What MOBAFire's voting system needs - Vote Decay!

[quote=Lugignaf]Damn it math majors... @_@ Formulas way too complex...[/quote]

[quote=JunSupport]Mmm... *sleepy yaaaawn* You bring up a good point, jhoijhoi. Exactly why I said this is open to critique. I have no intention of tackling that sort of math... (although I'm tempted) I wanted to be nice and give outdated votes a saving grace of a decay over time, but it is apparent that the math gods are not as benevolent. I will revert back to [color=#ff0000]my original cut-and-dry mercy-lacking prototype[/color] that I chose to heavily rework before posting the OP. Simplify the system and make it a measure used for more extreme cases. Change the decay to render a vote dormant (so it goes from 100% to zero value), and expand the requirement for a vote to be eligible to be decayed to a far more severe time such as 6 monthes (effectively 12 patches)? Also, change the requirements to be eligible for decay. Instead of going by voter count. (though we'll keep the minimum vote requirement and raise it up to 150) Let's go with rating! Guides above 85% are all subjected to decay (unless they have below 151 votes.) ( And, no, I don't find this too drastic a number to start with, since votes being decayed are 6 monthes old, and 80% used to be the norm for Mobafire. I'd like to see all guides above 90% take a bit of a hit. And that's not just because my guides are all below that number. *wink wink*) And to make it actually have some impact, heavily nerf the possibility of refreshing the decay timer by doubling the requires upvotes (while also expanding the decay timer). A guide now requires 6 upvotes to refresh the timer. The timer will be expanded to 2 weeks instead of 1 week. This both makes it a little easier and harder. Easier, because succeeding means you're in the clear for the remainder of the 2 weeks since the decay timer last reset, and technically the following 2 weeks until the deadline of the next decay timer. But, harder because 6 votes is still 6 votes, and it's all a matter of guide-quality and luck on viewers. [b]This however, would REQUIRE a user-notification system that alerted you that your vote has decayed (become outdated)[/b] So that users that still regularly browse the site can willing renew the guide if they so choose. This system is also mandatory, because most likely, some guides will immediately plumment from decay as 6 votes is pretty hard to get when you've already accumulated the usually 1000+ users who actual vote at all. Oh, and users refreshing their vote will count towards the required 6 upvotes per month. If done properly, this allows a guide is repeated cycle it's voters to show it is truly TRULY deserving of its place in mobafire. [color=#ffff00]^ I like this part most because it would make MOBAFire have a truly interactive and LIVING rating system that depends on its users as its life blood. What better way to attractive users then to be able to tell them that this site has a rating system that always checks ITSELF twice per year?[/color] Ignoring the notification-thingie for now, having votes simply be considered outdated and therefore not counted would return the math down to the realm of simpletons I think. All we have to do is recalculate the rating with the decayed votes excluded. In short: New more extreme-case-dealing Decay: -Votes older then 6 monthes lose 100% value. -Owners of the respective votes are immediately notified. -Decay timer: Votes will decay at the end of 1 patches (2 weeks total) -A guide requires 6 upvotes to avoid decay. -A guide is only eligible for decay if it hits over 85% and has 151+ votes total. -New: Users renewing their decayed votes contribute to required 6 upvotes 2 patches. (So, 6 votes isn't impossible. Just hard, especially if your guide is full of sheep votes.) The intention of this system is to constantly apply decay to highly rated guides, and only giving guides a small window of opportunity to DETER (not prevent) decay through the loyalty of users. Not all voters are active members, so there will always be decay with this system, and naturally forcing guides to show their best every patch cycle if they want to stay on top. And I shall dub thee.... [h2][b][color=#ff0000]Vote Decay - Guides' Nightmare: Beta 0.1[/color][/b] !![/h2] If you guys think this would be alright, feel free to express critiques, and I'll restructure the OP when I wake up after going to bed to have a well deserves snooze following my trip to the dentist to have my wires fitted and drink some delicious free lounge cappuccino machine caffeine. ... Oh wait. edit#XX: I wonder if anyone else edits their posts as much as I do. I must have clicked Post Reply at least 20 times now. Later. Scalding caffeine and metallic wires to my jaw await.[/quote]

[quote=Dorian182]Great idea.[/quote]

[quote=Temzilla][quote=jhoijhoi] Jun, I loved this idea when I saw it, but I think it'd be too convoluted to formulate. The scoring system already works off of the Bayesian average and I don't know how this would work with it. The math required for this idea - it boggles my mind. Bayesian Average: Calculating the Bayesian average uses the prior mean m and a constant C. C is assigned a value that is proportional to the typical data set size. The value is larger when the expected variation between data sets (within the larger population) is small. It is smaller, when the data sets are expected to vary substantially from one another. [img]http://upload.wikimedia.org/math/9/0/a/90a1df238dc2461747fb8031bea7cde0.png[/img] [/quote] To simplify this a bit, [quote]bayesian average = ( (Mean Votes*Mean Rating) + (Number of Votes * Rating) ) / (Mean Number of Votes + Number of Votes)[/quote] At the moment, the number of votes is a number, if there are 300 votes, then there are All you are doing is changing the weight of individual votes. I.E. A guide has 150 up votes, and 150 down votes. 100 upvotes are 6 months old, and so have a weight of 50, but 100 downvotes are brand new. This means the guide has a 50% rating for having 150 up votes 150 down votes, and with the new system it would have 100 up votes, and 125 down votes. You would simply calculate the total number of votes differently, the forumula doesn't actually change at all.[/quote]

