Recently Hearthpwn added a little widget or something that shows the win rate of the decks on the front page. IMO I feel like this is a bad move, and here's why:
1. Doesn't show the big picture: I saw pretty much the exact same miracle rogue list that people have been playing since JTU came out showing up on the front page again. Aside from my annoyance at someone reposting the same deck again, I noticed that it had a win rate of 41.2%. Here's the thing though. The deck is more difficult to pilot and as such, people just hopping on and playing it for the first time will be lowering the win rate of it. This in turn encourages people to downvote the deck or completely ignore it even though it's real win rate is likely higher for people who know the deck.
2. Encourages hive mind mentality: We already have TempoStorm and other websites ranking decks. I like Hearthpwn because the weird and janky decks sometimes get promoted even if they're not huge win rate monsters. Now, everyone on this site will likely be latching on to the top win rate decks and making decisions solely based on that. This means less chance of deck innovators or people with complex decks to get a chance to show off what they have (which I think is what makes this my favourite deck sharing website). It adds yet another barrier in getting your deck recognized. Someone could look at your deck and say "35% win rate, no thanks" and that could be it. People don't have to take any time to try out new things because they'll only swarm to that which is already successful.
3. Intentional win rate tanking: Let's say I'm really annoyed at a deck on the front page. Maybe the author is arrogant or it's getting upvotes without providing actual content in the guide. It's easy for me to just import the deck into the game and sit at rank 10, auto-conceding and lowering the win rate. You say that you don't think people would do this? Well I once made an anti-pirate warrior deck and conceded 40 times to get a few pirate warrior matches. Throw a show on in the background and I have no problem doing this when I'm tilted about something. If a deck isn't as popular yet, it's very easy to affect the %. This could even happen unintentionally where people who are terrible at the game play your deck and bring the win% down.
I'll bet most people are happy with the addition (especially since it's new and a cool feature) but I'm having a hard time with it. I'm not trying to bash the site programmers or anyone else who worked hard on it because I know it takes a lot of work to add new features like this. I just think it causes more harm than good. Anyone else agree/disagree?
1, decks can be difficult at first, players who play this deck without practicing will lower the win rate of the deck.
2, it reduces the people trying out the ''weird and janky decks'' and instead focus on the winrate, they'll use the decks with good winrate b'cuz: "they'll only swarm to that which is already successful".
3, you can intentionaly lower the winrate by auto -conceding.
I'm willing to bet more people will always more information about decks when its possible. You're seeing deck stats starting to pop up more and more in the community and it's one of those natural progressions we're all going to need to figure out.
Let's talk about the "hive mind mentality". This already happens on HearthPwn and I'm willing to bet we have had a fairly decent effect on the in-game meta over the past few years thanks to being the most popular Hearthstone & netdecking website. Adding additional information to decks that include their user winrates may indeed push people towards other decks and to further change what people are playing in-game, but us allowing authors to include their WR in the title of the deck already has done this as well.
My advice: Accept what the hive mind is playing and do your best to counter what the majority of people are queuing up with - it may be even easier now!
On winrates not being "accurate" due to difficulty of decks. This one is a complex issue due to a variety of factors.
Number of users playing the deck with Innkeeper tracking the games; Lower users with the deck means larger possibility for large swings in numbers.
Luck of a deck author advertising a high winrate; It may not be as good as it seems.
Timing; Over time the winrate of a deck can fall.
One interesting thing to me about the low winrate hard decks is that the winrate over time will likely improve due to people not wanting to play it because they couldn't even be bothered to check the deck out and more importantly, people who are still playing it have gotten better. Imagine a world where a gem of a deck can exist that was once on the front page of a huge site but isn't even tried to be played on the ladder because of the seemingly lower winrate; It's very cool!
