TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Pick a number from 1 to 10

218 pointsby alecperkinsabout 14 years ago

50 comments

alecperkinsabout 14 years ago
Pick a number, then read, please.<p>The goal at first was to just see what numbers people gravitate toward. I've heard lots of conjecture about how people pick 7 or 3 or 4 more than others, and for a variety of reasons, but had a hard time finding actual demonstration of this. Then, while implementing a choosing system, the problem became: how do you present the information so as not to bias it? This is why there are four different ways of picking. There are also a couple other metrics being measured, including a difference in phrasing (Pick a number… vs Pick a random number…) which may be interesting.<p>Apologies for any bugs or general wonkiness. The whole thing was a ~2 hour impulse project.<p>PS: The data will absolutely be shared! Just need time to do a breakdown of all the different permutations.
评论 #2375634 未加载
评论 #2375765 未加载
评论 #2376346 未加载
评论 #2375537 未加载
评论 #2375871 未加载
评论 #2375950 未加载
评论 #2375619 未加载
评论 #2375464 未加载
评论 #2375795 未加载
评论 #2375894 未加载
评论 #2376073 未加载
评论 #2376391 未加载
oniTonyabout 14 years ago
I was going to pick 4, but then realized that 4 has been my default "random" number, ever since <a href="http://xkcd.com/221/" rel="nofollow">http://xkcd.com/221/</a>
评论 #2375770 未加载
评论 #2375914 未加载
评论 #2375766 未加载
terhechteabout 14 years ago
I also chose 7. Human Random Distribution is probably not very random. I hope that he'll release the data.
评论 #2375549 未加载
评论 #2376522 未加载
评论 #2375873 未加载
teuobkabout 14 years ago
An interesting extension of this could be to add a poll here on HN asking, "Which number did you pick?" and then comparing the poll results to the actual results.
otabout 14 years ago
Are you trying to estimate the bias depending on the type of selector used (slider, input, options, ...)?<p>Very interesting concept :)<p>EDIT, forgot to add: if this is the case, have you thought of storing the type of selector in a cookie, so that refreshing the page gives always the same type?
评论 #2375411 未加载
评论 #2375390 未加载
Bolyubaabout 14 years ago
Did it twice before reading comments. 7 and than 3. Have no idea why. My wife just picked 7... Scary
steipeteabout 14 years ago
There are faster ways to provide data for /dev/random...
评论 #2375354 未加载
necroabout 14 years ago
Based on that interface I would predict...<p>middle number: for the graphical choices as it's common/easier to get a valid first click in the middle of the slider.<p>high number: for the key entry as if most people are right handed it's easier to get to the high numbers instead of going across the keyboard.<p>all that is predicated on the fact that people are lazy instead of random when using this.
评论 #2375621 未加载
joejohnsonabout 14 years ago
It might be biasing it's results by having it set to one by default...
评论 #2375444 未加载
rgbrgbabout 14 years ago
Excellent. Now give us the data!
niketdesaiabout 14 years ago
I am interested in finding out how many people decided not to choose a number at all, and if that fact could help mitigate the experiment's inherent bias.<p>As for people making multiple entries into the experiment, a simple IP filter could help reduce overall error.<p>I think it would be neat to offer an option to not choose a number, but rather a number to help offset selector's bias (in participating in the number choosing experiment knowingly). It's similar to the multiple choosing UIs.<p>Nonetheless, I can only appreciate simple things like this that lead to a spirited and educated read.<p>Cheers and can't wait to see "final" results and accompanying analysis.
clvvabout 14 years ago
Interesting project. I'm looking forward to see the results. This reminds me some of the mental algorithms I was thinking about: How do you generate a random number without electronic devices? How do you generate a random permutation without electronic devices? How do you generate a hash without electronic devices? How do you do the above most efficiently? How about without any pen, pencil or paper?<p>It will be cool if you can master mental cryptography.
alecperkinsabout 14 years ago
I posted a graph with the basic distribution on the results page (just hardcoded for the moment):<p><a href="http://nfrom1to10.appspot.com/results/" rel="nofollow">http://nfrom1to10.appspot.com/results/</a><p>A more in depth breakdown is coming. The number of responses has been far, far beyond what I expected — several orders of magnitude more. Once I get together a good way to efficiently track and display the graphs, I'll make them live.
roryokaneabout 14 years ago
Upon being shown a circle and asked to pick a random number, I thought for a bit and realized that no matter what I picked I wouldn't believe I had chosen it randomly. Then I wrote a one-line script to generate a random number from 1 to 10 and I put its output as my answer. Perhaps this defeated the purpose of the site, but I find it so hard to choose when I know my response is being measured.
Tichyabout 14 years ago
Can't wait to hear how many people picked 10 (as I did). If it hadn't been for science, I would probably have picked 7 like everybody else.<p>Once I thought it would be good to play the numbers "1 2 3 4 5 6" in the lottery because they are as likely as other numbers, and I thought people would not pick them. Turns out lots of people play "1 2 3 4 5 6".
评论 #2376444 未加载
Natsuabout 14 years ago
I was going to pick 7.1 until I found out that he was prepared for pedantry and restricted the set to natural numbers.
ilitiritabout 14 years ago
Here's another test: Ask the user to pick a number from 1 to 10 but randomize the order of the numbers each time.
davidjagoeabout 14 years ago
