Entry tags:
Snowflake Challenge #7 & #8
(Where's Challenge #6? I've decided to do it later when I have more time and am less panicked about work.)
Challenge #7 - In your own space, interact with a community or a fic.
Done. This one is always easy for me because commenting on fic tends to come quite easily to me. Also, I am really behind on the Quodo tag on AO3, so there's plenty of works for me to comment on at the moment. I am kind of rushing through my reading so I didn't leave as many comments as I liked, but I did leave comment on three different people's works, and I'm still in January in my backwards-moving fic backlog. I hope to leave a few more.
Challenge #8 - In your own space, create a quiz or a poll (or tell us your thoughts about answering quizzes/polls).
I was a bit inspired by
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
hamsterwoman's general ramblings about personality quizzes to share some of my thoughts on what I think make good and bad "What X are you?"-type personality/sorting quizzes.I used to really like taking personality/sorting quizzes and sharing the results on LJ/DW (I've got a quizzes tag (all f-locked)) but I feel like personality quizzes have really gone downhill since their heyday on... (looks at the quiz site I shared my results from the most) ...Blogthings.com? I don't actually know much about that site, but this post I found when Googling the site name just now suggests that it was one woman's project and that all the quizzes were written by her... Which I guess would explain the consistently high quality of the quizzes!
But anyway, ever since uQuiz became the dominant platform for making and sharing sorting quizzes, the chance I will enjoy taking a quiz AND also feel like the result I got is worth sharing has declined sharply, almost to zero. Occasionally you get an interesting quiz but most of the time, the experience is just very blah. Here's my theory as to why:
In order to design a good quiz, I think there are two main considerations: the first, main, and most critical trait of a good quiz is meaningfulness. That is, whatever result you get, it should say something interesting and insightful about you. Sometimes what the quiz predicts about you is just off the mark / totally wrong, but the important thing is to make a claim about the person based on their answers, and the test-taker then has something to ruminate on, about whether they agree with that assessment or not, like "Hm, I do see how I share this part of my personality with Aang, although overall I think I'm more like Suki," or whatever. The opposite of meaningfulness is arbitrariness or randomness. I think this is where so many uQuizzes fail, because they give you a result and it's just like "You are most similar to Uncle Iroh!" or whatever with NO explanation of why the quiz gave you that result. Even if the quiz is interesting or entertaining to take, if I get a result that is just like "You are X!" with no other insight into my personality or character or what the quiz was drawing out from the answers I was giving (even if totally off-base!), my desire to share the quiz result drops to zero.
The second key trait of a good quiz is opaqueness. After you have secured meaningfulness, then (where possible) it's also ideal to have high opacity, by which I mean that the result you get from the quiz is difficult to predict just looking at the answer set for a question. This is the property where the connection between answer choices and each possible quiz result has been somewhat obscured or obfuscated. The opposite of this property is transparency/obvious correspondence. This property is good to have, not only because it makes the quiz more entertaining to fill out (you don't want it to be super obvious to the quiz-taker at every single stage exactly what it means to select each answer and therefore also what result someone is going to get) but it also buttresses the aim of meaningfulness by making it harder for someone to easily game their answers and then get an entirely unsurprising (and therefore uninsightful) result.
Illustration of these principles
Let me illustrate with some examples:
1) High opacity, low meaningfulness: First, let's take the dreaded "Pick a song lyric" question, which is an unfortunate staple of uQuiz type quizzes. Generally speaking, I find these questions to be high opacity (which is good -- it is difficult to glean what result you will get when you pick a particular lyric), but low meaningfulness (which is bad -- it is often entirely unclear what picking a certain lyric says about you at all, let alone what connection it has to the character options).
I think the quiz-maker, when they write these questions, intends to pick out a suitable lyric for each quiz result option, but the problem is that the quiz-maker often has a very limited genre of songs they're familiar with/into, and the song lyrics chosen therefore tend to be super generic and similar in vibe/mood to the point of being almost interchangeable with each other... Also, most of the time when I encounter these questions I don't feel anything about the different lyrics at all and choose one at random based on individual words that stand out to me or whatever. If they say anything interesting about me OR the option I get at the end, that would be news to me...
So in short, these questions tend to go too far into the obscurantist direction and tip over into "meaningless." They are a good illustration of the "high opaqueness, low meaning" failure mode.
