Why bother with nearly three months of effort to collect this data?
April 24, 2019 12:20 PM   Subscribe

I found two identical packs of Skittles, among 468 packs with a total of 27,740 Skittles

Possibly Wrong wanted to know how many packs of Skittles you'd need to go through to get two identical packs (the same numbers of each different flavor). The initial estimate was 524 -- not too far from the actual number. (via kottke.org)
posted by Etrigan (54 comments total) 34 users marked this as a favorite
 
This is amazing.
posted by clawsoon at 12:38 PM on April 24, 2019 [2 favorites]


Ode to Joy by Miroslav Holub

You only love
when you love in vain.

Try another radio probe
when ten have failed,
take two hundred rabbits
when a hundred have died:

only this is science.

You ask the secret.
It has just one name:

again.
posted by lalochezia at 12:39 PM on April 24, 2019 [20 favorites]


wait the green ones are supposed to be apple? My childhood was a farce.
posted by HumanComplex at 12:43 PM on April 24, 2019 [5 favorites]


"Yeah, I learned from this experiment that I don’t actually like Skittles"

Heh.
posted by asperity at 12:46 PM on April 24, 2019 [9 favorites]


wait the green ones are supposed to be apple? My childhood was a farce.

What did you think they were? Broccoli?
posted by bluefly at 12:50 PM on April 24, 2019 [8 favorites]


wait the green ones are supposed to be apple? My childhood was a farce.

No! It wasn't! Apple is new. They changed it from lime to green apple in 2013.

Lime was better and I miss it dearly.
posted by Bulgaroktonos at 12:51 PM on April 24, 2019 [66 favorites]


They changed from lime to apple fairly recently.
posted by LionIndex at 12:52 PM on April 24, 2019 [1 favorite]


This is great. It's the kind of thing I'd want to do but weeks later I'd be stuck debugging the machine that sorts the Skittles by color and I'd probably give up and just eat all the Skittles while smashing the machine.
posted by Clinging to the Wreckage at 12:52 PM on April 24, 2019 [4 favorites]


wait the green ones are supposed to be apple? My childhood was a farce.

Anything green that isn't lime flavored can get fucked.
posted by bondcliff at 12:52 PM on April 24, 2019 [67 favorites]


candy doesn't need three citrus flavors
posted by prize bull octorok at 12:56 PM on April 24, 2019 [8 favorites]


Broccoli?

I'd seriously kill for a roasted garlic/brussel sprout one. Maybe a candied yam one too. Parmesan roasted cauliflower?
posted by bonehead at 12:58 PM on April 24, 2019 [3 favorites]


For a while I tried to use Skittles to correct low blood sugar episodes. I'd be playing rec league soccer, blood sugar dropping fast because I hadn't calibrated my insulin correctly, trying to jaw my way through a bag of Skittles which had somehow turned into a giant lump in my mouth. My body would be dying for sugar, and it would all be trapped in that disgusting giant lump, and my jaw muscles would be running out of steam trying to chew it.

It was an unpleasant experience. I've since switched to giant Rockets.
posted by clawsoon at 1:00 PM on April 24, 2019 [6 favorites]


can we talk about the creepy contagious skittles skin disease commercial that has been running in my market for, as best as I can recall, the last 20 years straight
posted by prize bull octorok at 1:01 PM on April 24, 2019 [11 favorites]


I'd seriously kill for a roasted garlic/brussel sprout one. Maybe a candied yam one too. Parmesan roasted cauliflower?

*grump, grumble* Fucking Harry Potter generation... *grumble, grump*
posted by The Tensor at 1:02 PM on April 24, 2019 [11 favorites]


This suggests that the dispenser that fills each pack targets an amortized rate of weight or perhaps volume, got jammed somehow resulting in an underfilled pack, and in getting “unjammed” overfilled the subsequent pack

If I know factory equipment, a few skittles got stuck in the chute and were then pushed down by the next batch. Each batch is prepared by weight which is approximate.

