Hacker News new | past | comments | ask | show | jobs | submit login
She tried 960 times (nytimes.com)
108 points by da5e on Sept 4, 2010 | hide | past | web | favorite | 51 comments

To me, this is an extremely sad story about the way we treat knowledge and learning. How absurd is it to cheer for someone that memorizes questions and answers without having contextual knowledge about the actual meaning behind those questions? It's not dedication, it's thick-headed stubbornness.

The lady has been at it for about five years now. If she did it the right way, she could have learned to learn, then learned to read and write, and then passed the exam, plus come out with other valuable skills. Doesn't this remind you of those college graduates that go through the motions, but when they sit down and are asked to do something on their own simply fail to even understand what they're asked? I remember back in college I would sometimes participate in study groups, and so often encounter students that relied simply on memorizing problem types and the ways to solve them without actually having any clue about what they were doing. It seemed more common with students hailing from asia, and surely had something to do with the way school and knowledge is treated there.

She's 60+ years old and has been selling vegetables and working on farms most of her life. She doesn't remind me of college students or graduates at all.

For her, learning how to learn is probably much more difficult than you think it is.

Well its not so much that learning to learn is more difficult, but learning to learn in a given domain. If you don't have someone to teach you, figuring out the ideal way to learn something can be tricky.

For example, the Rubiks Cube. I'm sure there are great ways to learn to solve this puzzle, but I learned using a very time consuming technique with a lot of failure. While I have a really good schema for book learning, I have a relatively poor schema for learning the optimal way to manipulate items in 3space, despite the fact that I live that space.

Learning to solve the puzzle would mean that you learned how to solve it without memorizing a technique.

There is a big difference between actually solving the cube and learning how to 'solve' it using some technique someone else came up with.

Solving it means that you analyzed the workings of the cube, figured out a way to return it to its original state without access to some outside source of information on how to do it.

Screwdrivers probably shouldn't count as a 'solution' in this sense ;)

surely had something to do with the way school and knowledge is treated there.

You mean like the fact she only started school at 15 and never finished it? We take things like literacy for granted, but they're actually quite difficult to acquire if you don't start young.

Which is not to say that I think her method was a good idea, just that there may not really have been other options.

I think I was not fully clear on who I meant treats school and knowledge. I was talking about the public, about the fact that we, and the media, exclaim and proclaim her diligence and hard working nature. That's the ironic and sad part.

Working smart is better than working hard (except, of course, when you need to work hard and smart). Imagine if instead of the current computerized mail sorting system we had thousands of post workers sitting there and sorting mail, diligently. Is that something to be applauded?

Her brute learning style reminds me of training artificial neural network from no prior knowledge: feeding problem-solution pairs to a Bayesian network again and again, hoping the ANN will reproduce the relationship somehow. We think her learning style sad and absurd, but such a statistical training technique logical and scientific. Just my two cents.

Just as ~90% of programming job applicants couldn't code fizz-buzz if their life depended on it, ~90% of drivers couldn't correct an oversteer safely... when their lives actually might depend on it.

> "Sajeonogi," or "Knocked down four times, rising up five."

Is Korean really that much more efficient than English, or is that word a label?

The instructors started to teach her after 949 tries - 949 tries, of watching this poor woman fail! - then she did learn, and got it on the 11th try after that. Teaching her was frustrating to them, I think mainly because they weren't really teaching her about driving, but a subset of civil administration and technology concepts, such as "regulations" and "emergency light".

The tragedy is that she could not afford "Middle school", despite dreaming of it so painfully that taking the driving test daily became a joyous wish fulfillment of attending school... an attitude of which the stereotypical student is too invisibly wealthy to properly conceive.

> > "Sajeonogi," or "Knocked down four times, rising up five."

> Is Korean really that much more efficient than English, or is that word a label?

No, it's a 사자성어 (sajasongoh), or 4 character idiom. They originally come from Chinese, in which they're called chéngyǔ (成语 or 成語, in simplified and traditional characters, respectively), meaning "set phrase", and also exist in Japanese, in which they're called 四字熟語 (yojijukugo, literally "four character Chinese idiom").