One thing that I think could help is if we were able to figure out a clean way to display data on winrates with more ways of filtering it. An example could be the legend rank winrate, or the winrate of the deck with only users who have played it a certain number of times. Of course, I'm not making any promises, but there are definitely other ways we can help tell the story. I don't think we'd do anything different on the front page widget though because we're already crazy strapped for space! Feedback is important here though.
I don't think it should really matter if people are going to downvote a "low winrate" deck, I already see this happen when a deck is advertised as an 80% winrate with a low author sample size and no one is able to replicate it. You also see a lot of people that are vocal about the "bullshit" winrate.
Winrate abuse is something we're going to be monitoring and adjusting our algorithms on over time where needed; It's a large concern of ours. Users intentionally trying to mess with our stats could be quietly banned from the system. I don't think it will be as big of an issue though a some may think as we've got tons of people running Innkeeper, a number that will continue to grow, and so it will only get harder to skew the numbers.
I appreciate your (and others) time to give us feedback on the new feature. It's something we're very much still working on and will be pushing more information that we gather from the game onto pages on the site in the future. Ultimately, the goal is to give people as many tools as possible so they can make the best decisions. I'm personally very excited about the things we're going to be able to do thanks to our Innkeeper application and hope that more people will install it so that we can continue delivering stat updates to the site.
Also, apologies for a bit of a messy post. I could have done a better job. If anything isn't clear, let me know and I can try and go over it again.
One thing that I think could help is if we were able to figure out a clean way to display data on winrates with more ways of filtering it. An example could be the legend rank winrate, or the winrate of the deck with only users who have played it a certain number of times. Of course, I'm not making any promises, but there are definitely other ways we can help tell the story. I don't think we'd do anything different on the front page widget though because we're already crazy strapped for space! Feedback is important here though.
This would be amazing.
Edit: To note as well, not all players may be using the deck tracker and more different decks with winrates can be found on other sites like metastats.net which brought to my attention that the deck must undergo at least 80-120 or more games like a cap threshold by the user(s) in the standard or wild format, an also show which region to it happened (got to legend) EU,NA Asia.
Although we just need to have more people sign up to InnKeeper to keep the database showing /consistent results and winrates if they play on PC. Hopefully Blizzard decides to also share the deck winrates from phones or implement it in some way but it can be a distraction as more fun decks may not be popular or brought to light to other players.
Thanks Flux for a solid response. Some of the fixes you've been hinting at (not promising of course!) would definitely solve a few of the problems I've listed on there. I understand it's tough to make a system that works best for everyone/everything so thanks for your hard work on it! I wish there was some sort of incentive for playing unique and completely different decks but that would have to come from within Hearthstone, not from Hearthpwn.
I do hope to see the win rate /average turn counters implemented on other facets of the site. And I encourage anyone using innkeeper to please track their games.
I don't get much in the way of visits for my decks because I'm an average player / poster, and posting some arbitrary win % would only hurt me in the process of becoming more solid in the gaming community.
If the Average player or even the new player posts a deck here and it is tracked at having a rather decent win rate / average turn count then those decks may finally see the light of day and make it to a more prominent spot in the community.
The fear factor of the OP and others is relevant, though as Flux pointed out, change and progress happen and sometimes you just gotta roll with it and hope for the best. Or be pleasantly surprised.
One thing that I think could help is if we were able to figure out a clean way to display data on winrates with more ways of filtering it. An example could be the legend rank winrate, or the winrate of the deck with only users who have played it a certain number of times. Of course, I'm not making any promises, but there are definitely other ways we can help tell the story. I don't think we'd do anything different on the front page widget though because we're already crazy strapped for space! Feedback is important here though.
Reply is a very good one overall, but I find this snippet most interesting. Since on the front page widget you're strapped for space, what would you think of incorporating the filter concepts into the algorithm itself, rather than in the UI?
Meaning: instead of calculating the raw win rate over all games, only the games from users who have played that deck a certain number of times and/or above a certain threshold rank are taken into account. This should filter out both occasional players (you can't evaluate a deck properly if you only play with it 3 games), and newbies who might lose mostly because of their inexperience, rather than the deck being bad.
