Hacker News Reader

Comment by bjourne | Hacker News Reader

Maybe a mathematician can explain whether this "trick" works or not? Intuitively, I can't see that knowing whether the M&M demand was filled makes it more probable that the other demands are filled.

Say you have a pile with five black or white marbles. You want them all to be black. So you check that the first marble in the pile is black (ie no brown m&m:s). Is it now more probable that the other four marbles also are black?

Because you are just checking one specific marble instead of sampling a number of randomly chosen marbles (which of course would increase the probability), I don't see how it can work.

View on HN

Because it's not like marbles. It's not a math thing.

It's more like a canary in the coal mine. If the canary dies, you should check on the air quality. If they went through the trouble of removing the brown m&ms you have a fairly strong indicator they actually read the document in the first place. And a strong indicator they have a good attention to detail.

Exactly, a concept also known as a "litmus test", if you pass the test you may still fail the actual, but if you fail the litmus test there's more or less no way you'll pass.

I'm all about data being presented with confidence bounds but if even 1 canary is dead I'm getting the hell out of there!

You assume independence between the color of the marbles. This is highly unlikely. Promoters probably don't read the terms randomly and completely randomly fulfill them. Likely whether one of the marbles is black or white tells you a lot about whether the other ones are black or white, especially if it's a marble that you selected before hand in a way that will often change depending on how the marbles are colored.

But you are assuming there exists a positive correlation between absence of brown M&M:s and whether the other demands are also filled or not. I don't think you can show that since the M&M demand was singled out. Had they instead randomly inspected once piece of equipment and found an error that would have been totally different.

This is not about rigorous statistics, it's about human nature. In the sixth grade, my math teacher offered a test at the beginning of the year that included quite a lot of complicated instructions. The last instruction told students to ignore all the previous instructions, and just write their name at the top of the paper and turn it in.

Guess how well the typical top scorers in the class did on this one.

That prank is horrible because the document is self-contradictory yet assumes a single correct interpretation, unless the document is very carefully written and formatted using an agreed semantics.

A well-written document would be a great illustration of Haskell-style lazy evaluation, though

I actually got that test in middle school. (I presume my teacher read about it and thought it was a cool idea.)

The one I got was very explicit: Read the entire test first, and only after you've read all the questions and instructions are you to begin work.

Let's just say that I failed and I was pissed about it, without any particularly good basis or justification, other than being 12-ish years old.

But you are assuming there exists a positive correlation between absence of brown M&M:s and whether the other demands are also filled or not. I don't think you can show that since the M&M demand was singled out.

Yes, that is the assumption. Many mathematical models make assumptions like that. It seems much more likely than the opposite assumption which is that they are independent.

Since you're attempting to model human behavior you are going to have to make assumptions that are ultimately incorrect. That's how humans process information. We join the dots and infer things in countless ways that mean we could be incorrect, but which none-the-less are better than wildly guessing (http://en.wikipedia.org/wiki/List_of_cognitive_biases). It would be probably better, if you are trying to model this, to think of it as a machine learning classifier. Brown M&Ms existing would be a prominent feature, not proof.

First of, now we know from another commentor that that the M&M-test didn't work at all since catering was handled by a completely different crew than those setting up the stage equipment (https://news.ycombinator.com/item?id=7754636).

Second, let's imagine that M&M:s and safety is handled by the same people. Then brown M&M:s could easily be negatively correlated with equipment errors because a competent technician would spend more time double-checking equipment than full-filling bullshit demands.

Its not that they expect to see a bowl of brown M+Ms - its that they expect to see someone come back to them during negotiations with - "M+M's - seriously?"

If no-one comes back they know they need to give that venue more attention.

Yeah, the quote on that page mentions an even better litmus test for this: "Provide us with 15A outlets that give 19A of electricity." This should be an immediate red flag for any facilities manager reviewing the contract before it's signed.

Reply on news.ycombinator.com