후기 영어편

Oh man! I’m writing this on the day I won the show. I just had a satisfying meal, a warm bath, and a sip of wine with some delicious cheese and grapes. Honestly I don’t know if I’ve ever been so happy.

My celebratory snack.

In light of my statistics-fueled victory, I find the following music appropriate to the tone of the remainder of this essay. (caution: contains profanity)

Why English

I did participate in a Korean show about the Korean language, but when it comes to writing I am more comfortable with English. Hence I started with this text and worked it back into the Korean version. I also wrote slightly more personal content here and got a little more wonky, so English seemed to be an appropriate barrier.

Me, from the inside

Since you saw what I look like on the outside on TV, I thought it would be fun to give you the full picture and show you what I look like on the inside, in the form of personality tests. The results of three tests are presented: MBTI, Big Five, and Dark Triad. While the Big-Five personality test seems to be the standard personality test for personality researchers, for whatever reason MBTI is popular in Korea; I wanted to make my results more interpretable to the Koreans reading this. Note that these results are from self-reported surveys, so they measure how I view myself, as opposed to how others view me. I have noticed that those two sometimes diverge significantly, so take these results with a grain of salt. These tests were done on the weekend before I went on the show, so that they would not be affected by any short-term psychological effects of my winning or losing.

Starting with MBTI, done over here:

The thing I don’t like about MBTI is perhaps what makes it appealing: it is discrete. As you can see, outside of being strongly introverted, on the other 3 main categories I am close to neutral. Hence it wouldn’t be strange if my categorization were INFJ or INTP or so on; yet I get a certain ISTJ. (I talked with the producers of the show today and they said I was definitely an ISTJ; go figure.) By the way, this site uses the Big Five personality traits as its fundamentals! It’s not actually MBTI! But whatever.

Next, Big Five, done over here:

I think this provides a better view of who I am: I am curious about a lot of things (high openness); I… am fine being alone, but in general like being in small groups (of 2-3 people) and dislike being in large and noisy groups (low extraversion); I do my best to never show it, but I can be negative and critical (high negative emotionality).

Next, Dark Triad. This is a really lesser-known one, but this paper (showing that dark and evil guys are rated as attractive by girls) caught my attention, so I wanted to give it a shot. I took the test here:

Oops, I’m not a dark and evil guy, maybe that’s why I’m not super popular. Well, I guess if you’re not romantically interested in me, you can think I’m a good person?

Lifestyle

I modified my lifestyle in some ways to perform better on the show, and I wanted to document it / show my effort off. (Also I wanted to show off some of the data I gather about myself.)

Jogging

Starting at July 7th morning, I started jogging in the morning, following the NHS Couch to 5k plan. The reason I went with jogging is primarily because of this Wikipedia article about the neurobiological benefits of aerobic exercise: it explains that “people who regularly perform aerobic exercise have greater scores on … declarative memory …”, and declarative memory is exactly what I wanted to improve.

Over the course of about 80 days I completed the Couch to 5k plan; regardless of the memory improvement, it felt good to improve my physical ability. It certainly felt like my memory was better after exercising, but that might be placebo in effect. Because of the strong subjective feeling that exercising was good, I kept on running for 30 minutes three times a week until the day of the show. After the show… you can probably guess.

Time

Starting from this April, I keep track of where I use my time with a 30-minute resolution. There are many interesting things you can do with this; one is to analyze my sleep pattern before and after the show event. Since Sep. 1st until the day I took the show (Nov. 10th) I generally tried to sleep at midnight and wake up at 07:30. After the show, I didn’t put any restriction on myself and just did whatever I wanted. (…Obviously while doing my best not to interfere with my main job, being a Ph.D. candidate. Love you Prof!) So how much did my sleep pattern change?

…Yeah, there is a ~1 hour shift, but it’s not that drastic, is it. Honestly I thought I could show you a really drastic graph where in September and October I sleep like some kind of monk but after the show I sleep like an extroverted undergrad, but my life is much more stable than I give it credit.

Since I have this data, I have a relatively accurate estimate of how much time I sunk into this project since April: 602.5 hours. Here is a month-by-month graph of time use:

Here, we can see that while the overall time I put in the project steadily decreased, my memorization time mostly stayed stable from April to August. After that I had a personal issue that limited my time investment in this project, but I kept on doing the Anki-prescribed reviews. The uptick of analysis in November is partly because I did some 벼락치기 analysis before the show, and partly because I continued to put in time in these analysis essays after the show.

