing it in Drupal's Views module. First sort option was vote count and second was node views.
Post new comment