Minimum games # could be, say, anything from 10 to 100. The higher the threshold, the more accurate the results, but a high value would also introduce some delay in the updates. IMHO a reasonable compromise might be 30 or 40, but I'll leave that to someone with more experience ;)
Minimum rank also should avoid the extremes. 20 would probably be too mild a filter, and 5 too elitist. Maybe 15, or 10?
I'd really like to see this feature reported on any deck's page in Hearthpwn. Seems like there are ways to minimize users intentionally screwing with the results. For example,
A match's result is only considered in computing a deck's performance if is satisfies the following requirements:
The match was in Ranked mode.
The user was rank 15 or better.
The user did not concede/lose before turn 3 [yes, I realize some extreme matches might get omitted in this scenario but I think this is a good idea to make it more difficult/longer for someone to try to fudge results].
I'd really like to see this feature reported on any deck's page in Hearthpwn. Seems like there are ways to minimize users intentionally screwing with the results. For example,
A match's result is only considered in computing a deck's performance if is satisfies the following requirements:
The match was in Ranked mode.
The user was rank 15 or better.
The user did not concede/lose before turn 3 [yes, I realize some extreme matches might get omitted in this scenario but I think this is a good idea to make it more difficult/longer for someone to try to fudge results].
We already only use ranked matches for the stat.
What makes a rank 15 player good enough over a rank 16 player that makes their numbers more worthy? I think we can better address this if we can find a good way to show a breakdown by rank on the decks themselves while retaining the overall on the widget.
With that said, no promises yet! We're still talking about the system and everyone's feedback has been helpful!
I'd really like to see this feature reported on any deck's page in Hearthpwn. Seems like there are ways to minimize users intentionally screwing with the results. For example,
A match's result is only considered in computing a deck's performance if is satisfies the following requirements:
The match was in Ranked mode.
The user was rank 15 or better.
The user did not concede/lose before turn 3 [yes, I realize some extreme matches might get omitted in this scenario but I think this is a good idea to make it more difficult/longer for someone to try to fudge results].
We already only use ranked matches for the stat.
What makes a rank 15 player good enough over a rank 16 player that makes their numbers more worthy? I think we can better address this if we can find a good way to show a breakdown by rank on the decks themselves while retaining the overall on the widget.
With that said, no promises yet! We're still talking about the system and everyone's feedback has been helpful!
Re: the rank point I'm thinking at rank 15 you don't lose stars (though that'd mean you'd want to limit it to rank 14 though the same effect happens at rank 10 and 5 so maybe just ditch the rank limitation and go with only the ranked games limitation.
Also, perhaps you could have some data log check and if someone racks up alot of losses in a row and in quick succession then perhaps that data is omitted and the perpetrator excluded from future stats. :)
Personally, I would like to see this on all decks, also. Or at least an option that a player posting a deck can toggle on/off, or between on to the deck builder only/on to everyone/off for everyone (maybe the toggle cant be turned off for main page decks, though, if the site prefers that).
In general, I trust what people who post their own stats with proof show, as being what their win rate was. Community win rates, I would expect to be much lower, as given a larger sample of players, it would be natural to regress toward the norm/center. So, I don't think the difference between self reported (w/proof) and community rates is really much of an issue.
The toggle idea seems good though, as some builders might not want to deal with the opinions of those who don't get that, or might prefer people judge without the 'hive mind' effect, whether their list sounds like fun or not.
I can see where I might want to see how the community is doing with my deck, but not have the whole community see it... especially fun decks, which most of mine are intended to be. Thus the three-option toggle suggestion. I like the idea of respecting the builder's preference (off main page) so that if a deck owner chose to have the community #s available, I could see them everywhere, but if they didn't, to allow the builder to see them personally for their own decks, at least (could help with testing tweaks, a lot, before claiming a change is 'worthy').
