Puzzle sheds light on government policy, corporate America and why no one likes to be wrong

Use falsifiability to your advantage. Consult the oracle

1 Like

The article says no one likes to hear they’re wrong, but after several YES tries I was desperate to get a wrong answer, and after two wrong answers I guessed correctly the rule. You learn more from mistakes than successes.

6 Likes

I may have been helped by my Math degree, but it took me 7 tries with 1 no to generate enough confidence to submit my answer.

Besides the information presented by the sample sequence, I also took into account that this was directed at a general audience (so no point in testing overly complicated solutions), and that it was testing/demonstrating flaws in general problem solving skills.

So I did two tests to verify that the x2 rule was sufficient, a third to verify that the x2 rule was not necessary, a fourth to test the simplest rule I could think of (ascending numbers), and a couple of additonal tests to verify that slightly more complicated rules were not necessary.

1 Like

Daniel Kahneman’s book “Thinking Fast and Slow” is my bible of cognitive biases. Confirmation bias is only the beginning; there are quite a number of predictable cognitive biases that get in the way of good judgement.

I got this in about the same time number of tries as Mark. Even though I realized that increasing was the most likely “trick” answer after just a few tries, it took a while before I was satisfied that there weren’t any other tricks.

This is my dilemna: my bias towards “negative thinking” tends to cause me to take a lot longer to reach answers because I tend to work by ruling things out rather than just accepting a proposition without verification. Kahneman’s book is helpful in teaching me to discriminate which situations it’s more useful to just “trust your gut” (System 1) rather than take the time to think rationally (System 2 in his nomenclature).

He addresses this well when he says that he thinks people with a positive bias are overall happier and more effective in life than those with a negative bias, even though the latter may be right more often.

4 Likes

one sec, taking a look. the ‘rule’ is stored in js, it doesn’t make an http call (according to console) to verify the rule. the event listener appears to be bound here:

document.addEventListener("click",e,!1)}},{1:10,2:8,ee:"QJf3ax",handle:"D5DuLP",loader:"G9z0Bl"}]

i am deobfuscating the event object, QJ, D5, and G9. we’ll see how far i get (all by hand, no automated helpers :D)


huh, i still haven’t found the evaluation code, but this is quite interesting. have i been living under a rock?

http://graphics8.nytimes.com/newsgraphics/2015/06/16/puzzle/3c265b64cd66937d132444db8e9f7edb6a231f29/nlp.min.js

2 Likes

Embarrassingly, the rule I came up with is correct but way more complicated than the real rule:

My rule: “The first number in the sequence is any number greater than 1. The second number in the sequence is the first number in the sequence multiplied by (any of its factors if it is not prime) or (itself if it is prime or is not a whole number). The third number in the sequence is the second number in the sequence multiplied by (any of its factors if it is not prime) or (itself if it is prime or is not a whole number).”

Also of note is that, no matter if I press the “I think I know it” button or the “I don’t want to play” link, nothing happens.

If this really is about civic life, I consider these valuable lessons learned:

a) You may not be playing the game you think you’re playing
b) You may not be playing the game at all

2 Likes

fuckit, i just used a debugger.

var rightWrong = (inputData[0] < 
inputData[1]) & (inputData[1] < 
inputData[2]) ? right : wrong;

the code is pretty tightly obfuscated. it would have taken forever to do it by hand. with the chrome debugger is was about 45 seconds :slight_smile:

Set breakpoint on tree modifications on the top level container for the results. Done in one.

4 Likes

You need to get a PhD in the social sciences before you’re allowed to debug the real world. And no, a JD doesn’t count.

3 Likes

No, it’s not:

Wrong. 1, 2, 3 is a valid sequence. So is 0, 1, 2; -1, 0, 1; etc. etc. etc.

3 Likes

After 6 NOs and 11 YESs, trying fractions and negatives, the rule I came up with was “Alternate doubling and quadrupling of the absolute value of the previous value in any order; zero is not a valid value.”

sigh I definitely needed more NOs.

2 Likes

Is this one of those times that I use my brain incorrectly by taking a less obvious approach to solving the problem, solving it faster and more accurately, but missing the life lessons that were implicit in the solution?

1 Like

If that was good for something, they’d care more to try to replicate their own field’s earlier experiments.

Given how flimsy the social stuff can be, this is a matter of quite some concern.

5 Likes

It could have been greatester (increasing) absolute value…

But MarjaE already nailed what I was trying to say, better, before I made the edit.

But your test doesn’t rule out increasing absolute value… or increasing absolute distance from -312 for that matter.

2 Likes

And yet you had not actually established that the domain was something other than the set of positive integers.

1 Like

Weirdly, I got it super fast by being super lazy.

I typed in 1,2,3 as my first try. Then 4,5,6, and so on… and after that it was just gravy. I didn’t have the mental energy left to bother with a complex puzzle, so I treated it carelessly and was “rewarded.”

Liberals are less likely to recall the many incorrect predictions over the decades, often strident and often from the left, that population growth would create widespread food shortages. It hasn’t.

I’m so sick of hearing this tired old trope get dragged out in every discussion of poverty. More interesting is what happens when cheap mechanization makes it so that the majority of the population cannot exchange labor for goods and services. That’s the apocalypse I’m interested in.

6 Likes

I tested it exactly once, got the answer I expected to get, wrote it up in the box and hit submit.

I’m proud to say that the rule I came up with would produce a correct result 100% of the time for numbers greater than 1. That’s nearly half of all numbers! Good enough for government work.

1 Like

I find it interesting that you think being a maths geek with experience of pattern solving helped you figure it out quickly, because I put my own quick solution (within 6 guesses, not 4) down to how very bad I am at maths. Basically, much like the young children, I simply couldn’t think up creative ways to a more complex, incorrect assumption so stumbled across the correct one almost by accident.

1 Like

I see this bias pop up at the software company I work at a lot. I’m a tech writer. I find an embarassing number of bugs because the QA guys often just test the functionality and ignore failure modes. When I play with a new feature to learn about it, tend to go at like that gorilla in the American Tourister ad. More times than I can remember, I’ve put the wrong value in a field or done something out of order or pushed some envelope and ended up crashing the system, corrupting data, and other things we really don’t want our customers to see.

I got it in about 10 guesses I think. It wasn’t until I hit my 4th guess that I realized I needed to see some negatives before I could tell what I was seeing.

2 Likes

This is an honest question–am I the only person that deofuscated the code and set breakpoints?

Do I have that much time to waste?
Am I that weird?

(One time I interviewed an architect for a small company we were thinking of acquiring, and when I asked about error log files for QA testing he said, “oh just use sysinternals dbgview.exe for that”.

I sat there for about eight seconds and said, " not all QA staff are 32/64 bit legacy and WoW kernel developers."

You had to have been there)

(There are like five people that will appreciate that anecdote. And sometimes I don’t think I’m one of them)

3 Likes