Outsmart your brain by knowing when you are wrong | Hacker News Reader

Outsmart your brain by knowing when you are wrong

troysimpson.co

47 pointsmrtron40 comments

For my 6th birthday party, my mother challenged all the kids to guess the number of Smarties (M&M-like candy) in a tube. Everyone guessed in the range of 15-20, except for one joker who yelled out 100. It ended up being 121.

This makes me think that underestimation is more common for the things you want lots of (chocolate, tickets); with overestimation being the norm for the things you want less of (enemy soldiers, incoming vehicle speed).

This would make sense from a what-happens-if-you're-wrong perspective.

Interesting. I did not know that Smarties were Europe-only.

The US has a candy called Smarties, but they are different entirely, and made by different companies.

Smarties in US: http://2.bp.blogspot.com/_Qm2M81VKS-s/SaC6lVU2-UI/AAAAAAAADn...

Smarties in Europe: http://www.sogoodblog.com/wp-content/uploads/2008/11/smartie...

Smarties exist in Canada as well. Though every seems to enjoy M&Ms more.

If you're applying a kind of maximin strategy, overestimating the things you want less of makes sense.

But underestimating the things you want more of is as likely to lead you to quickly exhaust your existing supply as to go out and find more.

Everyone guessed in the range of 15-20

Were the guesses public? Perhaps it was just the result of anchoring. Or overestimating others' abilities in comparison to one's own abilities (is there a term for this?).

This makes me think that underestimation is more common...

This one experience? And you were only six... not a whole lot of life experience at estimating things.

Don't try to count or guess, use simple math shortcuts.

I took a rough average of the double digit numbers on the screen (35), multiplied by the number of double digit numbers (16), and since 30 x 10 is 300, I guessed the total would be around 500.

This was all done within 5 seconds

I simply noticed that the bottom two rows had numbers clustered around 30. So I did 30 * 20 = 600. Not as close as your guess, but close enough :)

Yeah, same here. I noticed a few spots were blank, so I'd say around 540, but yeah, close enough.

Same here... I counted 14 boxes with numbers around 30, so that was 420. I rounded up to 500 because I knew there were skipped boxes. All under 5 seconds.

I think you are missing the point. No simple math tricks can overcome the fact that I am impressed at the amount on my credit card bill at the end of the month.

What the author tried to convey is more the incapacity of the brain to apprehend a large amount of small numbers.

Simple math will tell you to estimate your credit card bill using old credit card bills as a reference, not your gut feeling about how much you spend.

Slightly related to the post: If you are in Kuala Lumpur, don't bother going up the Petronas towers, you don't get up that high, the wait is long, the view isn't spectacular, and the infomercials about the worlds nicest little petroleum company gets pretty grating.

Don't miss the KL Tower though. It costs to go up (RM12, pocket change!), but the view is spectacular and you can stay as long as you like. The rotating restaurant is also very nice, but remember to book in advance.

I can confirm the infomercial before you go up the elevator is VERY grating.

However the whole day was amazing for me resulting from the aquarium visit right nearby the towers - the entrance is right where the picture of the towers is taken from.

Here is the pictures from it, truly amazing. It has an incredible 100 meter (!) underwater tunnel through a massive aquarium with a moving sidewalk.

http://taiwan.teebyte.com/malaysia-day-7-kl-aquarium-visit

On this topic- any recommendations good books/articles on Maths/statistics, or how the mind does estimation, and how to improve the same?

Back when I was in high school, I had read this book "How to develop a super power memory", by Harry Lorayne. I daresay, it was not a complete success for me :)

But you could pull-off some pretty neat memory tricks, like remembering lists of items, phone numbers. The reason I quote this-the author's reasoning was that the "memory" of human mind works best when it is irrational. For instance,if you have a list of unconnected items like carpet,guitar,laptop- make an illogical association of one item to the next.For instance, in your mind's eye,"see" something ridiculous like you strumming a carpet, and so on. This actually did work for me, back then.

So, has anyone tried out/read something similar for improving number estimation?

I'd vote for this: http://www.amazon.com/Secrets-Mental-Math-Mathemagicians-Cal... and this: http://en.wikibooks.org/wiki/Mental_Math

Nice links! Hadn't seen wikibooks before..

I wildly overestimated (700) due to the emphasis you placed on telling me how much I was likely to underestimate!

Maybe move that section to after the part where you ask the readers to guess?

My favorite example of something people are terrible at estimating is the number of leaves in a tree.

It's typically (rot13'd) bire n uhaqerq gubhfnaq.

... why was this rot13'd?

But cool. I could see that. Lots of leaves in trees...

I would be interested to see what people guess the total number is - post your estimate!

I guessed between 500 and 700 because there are 4x10 blocks (40) with a range between 0 and 40 tickets per block (time slot). But in fact the top two rows are almost empty, and the bottom two have many tickets left.

That makes:

Max = 40 slots x 40 tickets div 2 rows = 1600 div 2 = 800

Min = 40 slots x 10 (average number of tickets) = 400

Probable range: somewhere inside 400-800 so lets say 500-700! (that is not a "strong" reasoning, that is just what I thought).

Interesting - perhaps if you go the opposite way about how many tickets are already taken it will result in you overestimating the results.

I guessed 400. My brain initially gave me the number 250, but then I realised the numbers in the lower rows would add up to 100's pretty quickly, so I upped it a bit.

I guessed 300.

I guessed 600.

I guessed about 700, so I was way off, but on the other side. My reasoning was: the morning was pretty booked out, but most of afternoon slots (24 of them) had about 20 - 30 free. 24 * 30 = 720.

500, after being primed by estimates in the article

Two rows of ten boxes at the bottom have nontrivial amounts in them, a quick eyeballing gives 30 as a possible average amount -> 600 total.

530, really really close :o

500, based on eyeballing the number of big numbers (between 15 and 20, I took 20 to adjust for the small numbers) and eyeballing the average value of the big numbers (25, I took a number slightly lower than I estimated because I knew that the 20 was probably too high). Giving the estimate 20*25=500.

500. Thought process was something like "there are a lot of 30s in the list, at least 3 groups of 3, so a really conservative estimate would be 300. But there are about that many boxes again having numbers in the double digits below 30. So, uh, let's go with 500."

500, I remember coming up with an "average nonzero cell" estimate and then multiplying it by my estimate of how many of these there were. All done in under 5 seconds, a side effect of which is that I don't really remember what process the first estimate came from.

i surveyed the numbers, noticing bunches of 30-40 and few smaller numbers. so i got to roughly summing them up into hundreds.

in the end, i got to around 500. the bottom row alone gets you around 250 and the one above that another 200-250. the rest is peanuts.

i'm sure i took longer than 5 seconds for that, but 5 seconds is time i'd spend gladly on a good estimate, especially while standing in line.

500. But it was influenced by his warning not to underestimate.

Scans screen. 200. No, wait, that can't be right. Spots several two digit numbers which alone would add to two hundred. Looks more like . . . using previous sum as a "what 200 looks like" baseline. . . I'd say 500.

ha. what I've learned from reading the comments is that it's not necessarily necessary to compensate for this error. (I also guessed 400 as a lower bound.)

Most likely everyone commenting was compensating for error, as the author primes you by telling you about the error before asking you to guess. If he had written that people tend to grossly overestimate, I bet you'd have a very different result.

A/B test!

Reply on news.ycombinator.com