His data supports the thesis that the way candies are prepared is:
-> Each candy color made together in batches
-> They all feed into a mixer (again, by weight for equal distribution)
-> The mixer, where colors are already randomized, then feeds into individual boxes.
posted by vacapinta at 1:06 PM on April 24, 2019 [5 favorites]


Have any of these candy companies ever produced a limited run of 'All One Color' packets? When I was a child I would have killed for a roll of 'All Pink' Sweet-Tarts.
posted by Atom Eyes at 1:10 PM on April 24, 2019 [1 favorite]


Skittles had flavors? I sort of felt they tasted like crystallized sugar with a bit of lemon.

Didn't stop me from eating them back in the day; that ambiguous artificial flavor was just part of childhood candy.
posted by linux at 1:11 PM on April 24, 2019 [4 favorites]


"candy doesn't need three citrus flavors"

candy doesn't need you.
posted by komara at 1:18 PM on April 24, 2019 [34 favorites]


> candy doesn't need you.

Lord, don't I know it. (quiet sobbing)
posted by boo_radley at 1:27 PM on April 24, 2019 [10 favorites]


candy doesn't need three citrus flavors

You're right. That's not nearly enough.

It needs more than three. Grapefruit, tangerine, and yuzu, for example.
posted by evilangela at 1:27 PM on April 24, 2019 [28 favorites]


candy doesn't need three citrus flavors

I haven't had Skittles in years, but when I saw the comments about lime versus green apple I started trying to remember what the green ones tasted like. I didn't have to try very hard. Turns out my brain remembers all the flavors, all the small distinctions between yellow and green and orange, and then the red, and the purple, and my mouth is actually watering now on the strength of remembered flavors I haven't encountered in years. Who knew they were that vivid.
posted by trig at 1:31 PM on April 24, 2019 [11 favorites]


Bergamot skittles.
posted by amtho at 1:32 PM on April 24, 2019 [23 favorites]


I love things that are flavored with actual fruit.

I don't much care for skittles.

Why bother with nearly three months of effort to collect this data?
I like this machine, but can't help but think that all that effort could easily have been put to more productive use. I suppose I'd better actually read the damn thing.
posted by aspersioncast at 1:38 PM on April 24, 2019


Waste the rainbow...
posted by chavenet at 1:39 PM on April 24, 2019 [10 favorites]


I did a gifted summer program thing once where we studied the statistical distribution of M&Ms colors. It was an experiment that really needs to be repeated every so often.
posted by Foosnark at 1:43 PM on April 24, 2019 [1 favorite]


This is great! I think they really need to explore the sampling distribution of that estimate, though ;)

If you're nostalgic for the lime skittles, you can always make your own.
posted by Westringia F. at 1:47 PM on April 24, 2019


bondcliff: Anything green that isn't lime flavored can get fucked.

I'd like to hire you to have a word with the sick fuck at my last job who brought in gourmet jelly beans one day. Most of them were fruity flavours.

Except one.

While I have most definitely have nothing against savory flavours, there should have been clear signage indicating that the yellow ones were BUTTERED POPCORN-flavoured before I shoved a handful in my face, expecting lemon, or at bare minimum, something on the citrus-y spectrum.

If disappointment has a taste, it's definitely those buttered popcorn jelly beans.
posted by mandolin conspiracy at 2:04 PM on April 24, 2019 [12 favorites]


Oh, I always thought those were cool! (Though I think I had warning from the package that they existed.) Once I had a California Chardonnay that the only way I could describe the finish was "Jelly Belly brand buttered popcorn jelly beans." I didn't buy that wine again.

As for green Skittles, lime rules, green apple drools.
posted by Daily Alice at 2:42 PM on April 24, 2019 [3 favorites]


Anyone remember the person who smashed thousands of Skittles together to find the strongest one and sent it to the company? I can't find it.
posted by starfishprime at 2:50 PM on April 24, 2019 [2 favorites]


starfishprime: "Anyone remember the person who smashed thousands of Skittles together to find the strongest one and sent it to the company? I can't find it."

