Survey Results Pt. 3: Manifold's favorites

Alright, everyone. It’s time for the part I know you’ve all been waiting for. What were the winners of the Manifold’s favorites section, and which category had the mysterious upset?

For anyone who doesn’t know how this works, I ran a series of free response markets asking what Manifold’s favorite movie, book, TV show, print comic, webcomic, song, video game, board game, role-playing game, short story, poem, podcast, YouTube channel, and play are. In each of these markets, users could submit any work they wanted as an option, as long as it was a member of the right category of work. Then, on my Manifold survey, I included a section asking what users thought of each option. They could either rank it as “Excellent,” “Good,” “Okay,” or “Bad,” or select that they hadn’t watched it, read it, heard it, etc. To determine “Manifold’s favorite” in each category, I converted these qualitative ratings into scores: Excellent = 2, Good = 1, Okay = 0, and Bad = -1, and then I added up the scores from each respondent to get a total score. I used a total, rather than an average, so that a work has an advantage if more people actually know about it - it wouldn’t be Manifold’s favorite if it’s only beloved by a small but dedicated group of Manifolders.

The scores of every option, as well as some other information about them, like how many people gave it each ranking and the average score, can be found in this spreadsheet. But if you don’t want to spoil the winners, you should read ahead before clicking on that link.

Manifold’s favorite movie

This was the category with the most submissions, 58 in total. At close, the market on the subject predicted The Matrix as the most likely winner, thought it was only 18% confident. The Godfather was also at 18% but slightly behind it (18% is only the rounded probability displayed by the UI, I don’t know what the exact probabilities were).

To make this more suspenseful, I’ll go through the top 20 in reverse order. So we have:

20. Star Wars Episode V: The Empire Strikes Back - (84 points)

19. Oppenheimer - (85 points)

18. Dune (2021) - (89 points)

17. Arrival - (97 points)

16. Toy Story - (98 points)

15. Forrest Gump - (99 points)

14. Fight Club - (105 points)

13. The Shawshank Redemption - (106 points)

11. Pulp Fiction - (108 points, tied with Insterstellar)

11. Interstellar - (108 points, tied with Pulp Fiction)

10. Back to the Future - (111 points)

9. The Princess Bride - (116 points)

8. The Dark Knight - (118 points)

7. The Lord of the Rings: The Two Towers - (120 points)

6. Spirited Away - (121 points)

5. The Lord of the Rings: The Return of the King - (129 points)

4. Inception - (132 points)

3. The Lord of the Rings: The Fellowship of the Ring - (136 points)

2. Monty Python and the Holy Grail - (150 points)

And the winner is….

It was in fact The Matrix, with a total score of 164. 58 people ranked it as “Excellent”, 53 as “Good”, 17 as “Okay”, and only 5 as “Bad”. There were also only 12 people who said they hadn’t seen it, the smallest out of any movie. For reference, the numbers of people who said that they haven’t seen “The Star Wars Franchise” and “The Harry Potter movies” (the entire franchises were submitted as a single entry, in addition to The Empire Strikes Back being an option on its own) were 16 and 18, respectively.

So no big surprises in the movie category. The market actually managed to predict the winner correctly, despite being unconfident, and the runners up are all movies that I’m not surprised at all to see there. I guess the biggest surprise was that The Godfather, which the market considered second-most likely to win, didn’t make the top 20, but it was number 21.

Manifold’s favorite book

There were 12 total book submissions. At close, the market said that The Lord of the Rings was the most likely to win, and it was pretty conident (59%). So, who won? I’ll rank them in reverse order again.

12. Antifragile: Things That Gain From Disorder - (7 points)

11. Infinite Jest - (16 points)

10. The Player of Games - (28 points)

9. The Three-Body Problem - (50 points)

8. To Kill a Mockingbird - (54 points)

6. Foundation - (59 points, tied with Dune)

6. Dune - (59 points, tied with Foundation)

5. The Harry Potter series - (67 points)

4. Gödel Escher Bach: An Eternal Golden Braid - (74 points)

3. Harry Potter and the Methods of Rationality - (110 points)

2. The Lord of the Rings - (117 points)

The winner was The Hitchhiker’s Guide to the Galaxy, with 161 points. This is a slight upset, but not too surprising - Hitchhiker’s was in the lead spot on the market for a while even though it didn’t close there. In all, 67 people ranked it “Excellent”, 28 ranked it “Good”, 11 ranked it “Okay”, and only one person dared to rank it “Bad”. 33 said they hadn’t read it, which was smaller than every other option except the Harry Potter series, but it didn’t even need these extra readers to propel it to the top spot. It had the highest average rating of all of them as well.

