Summer of Math Exposition

Presented by 3Blue1Brown 3blue1brown

What a simple bluffing dice game reveals about different views on probability

Audience:

Tags: bayes-theoremconditional-probabilitiesdouble-treeunit-squaredice-game

Using the well-known bluffing dice game Maier (German; English: Mia), different perspectives on probabilities are explained. The focus is on understanding conditional probabilities and how Bayes’ formula can be applied in this context when using additional information from other players to estimate whether they are lying. Additionally, various research-based visualizations are presented that have already been shown empirically to be helpful strategies for dealing with conditional probabilities. Based on these considerations, decision rules are derived, which are then tested in a simulation to evaluate their practical applicability.


Analytics

6.39 Overall score*
61 Rank
12 Votes
8 Comments

Comments

3.5

I was glad that I’ve learned new game. At first it sounded really intriguing. I expected it to be combination of probability theory and bluffing, similar to poker. Problem is I don’t consider it bluffing, if there is only one choice. It looks like you don’t need to have any meaningful strategy to play once you decided you accept the number of previous player. You just say the number you see if it’s bigger than previous number and you make up high-enough random number otherwise. It’s not based on players personality, it’s not something where you can observe someones special strategy. Just play without any emotion as computer would and you can’t be predicted. Maybe there could be some side-channel that leaks information, like the blinking mentioned in the video. But this seems like huge stretch that’s not practically applicable.

I guess this video might be useful for teaching a kid about Laplace’s rule, but I think that could be done in simpler way, without 6 minutes explaining rules of the game. If you show me dice and explain me how can I calculate probability that I’ll roll number less than 3, you would teach me the same counting method, but it would be in 30seconds instead of 7 minutes. Then the video jumps to explaining the use of Bayes theorem, and it gets quite convoluted. There are probably better examples to teach Bayes theorem on with better motivation behind them. This explanation just took too long and was convoluted.

Few random technical details I’ve noticed throughout the video:

  • It’s little bit hard to understand the speech due to low audio quality, especially for me as non-native speaker. Some parts of video had music that was too loud and made the speech comprehension even more difficult.
  • At 4:32 I think it should be “C doesn’t believe him” instead of “B doesn’t believe him” if I understand the rules correctly.
  • Roll number text is cropped at the top of the screen.
4

The video achieves its goals, but could have benefitted a lot from more polish. My main gripe is that I believe the same information could have been conveyed more succinctly. There were some unexpected pauses that disrupted my attention. The music was much louder than the voiceover, though adequately placed. I think the video‘s main strength is the motivation since it nicely illustrates how easily statistics can be applied in real life, but its novelty apart from that seems limited.

7

The idea of using a lot math to do better in game is itself a pretty interesting things. And your video explained how the strategy work very well. and the background music is good. but some showing section is kind of too fast for me to understand what is going on before it end.

6.3

The number of different methods you visualized this with, starting with the dice roll grid, continuing through the double tree and the unit square, all work very well together in communicating this idea.

7.5

Great intro, quickly explained the background and what the video seeks to accomplish (does so in less than 30 seconds).

The overview was also well done, it’s nice to know what to expect from a video.

The animations fit the video, and were executed well.

Great application of probability and bayes rules in terms of game theory. I liked the merge of “hard” and “soft” observations, i.e. the percentage of rolling a number, and the percentage of a player lying based on if they blinked or not.

I’ve never heard of a double tree before, it was cool.

Overall, a really well made video. If I have any complaints, I’d say maybe speed up the pacing a bit, but that might just be me, as I’m already familiar with Bayesian Statistics, so someone learning for the first time might actually appreciate the slower pace. Also the audio mixing needs some improvement, the music at 26:39 is much louder than the other audio tracks.

3.9

I think this video is too long for students. Today I’ve seen 2 other videos on Bayes Theorem that are clearer, shorter and they also introduce some memorising techniques. I also want to suggest to display larger written (or just key words) despite small tables. Really, something can’t be read. Don’t give up!

6.5

I think my 9th or 10th grade student would enjoy learning about probability using the Maier game example, which this video utilizes well-enough. Teaching this subject in class I would ask more guided questions to establish the probabilities of each number combination. really liked the visualization of the unit square split in two and the two diverging graphs. maybe braking this video into two parts, one about Maier and one about conditional probability (blinking) could be better off.
My students would probably question the strategy of not lying on the first cast, If they knew on one was checking.

7.6

Cool video, I like it! There are quite a few typos in the displayed text though; it may be of value to team up with someone to check things over.