Rollback Post to RevisionRollBack
"Its beyond my control." - Viscount Valmont, as played by John Malkovich, Dangerous Liaisons
On one hand, the feature is good to unmask advertising titles with humongous winrates, which are calculated on winstreaks at best (when not false completely). Eg., a pool of 20-ish games is still kinda insignificant for statistical purposes.
On the other hand, the new feature surely lowers the winrate of many decks, because the deck is being piloted by many who, even if expert, don't completely know what they are doing, which is normal at first, when you use something made by someone else.
Now, i may miss something, but maybe the winrate should appear only when calculated on large enough numbers, and at ranks above 15 or even 10. Or you compute stats when the deck is being piloted by the same person from a certain rank to another (to avoid micrometa stats).
The point is that the winrate showed must be significant. It cannot include sparse and situational numbers of people just trying the deck for a couple of matches - before they even interiorise how to properly play it.
Tldr; the new feature is good if wisely applied to statistically significant numbers. NOT to any large numbers. Otherwise no number shown is more fair.
Agreed. It's going to expose a lot of inconsistent "legendary" decks which are only good for climbing from rank 1 and 3 stars right into legendary, which is what a lot of decks on the front page really are. A lot of people are naming their decks 70%+ winrate, when their stats clearly show that they squeaked by to legendary after rarely getting paired up against the top tier decks or classes. It will also show how a deck performs when the general player base uses it. Just take a look at Hearthstone replay site. Their top tier decks are nowhere near 70% winrate, because they have a much bigger sample size from different players.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
Recently Hearthpwn added a little widget or something that shows the win rate of the decks on the front page. IMO I feel like this is a bad move, and here's why:
1. Doesn't show the big picture: I saw pretty much the exact same miracle rogue list that people have been playing since JTU came out showing up on the front page again. Aside from my annoyance at someone reposting the same deck again, I noticed that it had a win rate of 41.2%. Here's the thing though. The deck is more difficult to pilot and as such, people just hopping on and playing it for the first time will be lowering the win rate of it. This in turn encourages people to downvote the deck or completely ignore it even though it's real win rate is likely higher for people who know the deck.
2. Encourages hive mind mentality: We already have TempoStorm and other websites ranking decks. I like Hearthpwn because the weird and janky decks sometimes get promoted even if they're not huge win rate monsters. Now, everyone on this site will likely be latching on to the top win rate decks and making decisions solely based on that. This means less chance of deck innovators or people with complex decks to get a chance to show off what they have (which I think is what makes this my favourite deck sharing website). It adds yet another barrier in getting your deck recognized. Someone could look at your deck and say "35% win rate, no thanks" and that could be it. People don't have to take any time to try out new things because they'll only swarm to that which is already successful.
3. Intentional win rate tanking: Let's say I'm really annoyed at a deck on the front page. Maybe the author is arrogant or it's getting upvotes without providing actual content in the guide. It's easy for me to just import the deck into the game and sit at rank 10, auto-conceding and lowering the win rate. You say that you don't think people would do this? Well I once made an anti-pirate warrior deck and conceded 40 times to get a few pirate warrior matches. Throw a show on in the background and I have no problem doing this when I'm tilted about something. If a deck isn't as popular yet, it's very easy to affect the %. This could even happen unintentionally where people who are terrible at the game play your deck and bring the win% down.
I'll bet most people are happy with the addition (especially since it's new and a cool feature) but I'm having a hard time with it. I'm not trying to bash the site programmers or anyone else who worked hard on it because I know it takes a lot of work to add new features like this. I just think it causes more harm than good. Anyone else agree/disagree?
Moved to Site Feedback & Support.
Intentional win rate tanking
Why would someone do such a thing, to change an unreal thing? Some pixels? If someone does that, please get a life.
Also your example was not intentional or hating someone.
Playing since 1 June 2014.