Manifold’s favorite TV show

There were 9 options here, and the market closed with Breaking Bad in the lead, at 44%. The scores were:

9. Nova - (17 points)

8. Succession - (25 points)

7. Doctor Who - (40 points)

6. Star Trek: The Next Generation - (44 points)

5. Bill Nye the Science Guy - (46 points)

3. Jeopardy! - (51 points, tied with Game of Thrones)

3. Game of Thrones - (51 points, tied with Jeopardy!)

2. Mythbusters - (100 points)

Another correct prediction! The favorite was indeed Breaking Bad with 109 points from 46 “Excellent” ratings, 22 “Good” ratings, 12 “Okay” ratings, 5 “Bad” ratings, and 59 people who hadn’t seen it. Ironically, even though Nova got last place, it was also the only show that not a single person ranked as “Bad”. Though that’s at in least in part because there weren’t that many people who watch it.

Manifold’s favorite print comic

This was the smallest category, with only 5 submissions. The most likely to win at close was Calvin and Hobbes, at 52%. The final rankings were:

5. Donald Duck - (18 points)

4. Peanuts - (27 points)

3. Watchmen - (56 points)

2. xkcd: volume 0 - (95 points)

Another correct prediction, with Calvin & Hobbes as the winner. It has 110 points in total, from 45 “Excellent” rankings, 24 “Good” rankings, 9 “Okay” rankings, 4 “Bad” rankings, and 54 people who hadn’t read it.

Manifold’s favorite webcomic

There were 12 total webcomic submissions. This one had a clear front-runner, with xkcd closing at 73%. In fact, there wasn’t a single time when xkcd was below 50% except on the day the market was created. Not only that, but a different market gave it a 62% chance of being Manifold’s favorite work overall. That’s a pretty crazy degree of confidence in a single work, but will it pay off? Here are the rankings:

12. Penny Arcade - ( -4 points)

10. Wild Life - (1 point, tied with Dilbert)

10. Dilbert (post- March 2023) - (1 point, tied with Wild Life)

9. Dumbing of Age - (6 points)

8. Gunnerkrigg Court - (9 points)

7. Questionable Content - (13 points)

6. Order of the Stick - (23 points)

5. Homestuck - (26 points)

4. Oglaf - (40 points)

3. Cyanide and Happiness - (42 points)

2. Saturday Morning Breakfast Cereal - (96 points)

The winner was of course, xkcd, in a shock to no one. It had a total of 193 points, with 79 “Excellent” ratings, 39 “Good” ratings, 11 “Okay” ratings, only 4 “Bad” ratings, and only 10 people who haven’t read it. And yeah, those people who bet at 62% confidence that it was Manifold’s favorite work overall were also right (slight spoiler for the other categories, I guess). This was a runaway victory, crushing even the closest competitor by almost 100 points.

Manifold’s favorite song

36 songs were submitted. I thought this one was the hardest to predict - I didn’t have the slightest idea which one would win - but the market ended up closing with “In the End” by Linkin Park at 26%. Honestly, I kind of wish I had noticed how high people were betting it because I definitely didn’t think it had a 26% chance of winning. Or maybe I’m glad that I didn’t notice it in time to push it down, let’s find out. As with movies, I’ll only be listing the top 20:

19. “Love Story” by Taylor Swift - (42 points, tied with “White and Nerdy”)

19. “White & Nerdy” by Weird Al Yankovic - (42 points, tied with “Love Story”)

18. “Love will tear us apart” by Joy Division - (45 points)

16. “Time” by Pink Floyd - (49 points, tied with “Holiday”)

16. “Holiday” by Green Day - (49 points, tied with “Time”)

15. “American Idiot” by Green Day - (50 points)

13. “(Don’t Fear) The Reaper” by Blue Oyster Cult - (55 points, tied with “Highway to Hell”)

13. “Highway to Hell” by AC/DC - (55 points, tied with “Don’t Fear the Reaper”)