[quote=Matt]This is all pretty complex. Before we discuss HOW to vote decay, we need to all agree that it's necessary in the first place. First of all I don't think it's fair to say a guide has a high score BECAUSE it has high vote count. Low scoring guides all have a low vote count obviously because people stop voting after it has already been sunk. Guides that are good and get more up votes than down votes will continue to do so, but this is how it is supposed to be. I'm still not sold on the notion of "sheep voting". However, an alternative reason for vote decay might be that really old votes were cast based on the champion at that time. Regardless of how up to date the author has kept their guide, those super old votes might not really apply several months out. Additionally, negative votes from early in the guide's life might also not apply, if the guide has been updated and improved consistently. A third issue is that changes to the site over time affect voting and scoring. A prime example is the period where C2V was on for everyone permanently. A lot of guides did get an unfair boost during that period that new guides miss out on. Are these good enough reasons to do it? What does everyone think? Any reasons I'm missing?[/quote]

[quote=jhoijhoi][quote=matt]Additionally, negative votes from early in the guide's life might also not apply, if the guide has been updated and improved consistently[/quote] I didn't even think of that. All votes mean all votes, so a decay, or window frame would remove negative AND positive votes. If we can't have a soft/hard reset due to votes not being like elo, then a way to remove irrelevant/old votes would be a step forward, I think :)[/quote]

[quote=JunSupport]Sorry for not updating the OP, everyone. Been rather busy with the new school year, and frankly, I have a hard time changing what is a large wall of text post-placement. Maybe I'll just wipe it and re-write it later... [quote=Matt] This is all pretty complex. Before we discuss HOW to vote decay, we need to all agree that it's necessary in the first place. First of all I don't think it's fair to say a guide has a high score BECAUSE it has high vote count. Low scoring guides all have a low vote count obviously because people stop voting after it has already been sunk. Guides that are good and get more up votes than down votes will continue to do so, but this is how it is supposed to be. I'm still not sold on the notion of "sheep voting". However, an alternative reason for vote decay might be that really old votes were cast based on the champion at that time. Regardless of how up to date the author has kept their guide, those super old votes might not really apply several months out. Additionally, negative votes from early in the guide's life might also not apply, if the guide has been updated and improved consistently. A third issue is that changes to the site over time affect voting and scoring. A prime example is the period where C2V was on for everyone permanently. A lot of guides did get an unfair boost during that period that new guides miss out on. Are these good enough reasons to do it? What does everyone think? Any reasons I'm missing? [/quote] -The reason I make it a requirement to have a high vote-count to be subjected to decay is because the goal here is to weaken and loosen the inflated guide ratings. Subsequently, guides of this nature always have a high vote count. I recognize this connection as a correlation, not a casual relationship. The correlation between number of votes and rating level is merely a good indicator of inflated ratings. Additionally, I feel old guides, inflated or not, should have older votes decayed to make the rankings of guides potentially more fluid to allow ease of access by other guide-writers of equal value. Elections and rankings occur on an annually basis because as time passes; circumstances change, accepted methods grow old, and past beliefs are put into question. -In regards to sheep voting, think of it this way: A lot of people probably report about "Troll Votes" on a regular basis, because they feel the vote is misinformed, mistaken, or even just plain stupid. Those three possibilities however, can apply to Sheep Votes as well. No one ever complains if an "idiot" votes positive and says "I got 23/0/5 with this guide. +1". No one ever questions such a vote, guide-writers are simply happy to get the upvote. This is an enduring bias that we need safeguards for because even the most objective of us don't want to irregularly try to undo our own success. -"those super old votes might not really apply several months out" This about sums up the major reason why votes should get decay. Out with the old, in with the new. In a voting case, this principle has no real flaws. Voting should be fluid and renewable to ensure guides aren't just riding on glory-days. -Changes to the site are most definitely also a factor in scoring. But, not just because of past-mistakes. The C2V period was definitely a counter-productive time, but that is only an intense peek into what was always happening. The main problem I feel is the site pushes forward with votes, but has no "double-check" system to go back and track the condition and progress of guides. The homepage is extremely geared toward bringing guides up to where they deserve right away, but there is currently no mechanic to check up on whether or not the guides still [i]deserve[/i] to be there once their time in the limelight has ended.[/quote]