M&Ms, not Skittles.
posted by namewithoutwords at 3:00 PM on April 24, 2019 [8 favorites]


Black Current Skittles. Taste the Forbidden Fruit!
posted by bonehead at 3:15 PM on April 24, 2019 [4 favorites]


Fucking Harry Potter generation...

Rowling stole from the best.
posted by bonehead at 3:17 PM on April 24, 2019 [1 favorite]


This is magnificent. On the surface this may seem like a "why the fuck would anybody do that?" and possibly move to dismiss it as a trivial waste of time. But in its own way it's a mathematical proof and the scientific method all rolled into one experiment. He:

1) formulated a series of questions based off of observation (the copy off of a Skittles bag);
2) created a hypothesis using both his knowledge and others, as well as observation, then made a prediction by applying math (a statistical hypothesis if you will);
3) tested the prediction by collecting data;
4) analyzed the data, including assumptions, to see if observations matched the hypothesis;
5) refined the hypothesis and proposed areas for future work/questions.
(The next step, of course, would be to reproduce the experiment.)

But here's what great: he also showed that the statistics worked. He painstakingly demonstrated that the mathematical equations behind the original prediction functioned as expected. Stats can be kind of abstract in nature, especially to people with non mathematical backgrounds and big numbers. It's one thing to "prove" an equation or a statistical test by flipping a coin 30 times; that's accessible. It's another matter to apply those equations to really large data sets that involve thousands, even millions, of numbers, even though the principles are the same and the equations have been proved, and say to others, "it still works." A mathematical background gives you the "trust" in those scaled up applications because you've been through the work (as well as understand the limitations) but for those without that it's hard to "trust" because our minds and our "gut instincts" quite so often desire "proof". So it can be extraordinarily difficult to take those data sets, those equations, and those methods to non-mathematicians and say, "Trust these results" and get them to believe it. Because without that kind of labor in one's background, it's hard to understand that the math does work and how - that you don't have to count out all the skittles by hand to "prove" it.

As science and mathematics have advanced, as computing power has become more powerful, and scientific problems have become more and more complex, this has become a factor in current scientific literacy. Without getting too much into the state of the world right now and those behind it, one of the problems with topics like vaccination and climate change that involve lots of data and statistical applications is that it's really easy for entities and individuals to use that non-understanding and say, "Statistics lie!" Or to say trust in one "scientist" who runs counter to all the other scientists because other scientists are "manipulating the data" for [reasons]. And alllll the hundred other associated narratives that people believe; a well crafted lie that serves as visceral "evidence" can appeal more strongly as the "proof" one desires than math one doesn't understand.

While I'm not unpacking the entire book-sized-topic of all the components of current scientific literacy (U.S. centric version) involved in this situation, here's one element: it's easier to believe those narratives if you lack certain knowledge. IMHO there's a lot of insecurity involved in the denial of scientific evidence; it's probable apathy about scientific issues also involves some insecurity. People don't like feeling dumb. If you don't understand a concept, sometimes it's easier to deny it than it is to believe it. Let's be clear: I'm disregarding the "absolutists" on certain issues, who are determined to follow a certain belief and not to change their minds - and I'm not talking about those who nitpick details/ play "what about" games/generally disregard logic (although there's some insecurity there as well). It's the people who feel insecure but who are open that I want to reach when it comes to educating the public about science. And they're out there. Those people are the ones I want to feel comfortable reading, hearing, and caring about science, without feeling insecure about their science background. I want people to feel capable. We need accessible scientific examples that can be broken down to an comprehensible premise and result. We may not be able to show them the math behind the scenes, but you can show them examples of science and math working - proof in action, so to speak.