12. “Sympathy for the Devil” by The Rolling Stones - (60 points)

11. “Lose Yourself” by Eminem - (61 points)

10. “Megalovania” by Toby Fox - (65 points)

9. “In the End” by Linkin Park - (67 points)

7. “Wish You Were Here” by Pink Floyd - (69 points, tied with “Sweet Child O’ Mine”)

7. “Sweet Child O’ Mine” by Guns N’ Roses - (69 points, tied with “Wish You Were Here”)

6. “Hotel California” by the Eagles - (78 points)

5. “Comfortably Numb” by Pink Floyd - (82 points)

4. “Starman” by David Bowie - (83 points)

3. “Under Pressure” by Queen - (102 points)

2. “Moonlight Sonata” by Beethoven - (107 points)

And the winner is… “Bohemian Rhapsody” by Queen, with 156 points, toaling 64 “Excellent”s, 32 “Good”s, 21 “Okay”s, 4 “Bad”s, and only 9 people who hadn’t heard of it. The only work that fewer people said they were unfamiliar with was the game of chess. Not that “Bohemian Rhapsody” needed that to win. It also had the highest average ranking out of every song except “Keep me in your heart for a while”, which probably only got such a high average because it was so obscure (only 9 people had heard of it).

This one technically marks a slight upset, since the leading option in the market got 9th place, but “Bohemian Rhapsody” was the second-highest option, so the traders didn’t do too bad of a job.

Manifold’s favorite video game

The market this time picked Minecraft as the most likely to win, closing at 41%. Though that percentage is misleading - it was actually much less confident until one trader spiked it at the very end. Before then, it was trading at 14%, and even lower than that earlier on. Will that spike pay off? Let’s see. There were 17 options:

17. Flappy Bird - ( -20 points)

15. Paper Mario Sticker Star - (8 points, tied with PES)

15. Pro Evolution Soccer - (8 points, tied with Paper Mario Sticker Star)

14. FIFA - (10 points)

13. Red Dead Redemption 2 - (16 points)

12. Baldur’s Gate 3 - (22 points)

11. Elden Ring - (27 points)

10. Grand Theft Auto V - (29 points)

9. Pacman - (31 points)

8. The Witcher 3 - (34 points)

7. Super Mario Galaxy - (37 points)

6. Snake - (42 points)

5. Terraria - (44 points)

4. The Legend of Zelda: Tears of the Kingdom - (46 points)

3. Skyrim - (56 points)

2. Minecraft - (113 points)

The winner was Portal 2, with 125 points. That’s a total of 53 “Excellent”s, 19 “Good”s, 5 “Okay”s, and not a single person who thought it was bad! There were also 49 people who hadn’t played it. This is an upset, given that it wasn’t rated as being very likely to win, but the most likely contender did get second place, and it did manage to get more than twice the score of every other contender.

This category had a lot of surprises overall. I didn’t even realize which option had won until I wrote down the list (I wrote the first paragraph of this section thinking Minecraft had won, after looking at the graphs on Google forms). And aside from the upset for the actual winner, we also had Paper Mario Sticker Star and Pro Evolution Soccer tied for second-to-last. This wouldn’t surprise me that much, since I’ve never played either game, except that, for a while, those two were predicted to be the most likely winners by the market (they both had long stretches of being in the lead before Minecraft pulled ahead). I was also pretty surprised that Snake got 6th place, though it’s pretty obvious why in hindsight: A lot more people have played it than most of the other games.

Manifold’s favorite board game

The market predicted that chess had a 53% chance of winning this one, and as I’ve already mentioned, it was the most well-known work on the entire survey. So did it win? Here are the results:

11. Chutes and Ladders - ( -49 points)

10. Monopoly - ( -22 points)

9. Othello - (41 points)

8. Coup - (49 points, tied with Pandemic)

7. Pandemic - (49 points, tied with Coup)

6. Ticket to Ride - (56 points)

5. Dominion - (59 points)

4. Settlers of Catan - (83 points)

3. Go - (84 points)

2. Codenames - (88 points)

The winner was indeed Chess. It had 154 points, and was considered by 59 people to be “Excellent”, 41 to be “Good”, 28 to be “Okay”, and 5 to be “Bad”. There were only three people who said they had never played it. No surprises there, but what does surprise me is just how much people hated on Chutes & Ladders and Monopoly. I mean sure, they’re not as good as the other games on this list, and maybe people voted “Bad” on them just to say they were bad by comparison, but they ended up being the two most hated works on the entire survey, with Monopoly just barely edging out Flappy Bird for that coveted second-to-last spot. It’s just surprising, given that they both seem innocent enough.