Happiness

Since 2019, I have been keeping track of my happiness 4 times each day, on a 1-7 Likert scale. The data has many caveats but I won’t bore you with those, let’s just accept the data as is. The question is, was I happier this year since I had the [우리말 겨루기] goal? Obviously there are a trillion confounding factors, but let’s just compare the average happiness of 2019 to 2020:

Can you see which year was happier? The answer is 2020; I was happier by 0.031 points. That kind of tiny difference could be upended by a few good or bad days on either year. So somewhat surprisingly, I wasn’t actually significantly more happy this year despite having this project in my life.

Since I report my happiness at random times of the day, I can also look at happiness by time of day. Here is the graph:

The peak at 12 o’clock is… lunch. (I really like eating.) I usually studied [우리말 겨루기] at 9~11 pm; you can see how I was at my worst mood at 9 and 10 o’clock, then I get much happier at 11, when my studies end. I apparently hated studying, funny. Maybe that’s why I wasn’t happier this year.

What’s next

I had fun this time preparing for [우리말 겨루기], and I am looking for things to do next. I want to use less than one hour a day for the next project, since I have to gear up on my real job, research; I’ve come up with a few things:

  • Let’s get this one out of the way: “normal” hobbies. For example, taking pictures, learning a musical instrument, trying carpentry, and so on. This is completely fine with me. I’m not really a hell-bent statistician; I just tend to like doing things that I am good at. This path very likely would make me no money, and the amount of fun would vary a lot based on what I choose to do and how compatible it is with me.
  • Next on the list is 2022 (Korean) Election Prediction, in which I would develop a model to predict who would become president and which party would win the municipal elections. I would be mostly following FiveThirtyEight here. Actually I did a small-scale project like this back in 2018, where me and my team got the municipal election results right for every 광역자치단체장 race. That model was flawed though and not publicized enough. This time, I think I could do much better both in statistical modelling and spreading the word about the model. This seems to be less fun on average; gathering data would be a real chore, and I’m not incredibly interested in Korean politics right now. I think if I make something that works though, there are many routes to monetization in this case.
  • Next is to be a game YouTuber that does quantitative analysis, a la Spirit Of The Law. This would be fun, and I wouldn’t fail, so to say, because the analysis wouldn’t be as difficult as election prediction for example. But the amount of money I would make would be subject to a lot of luck. Being a successful YouTuber might take more than 1 hour/day, which is another concern.
  • The last one is 수능 prediction. There are some direct parallels to my [우리말 겨루기] project: In both I predict what kinds of questions are more likely to come out; in both there is historic data that spans years that I can use for statistical analysis; in both, I start off with the belief that I have a comparative advantage because I doubt people have tried the statistical toolset. There are also some clear differences that might make this project difficult: making 수능 prediction results useful would be a non-trivial task, especially when trying to benefit myself (since it’s not me who would use the results); evaluation of how well I did in prediction may also not be well-defined. But if I succeed by some measure, I think this would make me the most money of any of these projects.

Influences

  • FiveThirtyEight, [The Big Short], [Moneyball], [Superforecasters]: These four items are listed in the order I got in contact with them, probably. Their common denominator is that they all involve statistic geek(s) who use data to see what other people failed to see. That’s so cool, you know? So yeah, I was definitely influenced by work like this, because I wanted to be someone as cool as that.
  • Gwern: Have you seen this guy’s site? Gwern’s cautiousness with statistics makes my website look like some kindergartener’s picture. I read a of lot of Gwern’s posts last year (starting with the tank post) and the use of statistics and scientific experiments to improve one’s life was interesting (e.g. the LSD microdosing post).
  • Money Stuff: Matt’s columns are the best; I often read them while having lunch. He often talks about weird financial products, which also inspired me to think of something intelligent that I could do. I probably read his columns from late 2018; I loved every single one about We (this one is a classic), I also liked getting introduced to weird news that I would have never seen, like this one about how the advent of online dating made dating itself more ‘efficient’, which in turn made everyone unhappy. Actually I just read the original paper for the online dating thing and it is really interesting! What is particularly interesting is the indirect evidence that dating has actually become more ‘efficient’: divorce rates are decreasing in the younger age group. If you didn’t know what kind of a person I am, now you know.
  • 쓰레기 머학생: (Which I saw when it was in 베스트 도전) This is a funny and harmless webcomic that actually significantly nudged me towards this project. Back then I was a little depressed at the things I felt I couldn’t do; the protagonist of the comic felt that she was socially isolated but tried to construct a satisfactory life by doing things she could do and she liked. I felt that she was somewhat similar to me, and that gave me some hope that I could construct a satisfactory life too. More importantly, instead of looking at the things I couldn’t do, I tried to look at the things I could do; [우리말 겨루기] felt like one of them.

