We started off with a pretty high maximum for now. We'll be checking voting data in the coming weeks and evaluating how the changes are affecting guide scores across the site. Adjusting these numbers is always possible based on what's best for the site.
With a max of 60 votes, individual downvotes have a pretty small impact on guide scores. However, groups of downvotes will have a more noticeable effect than before this change. For a guide with over 100 votes in a month, it was nearly impossible to reduce its score without a large influx of downvotes. Now it just requires a few to start moving.
Getting downvotes sucks as an author, but it's an important part of the system since it helps organize guides and gives other people a chance at the top spot. There's a certain amount of malleability that's necessary for guide scores to accurately reflect a guide's quality over time.
With a max of 60 votes, individual downvotes have a pretty small impact on guide scores. However, groups of downvotes will have a more noticeable effect than before this change. For a guide with over 100 votes in a month, it was nearly impossible to reduce its score without a large influx of downvotes. Now it just requires a few to start moving.
Getting downvotes sucks as an author, but it's an important part of the system since it helps organize guides and gives other people a chance at the top spot. There's a certain amount of malleability that's necessary for guide scores to accurately reflect a guide's quality over time.
Thanks to Lugignaf for the sig!
Vapora Dark wrote:
Anyone have any more issues with the site I can pester Psi about? I've run out of things to whine about. :/
Bringing back the new guides box on the main page could count
If you found me helpful give me some +Rep :)
Signature Shop!|Tryndamere Guide|Rengar Guide
Let the hunt begin! And by hunt I mean signature finding, so why don't you check out Ubnoxius's signature shop for some signatures?
Let the hunt begin! And by hunt I mean signature finding, so why don't you check out Ubnoxius's signature shop for some signatures?
You need to log in before commenting.
What's Changed?
Guide scores are based on the votes cast on a guide within the last 30 days. Older votes are considered to be "decayed" which means they no longer contribute to a guide's score. Before these changes, guide scores were calculated using a minimum of 15 votes (unless the guide had fewer votes than that of course) with an uncapped maximum, so popular guides that received hundreds of votes in the span of a month would be able to use all of those votes for their score.
We've made two changes to this system:
2. A new maximum threshold of 60 votes has been implemented (in a 30 day period).
Under the new scoring system, a guide will always use at least 30 votes to calculate its score before implementing any voting decay. If a guide gets more than 60 votes in a 30-day period, only their most recent 60 votes will count towards their guide's score.
Why Make These Changes?
Vote decay is an important factor in the scoring system because it allows guide scores to change over time. Without vote decay, we'd have guides with hundreds or thousands of votes whose scores would be basically set in stone. Since League of Legends changes over time and so do the guides, it's important that these scores are still malleable and can be meaningfully affected by new votes.
The Minimum
The minimum threshold controls how stable a guide's score is if it isn't getting much vote traffic. Lower minimums allow the guide score to change drastically with new votes, while higher minimums decrease the effect of newer votes. With a low minimum of 15, guides with low monthly voting traffic would usually have scores under 90% due to the low vote count and the relative power of downvotes in such a small number of total votes.
Raising the minimum to 30 should give lower traffic guides a little more stability in their scores so they can climb over time and compete with the top guides. Downvotes will still affect these guides, but not as heavily as before. Likewise, having very little voting activity will affect these guides less negatively. With a higher base voting pool to work with, scores should generally be higher overall, unless the guide receives a number of downvotes.
The Maximum
The new maximum threshold is intended to curb the stability of extremely popular guides. Many users will simply upvote whatever the top champion guide is, and those users can sometimes drown out voters who vote according to the guide's quality. It was difficult to budge a #1 guide's score without a lot of downvotes, making them difficult to compete with, even if they were outdated or a better guide could take the #1 spot.
With a maximum of 60 votes used to calculate the scores of popular guides, they still enjoy a more stable guide score with the higher voting pool, but that advantage has a limit. The scoring system is more certain of a guide's quality when it gets a higher amount of votes, so we wanted that higher certainty to be reflected in the scoring logic. However, curbing this stability allows other guides to be more competitive with the top guide.
Why Not Make The Minimum and Maximum the Same?
This option isn't off the table, but it would be a pretty large change to the voting system and would affect just about every guide. The obvious advantage to this idea is it would seem to make things more fair for less popular guides since everyone would be using the same scoring rules. However, there are some downsides to completely "evening the playing field" like this.
There's an inherent inequality with a Bayesian scoring system because the system is less certain of a guide's score when it has a lower vote count and more certain when it has a higher voting pool to work with. In order to even reach a 100% score, a guide needs to get several votes, and to keep a high score they'll need a really good ratio of upvotes to downvotes. These requirements relax the more votes a guide gets because the system is more sure of the guide's quality.
Having different numbers for minimum and maximum monthly votes mimics this process of growing certainty and rewards guide authors who receive a lot of positive votes from the community. If the maximum threshold is lowered too much, top guides become very unstable and the guide ranking would be constantly shifting. Likewise, if the minimum threshold is raised too high, votes cast months ago on low-traffic guides would be impacting the guide's score, even though the guide may have drastically changed or become entirely outdated in that span of time.
Feedback Appreciated!
We'll be watching guide scores to see what kind of impact these changes have and we can adjust the numbers if either the minimum or maximum threshold isn't having the desired effect. As always, your feedback is also appreciated since we want to create a system that everyone feels is fair and works well to sort guides in a way that makes authors feel rewarded and helps readers can find the best of the best.
Please leave us a comment with your thoughts and feel free to ask if you have any questions about these changes.