While there are many accessible examples out there, there's not many statistical ones. But here's one! It'd be great for students, certainly. But it's not necessary to understand the equations because one can understand the process and results. It's easy to observe a mathematically based prediction. Then there's tangible data - you can literally see it in a picture, you could comb through it yourself if you wanted to - that shows the accuracy of the prediction. (Also admirable: that collection methodology.) One could easily replicate it. There's no "manipulation" or "magic" involved. It doesn't involve a bias or external motivation - it's just counting Skittles. It doesn't involve a "small" number - one could transfer that understanding to other studies and scale it up. The scientific method involved is also comparable and transferable to other studies. It's an understandable project, and the process of grasping it could induce feeling capable of comprehending science and math after all. That's important! (Again, we're disregarding a certain determined subset of people.) Although this post was written by a mathematician for other mathematicians, it'd be very doable to write it up in such as way that it's understandable for a general audience (not saying easy - science education for the general public is hard). One could even walk people through the equations.

We need more examples like this - overall, it's accessible, visual, and complete. It's beautiful.
posted by barchan at 3:18 PM on April 24, 2019 [54 favorites]


^ tl;dr: This is a wonderful and accessible "experiment" that people could use as an example of statistics working as predicted with visual and understandable proof, for science education/improving science literacy (for students and for those members of the public who are open to that), and we need more examples like this. 5 stars, would replicate
posted by barchan at 3:27 PM on April 24, 2019 [8 favorites]


Bergamot skittles

Hey, I went to school with ol' Bergie Skittles!
posted by Greg_Ace at 3:40 PM on April 24, 2019 [4 favorites]


While there are many accessible examples out there, there's not many statistical ones. But here's one! It'd be great for students, certainly. But it's not necessary to understand the equations because one can understand the process and results.

First of all, thank you for this comment, barchan.

I was always a terrible math student in high school, but a lesson that stuck with me was in a finite math class where the teacher brought in Lotto 6/49 cards and had us figure out how much we'd have to spend on a single drawing to have a 50% chance of winning the jackpot. The answer was: you needed to spend $11 million. The jackpot was (IIRC) somewhere around two or three million dollars at the time - either way, it would have been a loss of the full $11 million, or the difference between the two. His point was "Even if you won in this scenario, you'd still be out money. Voila! Probability."

This is where the figurative penny dropped for me about lotteries and gambling in general - and I learned a little math that's actually stuck.
posted by mandolin conspiracy at 3:54 PM on April 24, 2019 [7 favorites]


But what happens when the jackpot gets to $25 million and your net expected value for a ticket is more than its cost? Do you empty your bank account and spend it on lottery tickets because the math says you should?
posted by any portmanteau in a storm at 5:22 PM on April 24, 2019


But what happens when the jackpot gets to $25 million and your net expected value for a ticket is more than its cost? Do you empty your bank account and spend it on lottery tickets because the math says you should?

This was $CAD circa 1992. I'd imagine the probability for Ontario's 6/49 is still equally shitty. The most recent advertised Ontario 6/49 jackpot is $5 million, so I'd imagine this particular lesson, taught to me in an Ontario school, still applies.
posted by mandolin conspiracy at 5:27 PM on April 24, 2019


Black Current Skittles. Taste the Forbidden Fruit!


Ah! I didn’t know that about blackcurrants being banned in much of the US!

Well, no matter...I prefer to take my blackcurrants in crème de cassis form, anyway.
posted by darkstar at 5:28 PM on April 24, 2019


What's going to stick with me is that while there may be an average of 59 Skittles per pack, the quantity can range between 55 and an astonishing 65. I'll never shop for candy the same way again.
posted by Flashman at 5:37 PM on April 24, 2019 [4 favorites]


candy doesn't need three citrus flavors

Or does it need only citrus flavours? I can't think of a time when a red, pink, or purple candy tasted good. And my favorite Jelly Belly flavour is grapefruit. A whole rainbow of citrus would be amazing.
posted by Margalo Epps at 8:50 PM on April 24, 2019 [2 favorites]


variable candy package contents, a new reason to bring the drug scale to the convenience store
posted by idiopath at 9:42 PM on April 24, 2019 [2 favorites]


Oh man, this is tremendous. A hearty hear hear to barchan's take on it.

Skittles had flavors? I sort of felt they tasted like crystallized sugar with a bit of lemon.