Review on Every Card: http://goo.gl/RTz806
Cards That Will Be Missed in Standart Next Year: http://goo.gl/adNMnn
tl;dr:
1, decks can be difficult at first, players who play this deck without practicing will lower the win rate of the deck.
2, it reduces the people trying out the ''weird and janky decks'' and instead focus on the winrate, they'll use the decks with good winrate b'cuz: "they'll only swarm to that which is already successful".
3, you can intentionaly lower the winrate by auto -conceding.
I agree btw
Mod Working Reaaaaally Hardly
I'm willing to bet more people will always more information about decks when its possible. You're seeing deck stats starting to pop up more and more in the community and it's one of those natural progressions we're all going to need to figure out.
Let's talk about the "hive mind mentality". This already happens on HearthPwn and I'm willing to bet we have had a fairly decent effect on the in-game meta over the past few years thanks to being the most popular Hearthstone & netdecking website. Adding additional information to decks that include their user winrates may indeed push people towards other decks and to further change what people are playing in-game, but us allowing authors to include their WR in the title of the deck already has done this as well.
My advice: Accept what the hive mind is playing and do your best to counter what the majority of people are queuing up with - it may be even easier now!
On winrates not being "accurate" due to difficulty of decks. This one is a complex issue due to a variety of factors.
One interesting thing to me about the low winrate hard decks is that the winrate over time will likely improve due to people not wanting to play it because they couldn't even be bothered to check the deck out and more importantly, people who are still playing it have gotten better. Imagine a world where a gem of a deck can exist that was once on the front page of a huge site but isn't even tried to be played on the ladder because of the seemingly lower winrate; It's very cool!
One thing that I think could help is if we were able to figure out a clean way to display data on winrates with more ways of filtering it. An example could be the legend rank winrate, or the winrate of the deck with only users who have played it a certain number of times. Of course, I'm not making any promises, but there are definitely other ways we can help tell the story. I don't think we'd do anything different on the front page widget though because we're already crazy strapped for space! Feedback is important here though.
I don't think it should really matter if people are going to downvote a "low winrate" deck, I already see this happen when a deck is advertised as an 80% winrate with a low author sample size and no one is able to replicate it. You also see a lot of people that are vocal about the "bullshit" winrate.
Winrate abuse is something we're going to be monitoring and adjusting our algorithms on over time where needed; It's a large concern of ours. Users intentionally trying to mess with our stats could be quietly banned from the system. I don't think it will be as big of an issue though a some may think as we've got tons of people running Innkeeper, a number that will continue to grow, and so it will only get harder to skew the numbers.
I appreciate your (and others) time to give us feedback on the new feature. It's something we're very much still working on and will be pushing more information that we gather from the game onto pages on the site in the future. Ultimately, the goal is to give people as many tools as possible so they can make the best decisions. I'm personally very excited about the things we're going to be able to do thanks to our Innkeeper application and hope that more people will install it so that we can continue delivering stat updates to the site.
Also, apologies for a bit of a messy post. I could have done a better job. If anything isn't clear, let me know and I can try and go over it again.
This would be amazing.
Edit: To note as well, not all players may be using the deck tracker and more different decks with winrates can be found on other sites like metastats.net which brought to my attention that the deck must undergo at least 80-120 or more games like a cap threshold by the user(s) in the standard or wild format, an also show which region to it happened (got to legend) EU,NA Asia.
Although we just need to have more people sign up to InnKeeper to keep the database showing /consistent results and winrates if they play on PC. Hopefully Blizzard decides to also share the deck winrates from phones or implement it in some way but it can be a distraction as more fun decks may not be popular or brought to light to other players.
#3182
Thanks Flux for a solid response. Some of the fixes you've been hinting at (not promising of course!) would definitely solve a few of the problems I've listed on there. I understand it's tough to make a system that works best for everyone/everything so thanks for your hard work on it! I wish there was some sort of incentive for playing unique and completely different decks but that would have to come from within Hearthstone, not from Hearthpwn.
