A bad cup of tea and the birth of modern statistics | Hacker News Reader

A bad cup of tea and the birth of modern statistics

sciencehistory.org

74 pointsapophasis19 comments

When I was a Google SRE, during one trip to Zurich I complained that I hated the lack of my standard vice, Diet Coke, in Europe (they have "Coke Light" instead). My boss didn't believe me that I could tell the difference... so he brought some back to the States and we ran a blind test. (I think we also added Pepsi and some other things.) My recollection is that I got every sample correct, but we definitely didn't have that many! I wonder what the probability of a false positive was in that case. (I no longer drink Diet Coke, but I still maintain that Coke Light -- and Coke Zero which I believe to be the same formula -- is gross.)

> I no longer drink Diet Coke, but I still maintain that Coke Light -- and Coke Zero which I believe to be the same formula -- is gross.

Here, we have Coke Light, and Coke No Sugar (formerly known as Zero). Coke Light probably tastes similar to Diet Coke, while Coke No Sugar has sucralose instead of aspartame to make it taste closer to regular sugar/HFCS Coke.

Coke No Sugar and Coke Zero are different formulas, at least in Australia. It was a big enough deal that major supermarkets refused to carry Coke No Sugar when it launched here next to Zero [1], and Dominos Pizza switched their contract from Coke to Pepsi over it. [2]

Personally I was a fan & regular drinker of Coke Zero. But I tried Coke No Sugar & didn't like it as much, and I switched all my spending to Pepsi Max once Coke Zero was discontinued. Interestingly, in Australia, Pepsi Max is owned by Asahi and is the #1 diet cola, even above the Coca-Cola brands.

[1] https://www.news.com.au/finance/business/retail/coke-no-suga...

[2] https://thewest.com.au/business/retail/woolworths-cans-new-c...

It seems “No Sugar” (which I just heard of for the first time) is supposed to replace Zero but it’s not quite the same. (Anyway, even the “same” type of Coke is not the same across markets and if I remember correctly the Zero brand has been used for quite different products).

https://www.news.com.au/finance/business/retail/coke-no-suga...

Edit: in which market do you say sucralose being used in “no sugar” coke? I can’t find any reference other than to “diet coke with splenda” (a brand of sucralose, which is used in combination with aspartame in that “diet” coke).

> in which market do you say sucralose being used in “no sugar” coke?

That would be Singapore. Here's a hotlink to an ingredient label images from one of our online supermarkets:

http://laz-img-sg.alicdn.com/p/54a80f275b284804aa4000e9fa899...

(original page: https://www.lazada.sg/products/coca-cola-zero-sugar-i3011107...)

I used to remember aspartame being present in Coke Zero, so Coke No Sugar (which seems to be marketed interchangeably as Coke Zero Sugar in this case) could actually be a new formula.

Interesting, thanks.

The zero brand seems to have been retired, and both "no sugar" and "light" use sucralose + acesulfame-K as sweetener (that's also the case for the US "diet coke with splenda” I mentioned, I mistakenly said it contains aspartame).

https://www.coca-cola.com.sg/brands/coca-cola/coca-cola-no-s...

It seems, however, that in other countries both "no sugar" and "diet" contain aspartame + acesulfame-K (and cyclamate in some cases, but it's forbidden in some countries).

>My boss didn't believe me that I could tell the difference... so he brought some back to the States and we ran a blind test.

Odd he didn't believe you, given the difference between the two is so striking.

Thought Diet Coke gave you Brain cancer and stopped. Switch to Coke Zero.

I think the latest white can coke good. Got some fibre inside and keep you slim :-))))

Reading the title of this article, I immediately thought of William S. Gosset, "The Guinness Brewer Who Revolutionized Statistics", earlier in the century. I am sure there are even older examples.

https://priceonomics.com/the-guinness-brewer-who-revolutioni...

When I visited the Guinness Brewery in Dublin a few years ago, I was pleased to note that they have a plaque honoring Gosset/"Student".

If I ever get to Dublin I want to see Hamilton's quaternion plaque on Brougham Bridge.

No reference to the " The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century"?

this is the book that introduced me to this origin story! It's a very well written, and short, book. Highly recommend.

Just to work out the odds, it's not 2^-8 since she knew there were four of each type of tea. 1 in 70 = 8 choose 4, which one can mental math as 8!/(4!4!) = 8765/4! = (8/(42))7(6/3)5*1 = 70.

My asterisks got formatted :( still works though

According to the story she chose after each cup, not at the end (so it's harder for her to make use of the 4-4 information); therefore, the odds calculation is not so straightforward right?

I was thinking the same thing! I think those would be her odds if she were permitted to go back and change her mind about which cups of tea had milk first. If she said the first 4 cups were milk first with some confidence, but on the fifth cup was absolutely certain it was milk first, she would have to go back and change a previous evaluation (because she knows there can be only 4). But because they were presented one at a time, and Bristol made her determination up front. It's basically a series of independent Bernoulli trials (i.e. Binomial distribution). So, assuming she had a 50% chance of guessing correctly, the probability would be (8 choose 4) * (1/2)^8.

But would this provide good or bad input noise for a atomic vector plotter? One might get very low levels of improbability...

Reply on news.ycombinator.com