2) Low opacity, high meaningfulness: To demonstrate the opposite end of the spectrum, let's take an alternative to the dreaded lyric question: what if we had a "What character from [canon] are you?" type quiz and instead of asking people to pick a lyric, we asked people to pick the quote they relate the most to, and picked out one quote from canon for each of the possible characters who are potential result options that we felt summarized that character in a nutshell.
This would be an example of a low opacity, high meaningful question because anyone who is familiar with the canon will be able to identify the representative character quotes and know what character they will be scored towards when they select quotes. However, if you do select quotes that get to the essence of the character, it's still possible for this to be a meaningful question, where selecting a quote that stands out to you could be an indication that you and that character are generally pretty similar.
3) Low opacity, low meaningfulness: Some questions are both low in opacity AND meaningfulness. For example, let's say I was writing a quiz to determine "Which Avatar nation do you belong to?" If I had a question like this: "Pick your favorite color: [options:] Red. Yellow. Green. Blue."
This question would be a low-opacity question, as in, it is extremely easy for the test taker to predict exactly how each answer maps to each result option. Not only is it a low-opacity question, though, but it's also a low-meaningful question too, because sorting someone into a nation based on their favorite color is a pretty shallow/arbitrary sort of thing to do and doesn't say much about your personality. Also, this question forces people to pick one of four -- very limited -- options for their favorite color and so a lot of people will be forced to pick a color that's not even their favorite and (combined with the low opaqueness) it's possible that choice could largely just be a reflection of which nation they would like to get as their result.
That's not to say low-opacity, low-meaningful questions are bad or have no place in a sorting quiz. In our Avatar nation example, even if we stuck to pretty shallow sorting dimensions, it could still be meaningful for someone whose favorite color is orange/yellow and who likes mountains and who likes windy weather and whose favorite animal of the options given is a lemur to feel satisfied that they are indeed closest to an Air Nomad, even if all of those questions are fairly shallow ones that pick out some surface-level resemblances only. But even if the sorting criteria is somewhat arbitrary, as long as there are enough different dimensions being measured, it can still feel satisfying to realize that your answers do most closely resemble what an Air Nomad might answer as opposed to a Water Tribe member and so you can still feel like the result is meaningful overall.
4) High opacity, high meaningfulness: As a case study, let me take a random Blogthings.com quiz that I found while looking at my old posts. The title of the quiz is, "What Color Should Your Eyes Be?" This is an example of a quiz that is opaque at the whole quiz level -- it's impossible for me to know in advance what result I'm going to get because I can't say I've ever thought about what color my eyes "should" be or even what that means!? What are the options, and how are the answers I give connected to those result options? It is a mystery. Because the results are high opacity, so are the questions.
However, this completely opaque quiz can still be meaningful. Here is the result I got: "Your eyes reflect: Innocence and sweetness. What's hidden behind your eyes: A calculating mind." I can no longer recall what the questions were, but given this result, I can still kind of evaluate whether I think this result fits me or not, and I'm guessing the questions were designed to sort me into the arbitrary personality archetypes that the quiz-maker came up with for eye colors. Because the results and questions are tied to personality, this is an example of a quiz with high meaningfulness even though the result categories featured in the quiz are just made-up nonsense.
Case study: Writing a good quiz
Okay so I've never actually written a finished quiz. But I have at least plotted out several quizzes and I think here is what I would advise based on that experience: To increase both meaningfulness and opacity, what I think works best is, first, analyze your set of result options and try to figure out what would distinguish between different pairs of options. Second, focus in particular on personality traits, and try to figure out a small set of personality traits that has a lot of variation across the result set and therefore is good at distinguishing different characters from each other. Using personality as the main thing being tested for in quizzes is both deep (meaningful) AND opaque (as you are not measuring the quiz-taker's direct resemblance to characters but rather which personality traits they embody which you THEN map onto the closest character after the fact, so personality forms a layer of indirectness).
Here is an example from one of my draft quizzes: A long time ago, I wanted to write a "What Soulcalibur character are you?" personality quiz written in PHP. The problem with Soulcalibur is that it has a HUGE cast. In the case of my quiz, I made a fairly non-exhaustive list of major characters and had 25 options. This would have been even more if I had added SCIV and SCV characters, but this quiz was planned before those games came out. Having 25 possible result options presents an interesting challenge. No way am I writing questions that give one answer per character -- that would be both a pain to write and tedious for the quiz-taker to scroll through 25 options for most of the questions. By necessity, then, I needed to come up with questions that group characters into clusters based on similar answers to a question.