I do hope to see the win rate /average turn counters implemented on other facets of the site. And I encourage anyone using innkeeper to please track their games.
I don't get much in the way of visits for my decks because I'm an average player / poster, and posting some arbitrary win % would only hurt me in the process of becoming more solid in the gaming community.
If the Average player or even the new player posts a deck here and it is tracked at having a rather decent win rate / average turn count then those decks may finally see the light of day and make it to a more prominent spot in the community.
The fear factor of the OP and others is relevant, though as Flux pointed out, change and progress happen and sometimes you just gotta roll with it and hope for the best. Or be pleasantly surprised.
Cute, ineffective, but cute.
I'd really like to see this feature reported on any deck's page in Hearthpwn. Seems like there are ways to minimize users intentionally screwing with the results. For example,
A match's result is only considered in computing a deck's performance if is satisfies the following requirements:
The match was in Ranked mode.
The user was rank 15 or better.
The user did not concede/lose before turn 3 [yes, I realize some extreme matches might get omitted in this scenario but I think this is a good idea to make it more difficult/longer for someone to try to fudge results].
Also, perhaps you could have some data log check and if someone racks up alot of losses in a row and in quick succession then perhaps that data is omitted and the perpetrator excluded from future stats. :)
Personally, I would like to see this on all decks, also. Or at least an option that a player posting a deck can toggle on/off, or between on to the deck builder only/on to everyone/off for everyone (maybe the toggle cant be turned off for main page decks, though, if the site prefers that).
In general, I trust what people who post their own stats with proof show, as being what their win rate was. Community win rates, I would expect to be much lower, as given a larger sample of players, it would be natural to regress toward the norm/center. So, I don't think the difference between self reported (w/proof) and community rates is really much of an issue.
The toggle idea seems good though, as some builders might not want to deal with the opinions of those who don't get that, or might prefer people judge without the 'hive mind' effect, whether their list sounds like fun or not.
I can see where I might want to see how the community is doing with my deck, but not have the whole community see it... especially fun decks, which most of mine are intended to be. Thus the three-option toggle suggestion. I like the idea of respecting the builder's preference (off main page) so that if a deck owner chose to have the community #s available, I could see them everywhere, but if they didn't, to allow the builder to see them personally for their own decks, at least (could help with testing tweaks, a lot, before claiming a change is 'worthy').
"Its beyond my control." - Viscount Valmont, as played by John Malkovich, Dangerous Liaisons
On one hand, the feature is good to unmask advertising titles with humongous winrates, which are calculated on winstreaks at best (when not false completely). Eg., a pool of 20-ish games is still kinda insignificant for statistical purposes.
On the other hand, the new feature surely lowers the winrate of many decks, because the deck is being piloted by many who, even if expert, don't completely know what they are doing, which is normal at first, when you use something made by someone else.
Now, i may miss something, but maybe the winrate should appear only when calculated on large enough numbers, and at ranks above 15 or even 10. Or you compute stats when the deck is being piloted by the same person from a certain rank to another (to avoid micrometa stats).
The point is that the winrate showed must be significant. It cannot include sparse and situational numbers of people just trying the deck for a couple of matches - before they even interiorise how to properly play it.
Tldr; the new feature is good if wisely applied to statistically significant numbers. NOT to any large numbers. Otherwise no number shown is more fair.
Agreed. It's going to expose a lot of inconsistent "legendary" decks which are only good for climbing from rank 1 and 3 stars right into legendary, which is what a lot of decks on the front page really are. A lot of people are naming their decks 70%+ winrate, when their stats clearly show that they squeaked by to legendary after rarely getting paired up against the top tier decks or classes. It will also show how a deck performs when the general player base uses it. Just take a look at Hearthstone replay site. Their top tier decks are nowhere near 70% winrate, because they have a much bigger sample size from different players.