Because of their origin in classical Chinese and their brevity, they're impossible to understand unless they're explained to you. In Asian countries, memorizing/understanding these proverbs is a big part of schooling.

I don't know the details of this particular sajasongoh, but I'm guessing that it either comes from classical Chinese or is a modified version of one that does. Modifying these proverbs to fit a particular situation is a common form of wordplay in Chinese, Japanese, and Korean. For example, the Chosun Ilbo (a big Korean newspaper) loves doing this in their headlines.

so it's like Brangelina

More like acronyms, I'd say. For example, "HTTP" = hyper-text transfer protocol.

It's funny that the word 사자성어 (meaning: 4-character idiom) is itself a 4-character idiom. East Asians like 4-character groups so much, they even read numbers in groups of 4 digits. For example, 1,234,567,890 would be read 12 억 3456 만 7890.

I doubt they like it. 4 is considered unlucky in East Asia. Elevators always read 1->2->3->5->...


Although 4 is considered unlucky in Chinese, decimal numbers is represented in 4-number groups, instead of 60 thousand, we'd have 6 wan, 800 million would be 8 yi, etc

It's probably a webdesigner-thing to think of 960 as a nice round number, but that's the first thing I thought upon reading that.

I confess: I thought "960 Times" was a newspaper fixed width stylesheet template of some sort at first.

Haha, I was thinking exactly the same thing :) Professional handicap I guess.

It's a relatively common number in computing. It's similar to how numbers such as 512, 256 and 1024 are intimately familiar to me.

It's not a power of 2 (although it is 512 + 256 + 128 + 64). Chess960 is the most popular variant of Chess, having 960 possible board layouts. There is the famous CSS framework, 960 Grid System. And it happens to be in the family of common display dimension, being twice the height of standard definition, 480, like the new iPhone and iPod, as well as half the width of an HD television (1080), 1920.

> It's not a power of 2 (although it is 512 + 256 + 128 + 64).

Well, if that counts as "round" then I have to remind you that any integer can be written as a unique sum of various powers of two (or ten, or eight, or ...).

That said, being able to quickly convert an arbitrary number into that form has proven useful more than a few times. But I admit that I never realized that playing a very old, free game called "Binary Blitz" would turn out to be quite so useful to me.

Yeah but it MUCH rarer that an integer can be written in consecutive powers of 2.

Define "rarer"? There's an infinite number of integers that can be written in consecutive powers of 2. So in some sense, there's exactly as many of those as there are of integers!

For fun, see if you can write a bijection between the integers and the "consecutive ones". Hint: realize that a "consecutive power of 2" number can be split into two numbers: a number of consecutive ones greater than zero followed by a number of consecutive zeroes (you can have zero of these), then think about diagonalization.

That said, I suppose you could define some sort of "density" of numbers with consecutive ones in a fixed-size range. Say that you want to know how many of them are between 0 and 2^x (i.e. how many such numbers are x bits long). Well, you get sum(1..x-1) different consecutive numbers (for 2^8, there are 7 ways to have a pair of ones, 6 ways to have 3 consecutive ones, as you can see by imagining sliding the pair or triplet: 11100000, 01110000, 00111000, ... 00000111) and there are 2^x possibilities total.

Add x-1 + x-2 + ... + 2 + 1 to itself backwards and you get: (x-1 + 1) + (x-2 + 2) + ... where there are x-1 terms in the series, which allows us to rewrite it as x * (x-1). This is double the original sum (because we added it to itself), so divide it by two and we've shown that sum(1..x-1) == (x^2 - x)/2. Now, divide that by 2^x and simplify to get D(x) = (x^2 - x)/2^(x+1) for the fraction of x-bit numbers that are consecutive powers of two.