As a result, I began to think about what tends to differentiate most of the SC characters from each other. I felt like, in the case of SC, the most noticeable differences are the personalities that they display when they fight someone (this variation is also present in the generic "custom character creation" voices too): some people are eager to fight (e.g. Mitsurugi, Mi-na, Yun-seong); others view it as an inconvenience and are annoyed and curt when facing an opponent (e.g. Taki, Astaroth), while others are sad to fight and apologize for being forced to beat you into the ground (e.g. Sophitia, Talim); some are playful and view battle like a game or form of entertainment (e.g. Raphael, Xianghua), while others are serious and somber when fighting (e.g. Kilik, Rock). Another obvious difference between characters is how evil-/good-aligned they are, which admittedly doesn't make a great test dimension, although with a Soulcalibur quiz, you could probably be a bit silly and put some edgy "Mwhahaha world conquest!" type answers as options and some people might go for them because they know not to take the quiz questions TOO seriously. But a slightly less ridiculous dimension might be how selfless or service-oriented a character is.
So I think a Soulcalibur personality quiz should probably primarily try to figure out where the test-taker falls on dimensions like Playful vs. Serious, Selfless vs. Self-absorbed, etc. Different Soulcalibur characters can fall at the extreme ends or be toward the center, and depending on the kind of answers the test-taker gives, you can sort them closer to certain characters. For example, Mitsurugi is more toward the serious side (although I don't think maximally) but is on the maximally self-absorbed end. Taki is extreme serious but leans selfless. Yun-seong is playful but extreme self-absorbed. And so on. A single question trying to determine which personality characteristics the test-taker has doesn't need to give too many answer options to be able to fairly accurately come up with a result that the test-taker can see their similarities with.
Not all the questions have to revolve around personality either. Shallow/arbitrary questions like "what's your favorite color" or "what's your hobby" or "pick a weapon to use" are fun to fill out and can also be useful as tiebreakers, especially if you have two characters with similar personalities that don't get differentiated by the personality questions.
Unfortunately, I tried out uQuiz just now and it doesn't really have a way to track one level of personality traits and then assign a final result depending on those intermediate values -- but other quiz formats could potentially do this. I suppose you can kind of approximate it by just assigning characters to different personality answers and hope it kind of averages out in the end.
If I were to design a quiz for Star Trek characters (yes, I've thought about it a little), some of the personality dimensions I would pull out would be how interpersonally warm a character is (as there are plenty of Star Trek characters who occupy the extreme personable end and the extreme emotionally detached end) and also generally how well-adjusted/cool-and-collected someone is (as there are characters who are cool under pressure and characters who are neurotic). Maybe something about the way the character approaches problems or what kinds of problems seem worth solving? I would need to think a bit more to go into more detail than that, but those are the kinds of things that stand out to me with the Star Trek cast vs. the Soulcalibur cast.
Why uQuizzes suck
So to summarize, here is why I think uQuizzes tend to suck compared to the "golden age" of personality quizzes. I'll separate these into things that I think are preventable (feasible to do on the uQuiz platform but people just tend to not do) and things where the platform itself is too limited to create good quizzes.
Improvements people should make to their uQuizzes
- Make your result blurbs meaningful! Don't just say "You are Aang!" in the results -- write a blurb that describes what kind of person will tend to get the Aang result in your quiz. This is so easy to fix and so few quiz-makers do this.
- In support of meaningful result blurbs, think about what actual personality traits your questions are picking out. What does it MEAN to get Sokka instead of Toph? What are the main sources of variation among the result options?
Limitations of the uQuiz format
- uQuiz doesn't allow you to track intermediate personality traits and then assign final results based on the nearest neighbor in that personality space, which is something you could achieve through custom PHP quizzes (for example). I think having this ability would help quizzes be much more meaningful and also insert a layer of opacity as well. Unfortunately, because of the limitation of the format, you end up getting answers needing to correspond more directly to the result choices, which is low opacity.