Regrets

  • I wish I had come off as more of a nerd on the show. On TV, I wimped out, simply going “metaphoric phrases are more efficient”, like a normal person. Instead, I should have said, “Metaphoric phrases actually have a study efficiency of about 0.2 points per word studied; that is ~5 times more efficient than more efficient than the average word we know, and ~30 times more efficient than a random word on the dictionary”. Or, “I want to do a non-competitive question; my research indicates that words that overlap with 자물쇠 문제 are ~80% 쓰기 문제”. You know? Despite imagining myself doing stupid stuff like that all year, I was woefully underprepared for the actual interview. I should have honed my nerdy lines in my simulations, but those were almost exclusively focused on evaluating my problem-solving skills.
  • A problem I noticed about my analysis, far into my studies when it was too late, was that I should have examined which (meaning, word) pairs come out, instead of which words come out on the show. That might have reduced my study quantity, and allowed me to study common-but-with-a-bazillion-meaning words, thus increasing my study efficiency. I guess I hope someone else does better analysis than me in the future?
  • I small regret that I have, since I didn’t get the 달인 in the end, is maybe I could have worked harder. But probably not. I wasn’t really in the mental state to do anything more for [우리말 겨루기] most of this year. Honestly I feel that this amount of effort is a lot, given my environment.

Acknowledgements

From data gathering to analysis to visualization to actually memorizing the questions, this was clearly a one-man project, but without the help and support of my peers I would not have been able to complete it. I would like to properly acknowledge the help I got on the way, and celebrate the social ties which facilitated the effort.

First of all, thanks to my parents and my advisor for being amused with this project and generally supportive. If either of them disapproved this side project, well… I would have done it anyway, but I wouldn’t feel so good about it. I am grateful for their understanding. I should also thank my labmates for their understanding here. None of them freaked out saying “well you should have kept that energy for research”; instead they were all supportive of my effort. Now to be honest I wasn’t 100% open about how much effort I put into this project; but they are good people, I don’t think they will get mad at me. (I hope.)

Thanks to H, J, H, C, W, and S for helping me with my simulation efforts. Simulations helped me identify how well I was doing, how much more I should do, and perhaps most importantly how stochastic the final score can be. Special thanks to H.C., who helped me with most simulations. Thanks to H for being particularly amused by this project, and motivating me to work harder. I deeply regret forgetting to take a picture of the last simulation; you guys are the best.

I wish I could enumerate all the friends I talked about this to, but I’m sure I would omit someone by mistake, and I don’t want anyone to be hurt. If I told you about this project in advance, I want to thank you for being nice to me: I never met any negative response about this project, and I am grateful for having such supportive friends. Special thanks to H, S, Y, D, and J for taking enough interest to listen to details about my analysis (which includes bearing my rants about statistics), and even providing feedback.

Thanks to W, T, and H for jogging with me. I can’t measure how much jogging helped my effort here, but it certainly kept me mentally stable, and I’m thankful to these friends who helped me stay motivated to jog.

Thanks to my parents, my labmates, H & T, S & C & H for taking 응원 영상 for me; they used some of their time to make these videos, and I would like to express my gratitude for them making that effort.

Thanks to the [우리말 겨루기] producers, for letting me be on their wonderful show and for taking interest in my website and analysis. One person, I don’t know her name, specifically noted the crowns I put on the winners and complemented me for that. As a person who loves hiding and finding details like that, I felt a particular satisfaction from that comment.

Finally, thanks to myself, for being willing to put 700+ hours in this project to achieve a childhood goal in a spectacular way. I don’t think I could have done much better, and I want to applaud myself for doing my best.