It doesn't take much analysis to see that it is decreasing when x grows after increasing initially:

Density of consecutive ones in 1-bit numbers = 0 Density of consecutive ones in 2-bit numbers = 0.25 Density of consecutive ones in 3-bit numbers = 0.375 Density of consecutive ones in 4-bit numbers = 0.375 Density of consecutive ones in 5-bit numbers = 0.3125 Density of consecutive ones in 6-bit numbers = 0.234375 Density of consecutive ones in 7-bit numbers = 0.1640625 Density of consecutive ones in 8-bit numbers = 0.109375

In short, "much rarer" depends on how big a number we're dealing with and the measure won't work for infinitely large numbers, because there are a countably infinite number of consecutive-power numbers, so they can be put into 1-to-1 correspondence with the integers, even though our D(x) decreases.

TL;DR: Math is crazy and relies on precise definitions. Be wary of intuition.

Similar thing happens all the time with powers of 2...

This is quite possibly the most inspiring story of human tenacity and sheer grit I've heard in quite a while. It really made me smile.

Clearly her method was not optimal but it... eventually worked for her. Just goes to show that perseverance pays off even when starting from absolute zero.

Good for Cha Sa-soon!

And she got a free car out of it:

In early August, Hyundai presented Ms. Cha with a $16,800 car.

Ms. Cha, whose name, coincidentally enough, is Korean for “vehicle,” now also appears on a prime-time television commercial for Hyundai.

Go Grandma!

She tried the WRITTEN test 960 times.

She passed the driving test in 4 tries which is probably like the average american teenager these days.

At $5 a pop though, she could have hired some personal tutors.

Either this is a common occurrence in South Korea or the media keeps recycling this story, I first read about this on Sept 6, 2009 and then again on May 7, 2010.


Perhaps she didn't realize that she could study for the test and then pass it after one try. According to the article she has almost no formal education, so that seems possible: "It was not until she turned 15 that she joined a formal school as a fourth grader. But her schooling ended there a few years later."

Bingo. She didn't have the formal education, and was basically rote-memorizing everying:

"What she was essentially doing while studying alone was memorizing as many questions — with their answers — as possible without always knowing what they were all about."

It's surprising that she didn't end up remembering by heart all possible test answers by then. It's also sad because, if she had asked an instructor politely to explain one complex traffic word for her each day, she would have reached the understanding she needed to pass the test far sooner. Then again, it sounds like she treated it like a hobby. If she was having fun, who's to stop her?

I spoke too soon. On the second page, the article says she did study, but did so "phonetically" and thus didn't understand what she was studying.

This reminds me of something: I saw a documentary about championship Scrabble players and a lot of them aren't English speakers or even Latin alphabet language speakers. They treat Scrabble as a visual pattern game rather than a verbal one.

Pretty much been my experience playing. It's more about memorizing lists of words without really bothering to know what they mean and figuring out which words to use when and how to place them to maximize your point total on each play.

Family scrabble rules: if you don't know a definition for a word (which can be found in the dictionary if challenged) you can't use the word.

I mean no offense by this but.. is English your mother tongue?

I used to play Scrabble a ton and adopted a more verbal, etymological style that, I'd assumed, would be more common amongst native English speakers (but maybe not!)

It is, but my primary strategy has been to commit a huge list of words to memory, especially the two-letter words to join new words together for point efficiency.

To me, this is less about the fact a single woman had to memorise a written driving test by failing it 949 times, and more about the mismatch between the test itself and the populace.

Surely she is fairly indicative of a rural inhabitant? Why is the test so inaccessible, and how many others are there like her?

>Why is the test so inaccessible, and how many others are there like her?

The test should ensure that a driver can read read and understand street signs both symbolic and textual ones and respond correctly ... oh yeah, and drive the relevant vehicle. They should also be able to service the car sufficiently to keep it safe - pump tyres, check fluids, know mostly when it needs attention from a mechanic.

Not everyone can do all these things hence the test appears "inaccessible".

It would be pretty darned difficult to make a standardized test "match" folks who are illiterate or just barely literate, which this Granny seems to have been. BTW, literacy rate in South Korea is around 98% (most of the remainder being old women like Ms. Cha who couldn't get education during colonial times) so I don't think the test would have been too inaccessible to the populace in general.

“She could read and write words phonetically but she could not understand most of the terminology, such as ‘regulations’ and ‘emergency light,’ ” said Ms. Park, the teacher.