- uQuiz also doesn't allow someone who has taken the quiz to browse through the full set of result options (I think). I think a feature like this would make quizzes more meaningful as it allows people to compare the result they got to other alternatives and see if there's another option that fits them better that they didn't get. Along with this, it would also be nice to have the option to see a ranked list of result options, ordered with the best fits at the top. That way, even if the person disagrees with or is confused by the top result they got, they might be able to see "Oh I see from the top three results that this quiz does generally get the kind of person I am" (or maybe not!).
Conclusion
Good quizzes need to be meaningful to the quiz-taker. This means that the result needs to make a claim about the quiz-taker which the quiz-taker can then chew on and see if they agree with or disagree with. Features like well-written result blurbs that make a prediction about the quiz-taker's personality based on their answers and providing the user the ability to look at the other results or see the scores of all results ranked would go a long way to improving the experience of taking quizzes and make me much more tempted to share the results I get. I think uQuiz is unfortunately quite lacking in many regards and I think people should consider more versatile hand-coded quizzes. That said, there's also some low-hanging fruit on the uQuiz platform that would allow for more engaging quizzes. But at the moment, I find the uQuiz offerings to be quite lacking and limited.
no subject
Most folk don't have the background or the understanding (or time, patience, and resources) to make a 'good' quiz. So I put quizzes that I don't know the pedigree of in the 'astrology' category. Sometimes, something will come up right, but that's as much luck as anything. It's for fun and not to be taken seriously.
Thank you for taking the time to share this with us. And thank you for participating in Snowflake!
no subject
I think even something a bit arbitrary and meaningless as astrology can at least be entertaining if it tries to make a concrete claim about your character/personality, but a lot of these quizzes don't even try to do that! I hope more people actually write result blurbs for their quizzes because even if the test doesn't actually do a good job of sorting you into an appropriate result, it would at least make the effort of taking the quiz feel more worth it. ^^;
no subject
Oh wow, I would not have guessed that was a one creator project!
This was very interesting to read, because this is very much the kind of stuff I like to ponder also. Wholeheartedly agreed with you on the "pick a lyric" questions -- I hate them, because I usually end up either picking the one lyric I know which song it's from (it is rare that my music taste coincides with the quiz creator's) or something that sounds OK based on the words (but then I'm missing the whole mood of the song). It's basically a wasted question for me, and in shorter quizzes probably skews the results a bunch, too.
It was very interesting to read through your drafting plans for a quiz with a lot of results options. I've thought about that a little before (that 17 Houses thing :P), and my assumption was that you pretty much have to pick some axes and sort into high/low (or high/low/medium) along those axes, kind of the way MBTI does. But that's essentially coming up with a brand new personality type system, or coopting an existing one and labeling points along that field with character/House/whatever names. (And, yeah, I was also thinking of the shallower questions like weapons and colors as tie-breaker questions.)
- uQuiz also doesn't allow someone who has taken the quiz to browse through the full set of result options (I think). I think a feature like this would make quizzes more meaningful as it allows people to compare the result they got to other alternatives
Yes! This is a big problem for me with "modern" quizzes -- I want to know what the other results are! (especially if the logic of the assignments is also explained, as you say in your first point). I especially like something some of the very "serious" quizzes that have a lot of options do, where they give you detailed results that show your percentage match to each character that is an option within them.
no subject
Yeah, I think that's the best way to do it. Coming up with brand new personality typing systems (tailored to a specific fandom) *is* pretty fun though (albeit quite a lot of work for something that ultimately doesn't really matter that much!). XD
I want to know what the other results are!
I wish uQuiz would change this. It's a small adjustment that would make their quizzes more interesting across the board.
I especially like something some of the very "serious" quizzes that have a lot of options do, where they give you detailed results that show your percentage match to each character that is an option within them.
Yep, that's always great, especially in cases where you're just not a particularly good fit for any character just because the whole cast isn't very like you in general, so the percentages carry some meaning as well/it does make a difference if the top result is like 90% vs. 60%. Or you can look at the worst matches and be like "Yep, that's accurate."
no subject
no subject
no subject
Sadly, I've stepped back from active writing myself at the moment for a variety of reasons but I still love leaving comments on various tags over at AO3 when I can. I'm sure those you leave comments for greatly appreciate them! ♥
no subject
And same, I'm sure your comments make the authors of those fics very happy! ^_^ It's fun to leave a little bit of positivity on someone else's work.
no subject