Sorry, Quantum Gambler, but it appears the degens have said otherwise.

Manifold’s favorite role-playing game

There were 22 different role-playing games submitted, thanks in large part to @evergreenemily.

But could any of them beat the clear front-runner, Dungeons & Dragons? The market didn’t think so, giving D&D a 53% chance of winning at close. Here were the rankings:

22. Gothic - (2 points)

21. Planescape Torment - (3 points)

20. Mario & Luigi: Dream Team - (6 points)

19. Lancer - (7 points)

18. Pokemon Sun/Moon - (10 points)

17. Divinity: Original Sin 2 - (12 points)

16. Pokémon X/Y - (13 points)

15. Final Fantasy VII - (14 points)

13. Earthbound - (15 points, tied with Shadowrun)

13. Shadowrun - (15 points, tied with Earthbound)

12. Pathfinder - (20 points)

11. The Elder Scrolls IV: Oblivion - (21 points)

10. Baldur’s Gate 3 - (23 points)

9. The Elder Scrolls III: Morrowind - (25 points)

8. Pokémon Gold/Silver/Crystal - (30 points)

7. The Witcher 3 - (33 points)

6. Pokémon Red/Blue/Yellow - (34 points)

4. Fallout: New Vegas - (38 points, tied with Disco Elysium)

4. Disco Elysium - (38 points, tied with Fallout: New Vegas)

3. Undertale - (53 points)

2. The Elder Scrolls V: Skyrim - (54 points)

The winner was Dungeons & Dragons, as expected, with 74 points (19 “Excellent”s, 38 “Good”s, 13 “Okay”s, 2 “Bad”s, and 47 people who hadn’t played it). This came down largely to which works were the most well-known. D&D won because a lot more people have played it than any other RPG listed here, which isn’t a surprise. Skyrim does, however, have the distinction of not a single person saying it was bad, despite having 52 people who had played it. Most of the others here were very well-regarded by people who had played them as well, but many were very obscure.

Manifold’s favorite short story

“The Last Question” was the front-runner here, with a 34% chance at close. Can it beat the other 10 entries? Let’s find out.

11. “A Casino Odyssey in Cyberspace” by Roger Williams - (9 points)

10. "BLIT" by David Langford - (12 points)

8. '—All You Zombies—' by Robert Heinlein - (14 points, tied with “The Lady or the Tiger?”)

8. "The Lady or the Tiger?" by Frank R. Stockton - (14 points, tied with '“All You Zombies”)

7. "The Colour Out of Space" by H. P. Lovecraft - (23 points)

6. “The Scorpion and the Frog” (unknown author) - (30 points)

5. "500 Million, But Not A Single One More" by Jai Dhyani - (38 points)

4. "Nightfall" by Isaac Asimov - (39 points)

3. “The Egg” by Andy Weir - (47 points)

2. "...And I'll Show You How Deep the Rabbit Hole Goes" by Scott Alexander - (81 points)

The winner was “The Last Question” by Isaac Asimov, with a total of 96 points, a great showing for a short story (most had pretty low scores because few people had read them). In total, it had 42 “Excellent”s, 14 “Good”s, 6 “Okay”s, 2 “Bad”s, and 54 people who hadn’t read it, making it tied with "...And I'll Show You How Deep the Rabbit Hole Goes" for the highest average rating of a short story among people who had read it as well. This category also had multiple well-known works that not a single person disliked: "...And I'll Show You How Deep the Rabbit Hole Goes", “Nightfall”, “The Colour Out of Space”, and “-All You Zombies-” all had 0 “Bad” rankings and at least 10 people familiar with them. Even the last place choice had no “Bad” rankings, but only 8 people had read it.

Manifold’s favorite poem

The front-runner for Manifold’s favorite poem was “Ozymandias” at 35%. How did this prediction do?

11. “The Dash” by Linda Ellis - (3 points)

10. "Dirge Without Music" by Edna St. Vincent Millay - (8 points)

9. "The Wasteland" by T.S. Elliot - (25 points)

8. "Hymn of Breaking Strain" by Rudyard Kipling - (27 points)

7. "If—" by Rudyard Kipling - (32 points)

6. "Howl" by Allen Ginsberg - (35 points)

5. "Do not go gentle into that good night" by Dylan Thomas - (55 points)

4. "The Road Not Taken" by Robert Frost - (64 points)

3. "The Raven" by Edgar Allan Poe - (68 points)

2. "Jabberwocky" by Lewis Carroll - (73 points)

Yet another accurate prediction. “Ozymandias” by Percy Shelley was the winner, with 79 points. It had 25 “Excellent” ratings, 29 “Good” ratings, 9 “Okay” ratings, and not a single “Bad” rating. “Hymn of Breaking Strain” and “Dirge Without Music” also had no haters, but the latter had only been read by 8 people. There were 52 people who had never read “Ozymandias”, which was lower than most other poems, but it was actually read by the fewest people out of the top four.

Manifold’s favorite podcast

Here, The Market Manipulation Podcast by Manifold’s own Conflux was given the highest chance of winning (35%). Here are the actual rankings:

12. All-in Podcast - (0 points)

10. Machine Learning Street Talk - (6 points, tied with Well There’s Your Problem)

10. Well There’s Your Problem - (6 points, tied with Machine Learning Street Talk)

9. Science & Futurism With Isaac Arthur - (8 points)

8. Cum Town / The Adam Friedland Show - (9 points)

7. Crystal Ballin' - (11 points)

6. The Market Manipulation Podcast - (13 points)

5. The FiveThirtyEight Politics Podcast - (14 points)

4. Welcome to Night Vale - (23 points)

3. The Dwarkesh Podcast - (25 points)

2. Lex Fridman Podcast - (26 points)

The winner was the 80,000 Hours Podcast with 30 points, 11 “Excellent” ratings, 8 “Good”s, 5 “Okay”s, and no “Bad” ratings (plus 86 people who had never listeneed to it). The Dwarkesh Podcast, Crystal Ballin’, Science & Futurism With Isaac Arthur, and Well There's Your Problem also had no “Bad” votes, though not as many people were familiar with them. Interestingly, the Lex Fridman Podcast, despite getting second place, also had the lowest average rating out of any podcast here, except for the All-in Podcast. It just had so many more people who knew about it to push it nearly into the lead.

This one marks an unsuccessful prediction. The front-runner didn’t even make it to the top five, but instead was stuck in the middle, along with the other Manifold podcast, Crystal Ballin’. Though the eventual winner was ranked as the third-most likely to win at close.

Manifold’s favorite YouTube channel

Destiny was the front runner for pretty much the entire time the market on this one was open. He closed at 63%, and was at around 40% for most of the time the market was open. This makes sense, since Destiny has a special relationship with Manifold that no other YouTube channel has. I don’t know the exact details, but it’s something along the lines of, “Destiny advertised Manifold on his channel, and this caused a bunch of his fans to join the site and make ‘stocks’ about Destiny and related figures.” So he has a big presence on the site, despite not actually being a user himself (unless he really is Wobbles). So, how did he fare?

I think it will actually be more dramatic if the list the scores here from highest to lowest.

The winner, by an overwhelming margin, was 3Blue1Brown, with 158 points. This isn’t that big of a surprise in and of itself: 3Blue1Brown was rated second-most likely to win. 3Blue1Brown didn’t just win, though: It had the highest average score (defined as the score divided by the number of people who had seen/read/played/etc. the work) out of any work whatsoever, with a spectacular 68 “Excellent”s, 23 “Good”s, and only 2 “Okay”s and 1 “Bad”. It was also very well-known, with only 35 people who said they hadn’t watched it. The other educational YouTube channels also did very well, making up the entire top five. But how did Destiny do? Let’s see the rankings:

2. Tom Scott - (114 points)

3. CGP Grey - (107 points)

4. Veritasium - (77 points)

5. Vsauce - (69 points)

6. Rational Animations - (47 points)

7. Game Maker's Toolkit - (30 points)

8. exurb1a and exurb2a - (29 points)