I see three possibilities. If these words are as normal in Korean as they are in English, then she failed to learn them because (possibility 1) she was so sheltered she actually had no exposure to basic knowledge required of any competent driver or (possibility 2) she has an intellectual disability that prevents her from learning new words and concepts without a lot of help. In these two cases, you can't blame the test.

Possibility three is that the words for used on the test are government legalese that is never used in real life, and she did not understand that learning what they meant was a better strategy than trying to memorize entire sentences from the driving manual. In this case the test is partly at fault, but an intellectual disability would probably still be involved. (Though perhaps you could also blame educational traditions that stress rote learning and glorify hard work as a sufficient solution for any problem.)

I guess all three of these scenarios require her to have some kind of intellectual disability, because even in the first scenario, a normal person could learn everything she needed from the test-prep books. Also, I doubt anyone would celebrate her persistence if it was a story of a person with normal intelligence who threw away so much time and money retaking a test she should have known she wasn't prepared to take again. It would be perverse or at best eccentric, not inspiring.

960 - 1 = ?

I wonder what are the odds of passing the written test with random answers. I assume it's multiple choice.

Here's what we know from the article: (1) 40 question test, (2) Multiple choice, (3) She passed it with "60 out of 100"

My assumptions: (1) It's graded on a simple "percentage right" basis, so 24/40 questions necessary right to pass, (2) 4 options for each question

Her score on an individual test is a random variable X following a binomial distribution with 40 trials and chance of 0.25 for each trial. Her chance of passing by guessing randomly, P(X >= 24), is an infinitesimally small 2.826E-6.

The probability that she fails all of the 960 tests, assuming independence of tests, is (1-p)^960. So the probability that she will pass at least one test is:

  1 - (1 - p)^960
Plugging in p = 2.826E-6, the chance is still practically 0, so a naive guessing strategy would not work.

However, under the above assumptions, she could practically guarantee her success by combining this guessing strategy with a simple test-taking strategy like eliminating 1 or 2 obviously wrong answers per question, or just remembering the answers to several of the same questions that are probably being recycled from test to test.

just remembering the answers to several of the same questions that are probably being recycled from test to test.

I assumed this is how she did it, but assuming all questions are recycled, 4 options in 40 questions, perfect memory, no knowledge, and a naïve strategy, she should have passed in just 29 tries (1.75 retries per question, 10 initially right, 24 needed). Mastermind isn't that hard a game. So clearly, some assumptions are wrong.

Change your assumptions to assume a larger set of questions, not 40.

The article says that her scores slowly crept upwards, not indicative of answering randomly.

The test givers must be quiet stubborn to, can't imagine something like that ever happening here for some kind of similar test, she would just get passed or turned away in the first 10.

Interesting question is that now that she got 60/100 does she understand things much better than 3 years ago, some form of understanding would spike the scores quickly, while the slow improvement can only seem to be the result of slightly better memorisation of question->answer each time without much better understanding.

Off-Topic: Most stories... they'd cover on TV news [or newspapers], [are] probably off-topic. A story that man bites dog I think would be in this category.

On-Topic: Anything that good hackers would find interesting. That includes... anything that gratifies one's intellectual curiosity.

That she took the test 960 times is interesting to know, but hardly gratifying.

Sadly, all she got was a drivers license. It's not like she hit success with her startup after 960 tries and became a happy member of the fym club.

She was a poor rural villager. She got the right to drive and a free $16,000 car. Its like getting magic powers and a king's ransom. I'd wager her life changed more for the better as a result of these events than most of ours would if we did get that big fym exit.

In her world, she joined indeed.

Well, after deducting 960 * $5 fee ~= $5k, plus 960 * unknown bus fee (let's assume it's rounds to $1k), it's more like a $10k car. Also, this car is given to her, so her value for the car may not be $16k (imagine she's given $16k towards the purchase of any car, she'll probably ended up with a different car). Finally, not sure of tax laws in Korean, so she may have to pay taxes on it before the $6k fee deductions. She still comes out ahead, but probably close to half of $16k value.

But her fame from the story is priceless.

Success is defined by the person.

Forgive me for being blunt, but this is an incredibly egotistical and snobbish statement to make.

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact