📢Creator requested N/A

This market did not attract as much traction as I had hoped, so I'm resolving to the market price as our best guess of the share of humans able to solve an easy Sudoku.

@Primer I don’t think this is correct, the correct thing to do would be to resolve it when we have more information.
Or at least to N/A, so it isn’t just an self-referential market again.

@dionisos Valid point! My motivation for this market was: Often people say "Humans can do X, but AI still can't do X" and I wanted to have something to point to gesturing "Not even most Humans can do X". I had hoped to find some resolution modality "along the way" via comments, and was obviously too optimistic. Probably there won't be any sources for resolution anytime soon, at the very least not undisputed ones. This is why I don't want to extend the closing time.

I'd prefer to have a resolution other than n/a, but people shouldn't lose Mana due to my preference. So I'll try to re-resolve as n/a.

@Primer Thanks for resolving it N/A

> I'd prefer to have a resolution other than n/a.

In my point of view, people should be able to look at resolved market, and think there were good evidences and at least partial agreement about the resolution (and not it to be mostly random).

> I had hoped to find some resolution modality "along the way" via comments

I am optimistic we would have been able to agree on some modality, but it was not really worth it to try, because what we would not have agreed to, is to pay the cost of the experimentation

unclear how you're ever going to do the study to answer this. Half the population of the world is totally innumerate tho.

I feel like title implies resolution based the likelihood it’s >50% whereas description does not. Would changing the title to what percentage of humans can solve make sense?

@JoshuaB Title changed. Was going for a more attention grabbing title originally, but asking for a percentage directly is the better fit.

Sudoku puzzles are generally considered to be solvable by most humans, especially when they are classified as "easy" puzzles. While there may not be a specific study examining the ability of "most" humans to solve easy Sudoku puzzles, it is widely recognized that these puzzles are designed to be accessible to a broad range of people, including those without advanced mathematical or logical skills.

Easy Sudoku puzzles typically involve a straightforward application of logical reasoning and deduction.

It's important to note that individual abilities and familiarity with Sudoku puzzles may vary. Some individuals may find them easier to solve due to prior exposure or experience with similar puzzles, while others may find them more challenging. Additionally, some people may prefer different types of puzzles or have varying levels of interest in solving them.

With that being said, from my personal experience having offered children between the age of 4-9 to choose between sudoku and chess after having explained both, they opt for chess because they said and I quote "the pieces look more fun", "I want to be a queen", or they just like moving them randomly.

the ones who picked sudoku got bored and didn’t bother going back to it.

I tried with my grandparents, but they prefer dominos because of familiarity and because "at their age learning something new was considered outrageous "

@sarakuqjad61f Thanks for joining in! Everything before "With that being said" came from an LLM, right?

If we understand "percentage of humans" to include e.g. children, people in developing countries, etc the current value seems a clear overestimate

@SGQ Yeah, I'm surprised as well to see this above 50%. Should I clarify "percentage of humans"? It seemed obvious that infants, people in nursing homes, mentally disabled people, people in 3rd world countries, etc. are humans and thus included.

people in 3rd world countries

Why are people in developing country such a disadvantaged demographics? Because of limited access to education?

According to Wikipedia, the global literacy rate for all people aged 15 and above is 86.3%. And it's probably even higher for people above 6; although I'd expect to drop dramatically under that age. If one made the assumption that whoever can read and write can solve easy Sudokus too, then most humans would be able to.

@Benx

Why are people in developing country such a disadvantaged demographics?

Maybe they're not. Do you think literacy is enough to be able to solve a Sudoku? I think having been exposed to a few logic puzzles might help quite a bit. I'm not able to check this, but I'm pretty sure only 1 of my grandparents could have solved one, I'm pretty sure my mother can't solve one (in a reasonable setting; she probably could with enough tries and undos), I did check both of my kids, they can't do it (in the same reasonable setting). All of them can read and write.

@Primer How old are your children ?
My expectation is that almost all people above 10 years old can solve it (but they would need some incentive, what I fear is that a lot of people, particularly children, will just think it is boring and not really try)

@dionisos

• 6 year old: Not a chance

• 10 year old: Starts guessing a number when he's out of ideas. Did about 5 Sudokus together when he was 8 or 9, he was happy to puzzle with me for half an hour. I'd guess chances are 50% he could do it now at sudoku.com (it immediatelly shows when an error was made) with 3 errors or less. But he's quite above average in logic and maths and has spent a few hours on Sudokus already.

• I'd bet 20:1 that a random 10 year old can't solve a Sudoku

@Primer In fact, I don’t think most 10 years old can do it, it was more of an age when I think they start to be able to do it (maybe 10% of them), and then I think it grows quickly from here, toward ~90% at 16.
Now I am quite curious and would like a some stats with the percentage of people by age ^^

I think I would still accept your 20:1 bet (I am more at something like 15:1 that the majority can’t do it), if there were some way to verify it.

@dionisos

Now I am quite curious

Me too!

@dionisos A little heads-up: Meanwhile, the 6 year old is 7 and can do easy Sudokus. So can the (meanwhile) 11 year old, as he taught the 7 year old. I think I wouldn't even take 15:1 now.

I think you should put a time limit on how long they have to solve the puzzle. If you locked people in a cage and gave them years to solve this puzzle, then I'm pretty sure most humans would be able to solve an easy Sudoku puzzle. At least, if you give them the puzzle in their language.

@MrR I'm hesitant to be too restrictive here, as I don't know what kind of data sources we might find (Maybe we stumble upon data from some vietnamese school giving Sudokus as homework assignments; I wouldn't want to exclude that data due to being restrictive on the timing). This is not meant as an adversarial task though, I want to allow the human to have enough time to correct an error, so I'd say at least half an hour. I want to exclude someone very dedicated bruteforcing through 20 random guesses, or szenarios like waiting 10 years for an infant to grow old enough.

I tried to capture this with "reasonable setting" in the criteria, but I'm open to suggestions to operationalize this!

@acc betting against human intelligence, classic bots prejudices

The site doesn’t work for me. Is it the same level of difficulty than this : https://sudoku.com/easy/ ?

@dionisos I think it is similar enough.

@Primer Ok, so I think the probability was too low (I fixed it).

@dionisos I think it's too high. All of my grandparents couldn't do it, both of my children can't. And that's in a Western country with good education.