02-Aug-2013, 1:27 PM
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">Before I contribute to this discussion on variety, I'd like to expound a little bit more on what Tom said in his original post about ratings, just so we're pretty much on the same page about what the types of levels at various ranges of averages in the results will <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-style:italic;">probably look like, at least if what's transpired so far is any indication. As someone who has been noting statistical trends in the voting across different types of levels (yes, this is why I'm encouraging people not to hold back votes they've decided on!), I can safely say that levels tend to fall into four categories with respect to how their ratings change along the voting process. This list is by descending order of % of levels of the voting pool:
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">- The "marmite" levels (~3.01 - 3.75). <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">This is the largest group of levels by far, and for good reason: they tend to inspire the most amount of disagreement among voters. These are the levels that are loved by some, hated by others, and in some cases, don't exactly inspire votes that carry much variance and affect the results. The average ends up remaining squarely in the middle in the end after bouncing around for a little while. Levels with concepts that polarize opinion between "annoying" and "genius" fall into here, and especially levels that feature some sort of game element or gameplay that veterans approach with some degree of negative bias (blobs, ice mazes, blue/invisible walls, etc.). Very easy levels also fit here too.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">- The rejects (~1.25 - 3.00). <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">These levels are the worst of the bunch. One of the consequences of a pre-voting elimination has been - at least for the most part - narrowing down the spectrum of quality that characterizes the levels being considered, which, in turn, affects the rating scale. Some 4s become 3s, some 3s become 2s, etc. People are more comfortable giving lower ratings to mediocre levels that obviously and objectively carry inferior design quality or gameplay, at least when compared to CCLP3's voting process, in which absolutely abysmal levels were up for consideration. The levels in this category for CCLP1 voting would have probably been the "marmite" levels of CCLP3's process. This category spans such a wide numerical range because there are high outlying votes that tend to raise the rating enough for some of the levels to appear to be in the same tier as the marmites. Levels in this category tend to start off with a 3.00 average and dive down from there.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;">- <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">The underdogs (~3.76 - 4.09). These are the levels that, most of the time, are perfectly great for CCLP1 but are at a disadvantage because they're not quite as "interesting" to voters as the levels in the top tier. Therefore, few 5s are awarded to these levels, which causes their averages to fall into this range whenever lower votes are casted. However, either because the design or gameplay is unique enough, some voters tend to give these levels high ratings, at least enough to save them from a marmite fate. Very difficult yet still CCLP1-appropriate levels tend to fit into here too, as well as easy levels that fit the "unique enough" bill.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">- The populist levels. (~4.10 - 4.75). <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">These are the levels that tend to have some degree of reason for multiple types of voters to enjoy them, and for that reason, they are the smallest group. Getting a high average in CCLP1 voting usually involves including enough variety such that one or a few of its components win someone over or creating a "playing landscape" that's so enjoyable that people can't help but give it a high rating. Levels in this category tend to start off high average-wise and stay that way, although they may drop a bit. They also tend to be medium-difficulty and are designed very well.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">So with this in mind, my predictions for the results are:
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">- We'll have about 100 "populist" levels by the end of voting. Most of them will be quite worthy of inclusion into the set, and many of them probably will form the 90-130 range of CCLP1. We'll also probably have to eliminate many for variety's sake, and that will most likely one of the hardest parts of post-voting set assembly.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">- We'll probably include many of the "underdogs," many of which will be easy levels that deserve to be in the set or a few deserving difficult levels in the 130+ range that inspired enough disagreement to not be in the "populist" tier.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">- The "marmite" category is where we'll probably have to do a lot of work. We'll have to dig deep here to find blob levels, ice levels, mazes, and other not-quite-as-interesting-to-a-veteran levels that would add a lot of variety to CCLP1. How we decide what we'll include will probably depend on what we need with respect to gameplay, difficulty, design style, etc., and the ratings from voting would also probably come into play as well. Many of the not-quite-so-unique-but-still-fantastic easy levels will also come from this group too.
- We'd probably eliminate the levels below the 3.00 threshold.
I agree with the use of a spreadsheet to ensure variety. My suggestion is to use difficulty (perhaps expressed by ranges of levels within which a given level would appropriately fit), gameplay type (puzzle, action, block-pushing, maze, melee, variety, etc.), design style (matriced, symmetrical, modular, "organic," etc.), "feel" (claustrophobic, open, etc.), and of course, predominant tile. While CC1 had variety, I think it could have easily had more in some areas and less in others. We definitely have the capability to produce a truly diverse set (albeit one with consistent top-notch design quality the entire way through) with the levels up for consideration. My suggestion is to reserve a spot for each game element in each third of the set for "predominant tile" levels. We'd have to be flexible with this, though, as some gimmick-exploring "variety" levels may contain a significant enough amount of a certain game element to be eligible for these spots, and the difficulty tiers are also flexible as well with "relief" levels to be considered and what not. Some types of levels may be included just by virtue of not having many in the "populist" tier to begin with (I'm especially thinking of the "contentious" game elements, mazes, and the "open" levels). Perhaps we can also have a "staff-only" yes/no vote with some of the marmite levels to determine how deserving they are of inclusion. In the end, we'll just have to play it by ear.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">- The "marmite" levels (~3.01 - 3.75). <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">This is the largest group of levels by far, and for good reason: they tend to inspire the most amount of disagreement among voters. These are the levels that are loved by some, hated by others, and in some cases, don't exactly inspire votes that carry much variance and affect the results. The average ends up remaining squarely in the middle in the end after bouncing around for a little while. Levels with concepts that polarize opinion between "annoying" and "genius" fall into here, and especially levels that feature some sort of game element or gameplay that veterans approach with some degree of negative bias (blobs, ice mazes, blue/invisible walls, etc.). Very easy levels also fit here too.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">- The rejects (~1.25 - 3.00). <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">These levels are the worst of the bunch. One of the consequences of a pre-voting elimination has been - at least for the most part - narrowing down the spectrum of quality that characterizes the levels being considered, which, in turn, affects the rating scale. Some 4s become 3s, some 3s become 2s, etc. People are more comfortable giving lower ratings to mediocre levels that obviously and objectively carry inferior design quality or gameplay, at least when compared to CCLP3's voting process, in which absolutely abysmal levels were up for consideration. The levels in this category for CCLP1 voting would have probably been the "marmite" levels of CCLP3's process. This category spans such a wide numerical range because there are high outlying votes that tend to raise the rating enough for some of the levels to appear to be in the same tier as the marmites. Levels in this category tend to start off with a 3.00 average and dive down from there.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;">- <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">The underdogs (~3.76 - 4.09). These are the levels that, most of the time, are perfectly great for CCLP1 but are at a disadvantage because they're not quite as "interesting" to voters as the levels in the top tier. Therefore, few 5s are awarded to these levels, which causes their averages to fall into this range whenever lower votes are casted. However, either because the design or gameplay is unique enough, some voters tend to give these levels high ratings, at least enough to save them from a marmite fate. Very difficult yet still CCLP1-appropriate levels tend to fit into here too, as well as easy levels that fit the "unique enough" bill.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;font-weight:bold;font-style:italic;">- The populist levels. (~4.10 - 4.75). <span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">These are the levels that tend to have some degree of reason for multiple types of voters to enjoy them, and for that reason, they are the smallest group. Getting a high average in CCLP1 voting usually involves including enough variety such that one or a few of its components win someone over or creating a "playing landscape" that's so enjoyable that people can't help but give it a high rating. Levels in this category tend to start off high average-wise and stay that way, although they may drop a bit. They also tend to be medium-difficulty and are designed very well.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">So with this in mind, my predictions for the results are:
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">- We'll have about 100 "populist" levels by the end of voting. Most of them will be quite worthy of inclusion into the set, and many of them probably will form the 90-130 range of CCLP1. We'll also probably have to eliminate many for variety's sake, and that will most likely one of the hardest parts of post-voting set assembly.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">- We'll probably include many of the "underdogs," many of which will be easy levels that deserve to be in the set or a few deserving difficult levels in the 130+ range that inspired enough disagreement to not be in the "populist" tier.
<span style="font-family:Arial;color:rgb(0,0,0);background-color:transparent;">- The "marmite" category is where we'll probably have to do a lot of work. We'll have to dig deep here to find blob levels, ice levels, mazes, and other not-quite-as-interesting-to-a-veteran levels that would add a lot of variety to CCLP1. How we decide what we'll include will probably depend on what we need with respect to gameplay, difficulty, design style, etc., and the ratings from voting would also probably come into play as well. Many of the not-quite-so-unique-but-still-fantastic easy levels will also come from this group too.
- We'd probably eliminate the levels below the 3.00 threshold.
I agree with the use of a spreadsheet to ensure variety. My suggestion is to use difficulty (perhaps expressed by ranges of levels within which a given level would appropriately fit), gameplay type (puzzle, action, block-pushing, maze, melee, variety, etc.), design style (matriced, symmetrical, modular, "organic," etc.), "feel" (claustrophobic, open, etc.), and of course, predominant tile. While CC1 had variety, I think it could have easily had more in some areas and less in others. We definitely have the capability to produce a truly diverse set (albeit one with consistent top-notch design quality the entire way through) with the levels up for consideration. My suggestion is to reserve a spot for each game element in each third of the set for "predominant tile" levels. We'd have to be flexible with this, though, as some gimmick-exploring "variety" levels may contain a significant enough amount of a certain game element to be eligible for these spots, and the difficulty tiers are also flexible as well with "relief" levels to be considered and what not. Some types of levels may be included just by virtue of not having many in the "populist" tier to begin with (I'm especially thinking of the "contentious" game elements, mazes, and the "open" levels). Perhaps we can also have a "staff-only" yes/no vote with some of the marmite levels to determine how deserving they are of inclusion. In the end, we'll just have to play it by ear.