9. Half as Interesting - (26 points)

10. Mr. Beast - (10 points, tied with Nexpo)

10. Nexpo - (10 points, tied with Mr. Beast)

11. Something Witty Entertainment - (8 points, tied with Posy)

11. Posy (PosyMusic) - (8 points, tied with SWE)

12. Destiny - (7 points)

13. The Cursed Judge - (0 points)

That’s right. Not only did Destiny not win, but he got nearly dead last, only managing to beat out a YouTube channel too obscure to get any points (only 2 people had watched it). How did the market get this so spectacularly wrong? Well, there are two main factors. First of all, opinion on Destiny was very divided. In fact, he had the exact same number of fans (“Good” or “Excellent” ratings) as haters (“Bad” ratings). The only reason his score turned out positive was becase the “Excellent” ratings give an extra point to outweigh the “Bad” ratings. The second reason is that he is just way less popular on Manifold, or at least among my Manifold survey respondents, than anyone thought. In fact, only 30 people had even watched his channel. Compare this to 3Blue1Brown, who had been watched by 94 respondents.

I have a few theories as to why this might be. First of all, while he is popular among a small ingroup of Destiny fans, that doesn’t translate to the broad popularity on Manifold needed to win the Manifold’s favorites competition. I’m also not sure how many Destiny fans there really are on Manifold. Sure, there were those stock markets that got thousands of traders, but those have sort of disappeared by now, and I haven’t seen much of the Destiny community lately - my guess is that a lot of the Destiny fans just signed up for the stocks and eventually got bored. Another possibility is that there’s selection bias on the survey. Since it was mainly advertised through the markets that I made on it, the results were always going to be at least somewhat biased towards my interests, since people who share my interests are more likely to trade on my markets. I tried to spread it as far as possible to reduce this effect, but there was no way to completely counter it. I have never watched Destiny and rarely interact with the Destiny side of Manifold, so it’s certainly possible that there are still a lot of Destiny fans here who never even heard about my survey.

Of course, to some extent, this is post-hoc rationalization of why the market got it so wrong. I don’t actually know for sure why this prediction failed so badly. But I will brag a bit about predicting that Destiny was overvalued in that market (just ignore the part where I also didn’t have faith that 3Blue1Brown would win).

Manifold’s favorite play

We’ve reached the final category. According to the market prices, Hamilton was the front runner at 50%, with Hamlet in second at 17%. Who will win the War of the Hams? There’s only one way to find out. There were only 6 submissions:

6. Cats - ( -3 points)

5. Waiting for Godot - (24 points)

4. Sweeney Todd: The Demon Barber of Fleet Street - (27 points)

3. Les Miserables - (53 points)

The top 2 was an extremely close match-up. Hamilton had 59 points, but Hamlet just managed to edge it out by a single point. In the closest finish of the night, Hamlet wins Manifold’s favorite play with 60 points, 22 “Excellent” rankings, 20 “Good” rankings, 12 “Okay” rankings, 4 “Bad” rankings, and 59 people who hadn’t seen it. This is another weird case, though, where Hamlet actually had the lowest average rating except for Cats, and it only won because more people had seen it than any other play. Though it’s worth noting that all of the plays here (except Cats) had good average scores, and they’re all pretty close to each other, so it’s not like people didn’t like Hamlet.

Which works had the highest average ratings?

This concludes the actual judgement of the winners, but I did have a few other things I wanted to get from the data. As we just saw, Manifold’s favorite in a category wasn’t always the work that was rated the highest by the people who were familiar with it. It also had to actually be popular enough that lots of people knew it. I think this is the most faithful interpretation of what Manifold’s favorite is - otherwise, the winners could be works with a devoted cult following but that most people on Manifold had never heard of - but it’s still interesting to see what works Manifold ranks as the best.

So here were the works from each category with the highest average rating:

Movie: The Shawshank Redemption (1.43)

Book: The Hitchhiker's Guide to the Galaxy (1.50)

TV show: Breaking Bad (1.28)

Print comic: xkcd: volume 0 (1.44)

Webcomic: xkcd (1.45)

Song: Keep me in your heart for a while (1.33)

Video game: The Legend of Zelda: Tears of the Kingdom (1.64)

Board game: Chess (1.16)

Role-playing game: Fallout: New Vegas (1.41)

Short story: tie between “The Last Question” and “…And I’ll Show You How Deep the Rabbit Hole Goes”, each with exactly 1.50 points on average.

Poem: Hymn of Breaking Strain (1.29)

Podcast: The Dwarkesh Podcast (1.47)

YouTube channel: 3Blue1Brown (1.68)

Play: Sweeny Todd (1.29)

So, eight of the categories end of with a completely different winner, and one ends up with the previous winner tied with another option. There is a problem with this methodology, however. Some of the works on the survey were very obscure, and as a result, they only got a few rankings. This means that their average rankings aren’t that meaningful, since the sample size is small. If we exclude any works that had fewer than ten people who had watched/read/heard/played them, “Bohemian Rhapsody” goes back to being the best-ranked song (1.29). Here are the top five in each category when sorted by average rating, ignoring those with fewer than ten ratings:

Movies

1. The Shawshank Redemption (1.43)

2. Spirited Away (1.42)

3. Monty Python and the Holy Grail (1.40)

4. Goodfellas (1.36)

5. The Princess Bride (1.32)

Books

1. The Hitchhiker's Guide to the Galaxy (1.50)

2. Harry Potter and the Methods of Rationality (1.34)

3. The Lord of the Rings (1.26)

4. Gödel Escher Bach: An Eternal Golden Braid (1.23)

5. Foundation (1.09)

TV Shows

1. Breaking Bad (1.28)

2. Mythbusters (1.14)

3. Star Trek: The Next Generation (1.00)

4. Nova (0.89)

5. Bill Nye the Science Guy (0.88)

Print Comics

1. xkcd: volume 0 (1.44)

2. Calvin & Hobbes (1.34)

3. Watchmen (1.33)

4. Donald Duck (0.58)

5. Peanuts (0.47)

Webcomics

1. xkcd (1.45)

2. Saturday Morning Breakfast Cereal (1.10)

3. Homestuck (0.87)

4. Order of the Stick (0.77)

5. Oglaf (0.74)

Gunnerkrigg Court (1.29) and Dumbing of Age (1.20) would have been on this list, but they each only had 9 readers.

Songs

1. Bohemian Rhapsody (1.29)

2. Comfortably Numb (1.28)

3. The Ruler of Everything (1.28)

4. Love will tear us apart (1.25)

5. Moonlight Sonata (1.22)

“Keep me in your heart for a while” would have been included, but it only had 9 people who had heard it. Also, note that these are rounded to the nearest hundredth - “Comfortably Numb” did slightly edge out “The Ruler of Everything” (albeit by a statsistically insignifant margin) in the average.

Video Games

1. The Legend of Zelda: Tears of the Kingdom (1.64)

2. Portal 2 (1.62)

3. Baldur's Gate 3 (1.29)

4. The Witcher 3 (1.26)

5. Minecraft (1.23)

Board Games

1. Chess (1.16)

2. Go (1.14)

3. Dominion (1.11)

4. Codenames (1.09)

5. Coup (1.00)

Role-Playing Games

1. Fallout: New Vegas (1.41)

2. The Witcher 3 (1.38)

3. Disco Elysium (1.27)

4. Baldur's Gate 3 (1.21)

5. Undertale (1.13)

Lancer (1.40) also would have been included, but only 5 people had played it.

Short Stories

1. “...And I'll Show You How Deep the Rabbit Hole Goes” / “The Last question” (1.50, tie)

3. 500 Million, But Not A Single One More (1.41)

4. '—All You Zombies—' (1.40)

5. Nightfall (1.39)

Poems

1. Hymn of Breaking Strain (1.29)

2. Ozymandias (1.25)

3. Jabberwocky (1.07)

4. The Wasteland (1.00)

5. Howl (0.97)

“Dirge Without Music” (1.00) would have been included, but only 8 people had read it.

Podcasts

1. The Dwarkesh Podcast (1.47)

2. 80,000 Hours Podcast (1.25)

3. Crystal Ballin' (0.85)

4. Welcome to Night Vale (0.68)

5. The Market Manipulation Podcast (0.65)

Science & Futurism With Isaac Arthur (1.00), Well There's Your Problem (0.86), and Machine Learning Street Talk (0.67) would also have been included, but they only had 8, 7, and 9 listeners, respectively.

YouTube Channels

1. 3Blue1Brown (1.68)

2. CGP Grey (1.19)

3. Tom Scott (1.19)

4. exurb1a and exurb2a / Rational animations (1.00, tie)

Posy (1.33), Nexpo (1.25), and Something Witty Entertainment (1.14) would have been included, but they only had 6, 8, and 7 viewers, respectively.

Plays

1. Sweeney Todd: The Demon Barber of Fleet Street (1.29)

2. Les Miserables (1.18)

3. Waiting for Godot (1.09)

4. Hamilton (1.05)

5. Hamlet (1.03)

The Most Divisive Works

So far, I’ve been focusing on works that Manifold likes. This encompasses most of the works. After all, people typically only submitted a work if they like it, or if they think enough other people might like it for it to win. There were a few exceptions, however.

There were six works that had negative scores: the board games Chutes & Ladders (-49) and Monopoly (-22), the video game Flappy Bird (-20), the webcomic Penny Arcade (-4), the play Cats (-3), and the 2017 movie It (-1). I already talked about Chutes & Ladders and Monopoly, and Flappy Bird doesn’t really surprise me. I’ve never read Penny Arcade, so I can’t really say whether it deserves to be here. I was kind of surprised to see It here because it seemed like a fairly normal movie that wouldn’t have been bad enough to get a negative score. It just felt average to me, but I guess that was exactly why it ended up here.

Cats was the work that I was referring to in my first post when I said, “There was at least one that I submitted just to find out what the people who have seen it think, because everything I’ve heard about it makes it sound bad, and yet it’s well-regarded.” Apparently, it’s not well-regarded by Manifolders because a plurality of those who saw it did in fact think it was bad. Though it’s possible some people had it confused with the movie adaptation which I know is very badly regarded. Anyway, all I can say is Release the Butthole Cut.

While these works had the most negative opinions on them, I also wanted to learn which works had the most divided opinion. The simplest way to do this is to determine which works had the highest standard deviation of scores. However, I excluded any works that had fewer than 10 rankings because that’s just not enough to decide whether they’re divisive or not. In addition, in my market on the subject, I said I would use the corrected sample standard deviation, rather than the uncorrected sample standard deviation, since this is an unbiased estimator for the variance. The only difference there is dividing by the number of rankings minus one, rather than the raw number of rankings. I don’t actually know if this was the right decision, though: the unbiased sample standard deviation gives larger values than regular sample standard deviation for works with fewer rankings, so using it could have made it more likely that a very obscure work was the “most divisive”. It turns out it didn’t matter, though. The most divisive work, by either metric, was Infinite Jest, with a standard deviation of 1.20, or a corrected sample standard deviation of 1.22. And the second-most divisive was “BLIT”, with a standard deviation of 1.16 and cssd of 1.22. Third place did depend on which metric was used. For the regular standard deviation, it was Destiny (1.15, 1.17), but for cssd, it was Pro Evolution Soccer (1.14, 1.19).

The market on this one didn’t do very well. It gave Destiny a 42% chance of being the most divisive, which was the closest prediction it made, but that was probably too much confidence to put in a single answer out of hundreds. Out of the other five options that were explicitly submitted (Mr. Beast, Game of Thrones, “Love Story”, Homestuck, and The Godfather), only Homestuck was in the top 20 (it was the 8th most divisive using either measure). The rest were probably even further down - I didn’t look farther than the top 20. This wouldn’t be a bad prediction on its own, if the market hadn’t considered any of the submitted options to be likely to be the most divisive, but “Other” closed at 23%, meaning that traders were pretty confident that one of the options submitted would be the most divisive. To be fair, though, it was much higher until the very end.

Conclusion

Manifold’s favorites in each category are:

Movie: The Matrix

Book: The Hitchhiker’s Guide to the Galaxy

TV show: Breaking Bad

Print comic: Calvin & Hobbes

Webcomic: xkcd

Song: “Bohemian Rhapsody” by Queen

Video game: Portal 2

Board game: Chess

Role-playing game: Dungeons and Dragons

Short story: “The Last Question” by Isaac Asimov

Poem: “Ozymandias” by Percy Shelley

Podcast: 80,000 Hours Podcast

YouTube channel: 3Blue1Brown

Play: Hamlet

And the overall winner, with the absolute largest point total, was xkcd.

The work with the highest average rating was 3Blue1Brown, with an average rating of 1.68, one of the few averages to round to “Excellent”. And all this despite not even being the front-runner according to the market. Oh, how they fool ya!