LR Thread Recommended Game Polls Discussion Thread - Page 3

digmouse

China6331 Posts

July 30 2013 05:52 GMT

#41

I think the current Y/N/I system is just fine, which gives people enough information about a game's quality without being a mess, also it makes compiling good game list like what juicyjames do much much easier, have you considered if you move on to a system like stars, will there be potential problem for him doing it in TWinSC2?

lichter

1001 YEARS KESPAJAIL22272 Posts

July 30 2013 07:53 GMT

#42

On July 30 2013 14:52 digmouse wrote:
I think the current Y/N/I system is just fine, which gives people enough information about a game's quality without being a mess, also it makes compiling good game list like what juicyjames do much much easier, have you considered if you move on to a system like stars, will there be potential problem for him doing it in TWinSC2?

juicyjames has commented on it, and basically the conclusion is similar to what I posted in the OP:
Y/N/I is fine, but it doesn't separate good from great games. It is also difficult to get an accurate score.
5-star should be ideal but people don't think about their votes. It's also problematic when vote counts are low.

So basically he thinks the most important thing is to get more people voting on the polls, which I think everyone agrees with. If you have suggestions to make them more popular to use please tell us.

Ammanas

Slovakia2166 Posts

July 30 2013 13:09 GMT

#43

Just want to say, that today's OSL Ro4 is nice example, why Y/N/Time polls are not working as intended (imo). Every single game of those series was god-awful and every single one is recommended. Now, it was not awful because it was cheesy, it was awful because that was unscouted cheese with close to no defense, basically no fight back from the loser. The winner also didn't have to show excellent micro or anything. Yet people recommended those games, because it was an underdog (and kinda fan-favorite also) winning against huge favorite. So yeah, that's why I think the current system doesn't work properly - it doesn't provide any useful info about the quality of games to the people it is aimed at - the people who want to watch VoDs.

Melliflue

United Kingdom1389 Posts

July 30 2013 14:19 GMT

#44

On July 30 2013 22:09 Ammanas wrote:
Just want to say, that today's OSL Ro4 is nice example, why Y/N/Time polls are not working as intended (imo). Every single game of those series was god-awful and every single one is recommended. Now, it was not awful because it was cheesy, it was awful because that was unscouted cheese with close to no defense, basically no fight back from the loser. The winner also didn't have to show excellent micro or anything. Yet people recommended those games, because it was an underdog (and kinda fan-favorite also) winning against huge favorite. So yeah, that's why I think the current system doesn't work properly - it doesn't provide any useful info about the quality of games to the people it is aimed at - the people who want to watch VoDs.

+ Show Spoiler [Possible OSL Ro4 Spoilers] +

I'm not sure about that. The one thing I noticed reading the polls was how ineffective the anti-spoiler polls are. The polls for those games had maybe 1/5 the number of votes of the others. This is a very good hint that those games weren't played. But this may be a different problem entirely.

On July 29 2013 23:22 lichter wrote:

Show nested quote +

The first game of MKP vs Bbyong was on Whirlwind, and that was the terrible game where + Show Spoiler +

MKP proxies everything at the Xel'naga, Bbyong scouts it immediately, yet somehow loses

. It was a terrible though funny game. The game I am talking about was the one on Anaconda which is slightly better, but incredibly tense and + Show Spoiler +

came down to a base trade, single digit supplies, and tiny margins of error which made it exciting

Oh yeah, you're right. I just remembered the Gumiho vs soO game being bad and remembered MKP vs Bbyong started badly too. I forgot about that game 1

opterown

Australia54784 Posts

July 30 2013 20:31 GMT

#45

if you read the thread, many many people found the games entertaining

AgentW

United States7725 Posts

July 30 2013 20:39 GMT

#46

I concur, I found the games hugely entertaining (watched VoDs afterwards, didn't see the polls). Micro was pretty good, especially the hellions against hellions in game four.

Personally, I like the Y/N/M polls because I'm familiar with them. However, I found the format lichter (?) was using last night for WCS AM Code A Ro40 really good with the stars and descriptions. The descriptions took a minute to get used to, but considering the games are over, there's certainly time to take a read. I feel the format was better than the current system but might take a bit to become fully familiar.

opterown

Australia54784 Posts

July 30 2013 20:42 GMT

#47

the problem with the polls was that they took so damn long to read that i was almost cbb by the end of it ;p

AgentW

United States7725 Posts

July 30 2013 20:45 GMT

#48

That's basically what I just said. There would be some time needed for everyone to get used to them (the chances of this happening I think would be pretty minimal, people would just see the stars and vote). However, since the game just ended, if you can catch people and get them to vote before they go grab a snack or use the bathroom, it might work.

lichter

1001 YEARS KESPAJAIL22272 Posts

July 31 2013 04:09 GMT

#49

I really liked the outcome of those polls as well. Not perfect, but pretty good. I'm going to keep using them for the OPs I handle and hopefully others will give it a try too, so that we can see if vote distributions look more 'intelligent' because the descriptions make people think more about their votes.

lichter

1001 YEARS KESPAJAIL22272 Posts

August 01 2013 08:27 GMT

#50

Come on bros discuss

juicyjames

United States3815 Posts

August 01 2013 08:30 GMT

#51

Tried out your new system of polls for:

[WCS AM] RO40 Day 3 Challenger Season 2 2013
http://www.teamliquid.net/forum/viewmessage.php?topic_id=423447

I'm not sure if we can take much from it, though.

lichter

1001 YEARS KESPAJAIL22272 Posts

August 01 2013 08:32 GMT

#52

On August 01 2013 17:30 juicyjames wrote:
Tried out your new system of polls for:

[WCS AM] RO40 Day 3 Challenger Season 2 2013
http://www.teamliquid.net/forum/viewmessage.php?topic_id=423447

I'm not sure if we can take much from it, though.

Wow that's a bad day

opterown

Australia54784 Posts

August 01 2013 09:54 GMT

#53

ouch those polls

graNite

Germany4434 Posts

August 01 2013 10:54 GMT

#54

I dont get why I can see the results of a poll before voting, this definitly hurts the outcome.

Melliflue

United Kingdom1389 Posts

August 01 2013 11:27 GMT

#55

On August 01 2013 19:54 graNite wrote:
I dont get why I can see the results of a poll before voting, this definitly hurts the outcome.

But it has to be possible to see the results of the poll before voting, otherwise people who check the poll afterwards to see if the game was recommended or not would have to vote before even knowing whether or not they should watch it.

graNite

Germany4434 Posts

August 01 2013 11:49 GMT

#56

Yeah of course, just hide the results, show the vote buttons and an additional one "show results" how simple is that. or even better: make two different ways to interprete the pollid in the forum, eg.

[pollvote]234897[/pollvote]

only shows poll 234897 and its answers while

[pollresults]234897[/pollresults]

shows the results.

lichter

1001 YEARS KESPAJAIL22272 Posts

August 02 2013 04:13 GMT

#57

On August 01 2013 20:49 graNite wrote:
Yeah of course, just hide the results, show the vote buttons and an additional one "show results" how simple is that. or even better: make two different ways to interprete the pollid in the forum, eg.

Show nested quote +

only shows poll 234897 and its answers while

Show nested quote +

shows the results.

There are two degrees or ways to 'obscure' results.

One is to make the results completely hidden, as you suggested, requiring an additional click to see them.

Another is to show the averaged or compiled result while only hiding the distribution.

The first method ensures that there is little or no influence made by previous votes on current voters. This type of influence can be both positive and negative. On the one hand it can make people think about their votes more, instead of voting on impulse (which is likely highly biased). On the other, it could just make people vote for or against the majority (some people are just assholes, some people just want to agree). Hiding the results removes both influences. I am uncertain if the net result is a more accurate or more inaccurate poll. I suppose it is up to preferences.

The second method does influence voters too, but in a more subtle way. One consequence of only hiding distributions is that voters will tend to try to 'force' the compiled result to match their opinion. If they believe the average should be higher, they vote for a score higher than they otherwise would. If they believe the score should be lower, they accordingly vote lower than they would have. If the score is about right, then they vote as they would. This attempt to 'balance' the average has pros and cons. One positive is that it can negate some of the inaccuracies of low vote counts, assuming those that do vote do so intelligently. The 10 voters will end up trying to balance each other out to reach an agreeable mean. It doesn't solve the problem, but it is a slight improvement. However it can also give the poll results greater variance if there are voters who have extreme opinions, since they will try to cancel each other out. Low vote counts is a problem, is what I'm saying.

Anyway I do agree that some kind of result obfuscation is necessary, but which do you prefer: all the results or only its distribution?

graNite

Germany4434 Posts

August 02 2013 07:37 GMT

#58

I think the "hidden results"-version just gives the most honest vote, thats why I think you should use it.

tl2212

Belize731 Posts

August 02 2013 08:14 GMT

#59

I think the criteria poll is cool but you only should include 5 options, Level of play, Entertainment Value, Build Order, Micro, and No / not really. This would actually make the results more like a 5 star poll in distribution, but i think it would still convey a lot of detail about the game and give a good sense of whether the game is recommended or not.

Poll: What was awesome about Bisu vs Flash game 1?

High level of play (1)

33%

High entertainment value (1)

33%

None, not recommended (1)

33%

Awesome build orders (0)

0%

Sick sick micro (0)

0%

3 total votes

Your vote: What was awesome about Bisu vs Flash game 1?

(Vote): High level of play
(Vote): Awesome build orders
(Vote): Sick sick micro
(Vote): High entertainment value
(Vote): None, not recommended

like this. its simple and effective and if the game isn't recommended, people can just vote no. the way you have it now with 17 options is too overthought imo

also its funny how you talk about the low voter turnout, and then most of the polls in the OP have like 5 or 10 votes.

However this OP is great and very well done! i would rate it 5 stars if i could

its awesome how much you guys care about the community here

lichter

1001 YEARS KESPAJAIL22272 Posts

August 02 2013 08:20 GMT

#60

The problem with that kind of poll is that some people may want to vote for multiple options. Some games have them all, and the poll should reflect that. If only one criteria can be voted on, you will have situations where the votes are split between Entertainment and Level of Play, for example. 50 vote one, 50 vote for the other. The way the polls are structured, it will look "less good" than a game where 70 people vote for level of play.

To me, the only way a criteria based poll can work is if we allow checkboxes, as I included as an example. This makes the poll soooooo much simpler with the option to vote for multiple criteria. It's also only 1 or 2 clicks more, and doesn't require a lot of time to go through to understand the results.

I will bug waxy to bug R1CH to find a way to checkboxes work.

Prev 1 2 3 4 Next All

Please or register to reply.

LR Thread Recommended Game Polls Discussion Thread - Page 3

Completed

Ongoing

Upcoming