Skittles are the Skeksies to La Croix's Mystics, or maybe vice versa.
posted by cortex at 10:01 PM on April 24, 2019 [4 favorites]


I have distict feelings about skittles flavors and I am not a citrus candy fan at all. Avoid. Avoid. Avoid. I like bananna runts, soo I have been told I am not okay.

I thinkn best example of taste influence is from a friend mom who told her kids that Teddy Graham's with the paws up taste better than paws down because she just wanted some cookies.
She got cookies and she also witnessed subsequent fights with her children over eating the good cookies and not sharing when they're were perfectly good cookies in the bag.
posted by AlexiaSky at 2:28 AM on April 25, 2019 [3 favorites]


But what happens when the jackpot gets to $25 million and your net expected value for a ticket is more than its cost? Do you empty your bank account and spend it on lottery tickets because the math says you should?

The math changes because ticket buying syndicates enter the game and a lot more people decide to buy tickets so you have to factor in the rapidly increased risk of a shared prize.
posted by srboisvert at 4:35 AM on April 25, 2019 [2 favorites]


I learned from this experiment that I don’t actually like Skittles

Me either. I have yet to forgive the first Skittles I ever tasted for their complete lack of chocolate.

M&Ms I can just about accept as an inferior substitute for Smarties, but removing the chocolate altogether from my brightly coloured food pellets is complete overreach by the imperialist running dogs of US candy.
posted by flabdablet at 5:57 AM on April 25, 2019 [1 favorite]


One of the wonderful ladies in our board game group brought a whole box of "all reds" Starburst to our last board game day. The whole box was just strawberry, cherry, and fruit punch. Best day ever.

Back to Skittles. The "Sweet Heat" (black package) variety are SO GOOD.
posted by xedrik at 7:08 AM on April 25, 2019 [1 favorite]


weeks later I'd be stuck debugging the machine that sorts the Skittles by color and I'd probably give up and just eat all the Skittles while smashing the machine

Yep. And my kids would be like "Finally! Can I have our stuff back now?" And I'd be like: "Mindstorms, yes. Skittles, not so much."
posted by The Bellman at 7:33 AM on April 25, 2019 [1 favorite]


It'd be great for students, certainly.

But if you do get lots doing it, then you should run the next statistical prediction - about how many students have to run this experiment before you expect one to get a match in the first ten packs?
posted by solotoro at 7:42 AM on April 25, 2019 [1 favorite]


One of the wonderful ladies in our board game group brought a whole box of "all reds" Starburst to our last board game day. The whole box was just strawberry, cherry, and fruit punch. Best day ever.

I've had those and for some reason they tasted different. Like maybe the reds and pinks need the leaking flavors of the others to get the taste right.
posted by LizBoBiz at 8:31 AM on April 25, 2019


But what happens when the jackpot gets to $25 million and your net expected value for a ticket is more than its cost? Do you empty your bank account and spend it on lottery tickets because the math says you should?

The problem is that as the jackpot goes up, more people buy tickets, so the chances that you'll have to share the jackpot goes up. So it's really hard to get to that point.
posted by madcaptenor at 5:48 PM on April 25, 2019


Ticket buying syndicates exist? I guess that makes sense. You could likely make an accurate model of the extra ticket purchases based on past jackpots to work out what the profit making point would be for a given high jackpot. What are the mechanics of buying 11+ million lottery tickets anyway?
posted by any portmanteau in a storm at 7:47 PM on April 25, 2019 [1 favorite]


What are the mechanics of buying 11+ million lottery tickets anyway?

Funny that you mention that, people have been successful in buying up all the lottery numbers.
posted by mmascolino at 6:38 AM on April 26, 2019 [1 favorite]


But what happens when the jackpot gets to $25 million and your net expected value for a ticket is more than its cost? Do you empty your bank account and spend it on lottery tickets because the math says you should?

If you happen to have $11 million lying around to put on a 50-50 double-or-nothing chance, go for it!
posted by ErisLordFreedom at 1:55 PM on April 26, 2019


« Older "then surely the song would be your best friend"   |   What Has Irony Done for us Lately? Newer »


This thread has been archived and is closed to new comments