Brilliant idea, but haven't you built in bias by unblinding the experiment? I followed the link, got one implementation and keyed in '5'. Then I read your write-up and realised that there are more implementations, reloaded the page and thought deeply about the number I would/should choose on the slider.
coprologabout 14 years ago
yesterday i was improving my excel skills about statistics, instead of use the random funcion i thought to ask a number from 1 to 10 to some friend and see the histogram. just after some question i was surprised about the results and so i asked more people. up to now i asked 24 people(46 number in total,the first 2 was just one question), i also asked a second number after recived the first. my result are: that 7 win with 31% of answer followed by 5 with 17% the last are 10 (zero answer), 1 with 1 answer and 2 and 6 with 2 answer each. you have to note that 83% say a odd number as a first one.noone told me a number that wasnt integer. whats going on? the majority of people was italian, some spanish and some german.Guido
solipsistabout 14 years ago
After you've entered your unbiased data, make sure to refresh the page and see what else you could have been confronted with. You'll notice the different variables in the experiment - just make sure to do this <i>after</i> in order for it to not influence your decision.
mcorrientesabout 14 years ago
Even if people have to pick a number from 1 to 20, 7 would be the second most chosen one.<p><a href="http://scienceblogs.com/cognitivedaily/2007/02/is_17_the_most_random_number.php" rel="nofollow">http://scienceblogs.com/cognitivedaily/2007/02/is_17_the_mos...</a>
robryanabout 14 years ago
I'd imagine 7 would be less popular or people that have previously heard of a bias towards 7, would be hard to have a test group though that has never heard anything at all about a bias. It's still the highest so far in this experiment though.
protomythabout 14 years ago
I would imagine that if you did this survey in a crowd of sports fans, you would get a different distribution based on town or player followed. Plus, if I remember right 7 and 4 and not terribly lucky numbers in China.
lucasjungabout 14 years ago
I picked 6 because it was closest to my cursor when the row of numbers popped up.
reidababout 14 years ago
I'm curious to know if the typing prompt scores fewer 10's because of increased typing effort and if the hover-to-reveal circle gets more centrally-located numbers due to Fitt's law
Vivtekabout 14 years ago
I swiped the mouse quickly, ended up at 5, and managed to resist the temptation to change it to something "more random".<p>How many people leave it at 1? This is kinda neat.<p>Also, it was nice of science to thank me.
评论 #2375456 未加载
评论 #2375434 未加载
frooxieabout 14 years ago
I picked 7, because that's what I expected other people to gravitate to, and I wanted to see if you had some fun response if I picked the most common alternative. :P
tommiabout 14 years ago
I chose 7 with slider preset at 1. Didn't want to leave it at 1 and throwing the slider "left" it there.<p>Referrer might be interesting information. Do geeks behave differently etc.
BasDirksabout 14 years ago
I picked 8 because I could not find a reason for picking 8.
Andrenidabout 14 years ago
I picked 6 purely because it's in the middle of my keyboard. I don't think "type a random number" is the same as "think of or say a random number"...
WingForwardabout 14 years ago
What is the point of the asterisk at the end of the request?<p>Oh, and something about the page with the circle on it prevents my browser from reloading.
CallMeVabout 14 years ago
I accidentally chose my number before I could do more than deliberate on whether to pick that number or go for another one.
mailarchisabout 14 years ago
There is an interesting observation. Try selecting a random number multiple times with page refresh. The ui keeps changing.
d2zoabout 14 years ago
Hmm, might be interesting to see the results of numbers presented in a ring, equidistant from the cursor...
评论 #2376467 未加载
driverdanabout 14 years ago
I was already aware that most people choose 7 in this situation so I selected my favorite number 8.
评论 #2375980 未加载
noseabout 14 years ago
I'd like to see a heatmap for the website. I think you can do it with google analytics.
bauchidgwabout 14 years ago
pleas deploy <a href="http://code.google.com/p/jquery-ui-for-ipad-and-iphone/" rel="nofollow">http://code.google.com/p/jquery-ui-for-ipad-and-iphone/</a> or something likt that to make the number by bar-thingie working in the ipad
评论 #2376210 未加载
tejaswiyabout 14 years ago
Actually, I've heard a lot of people pick 2 or 7. Can you confirm if this is true?
评论 #2375345 未加载
评论 #2375358 未加载
评论 #2375363 未加载
hammockabout 14 years ago
Is this an actual app or is it just a trojan horse designed to collect our data?!
评论 #2375500 未加载
评论 #2375448 未加载
eliabout 14 years ago
You shouldn't have it start on "1" -- you should make people have to slide it.
nikcubabout 14 years ago
Thought of submitting this to mechanical turk in order to get a lot of data?
评论 #2377269 未加载
kirpekarabout 14 years ago
What is the point of this numerical experiment? (I"m serious)
AbyCodesabout 14 years ago
Without something like facebook or twitter login, the system will be abused. There is no stopping someone who just reloads ( or even make an automated script heh ) and keeps on choosing, resulting in flawed data. Just saying.
评论 #2375580 未加载
samuel1604about 14 years ago
It's all about the position of the mouse I would believe.
评论 #2375675 未加载
Zolomonabout 14 years ago
I had a favourite number, and I picked it. It was 4.
zyfoabout 14 years ago
Reminds me of Benford's Law [1], which stipulates that in many lists of numbers from real-world data, the leading digit is 1 30% of the time, and larger digits occur as the leading digit with lower and lower frequency (logarithmically). This has been used to detect made-up numbers in accounting which later turned out to be fraudulent behaviour.<p>1: <a href="http://www.rexswain.com/benford.html" rel="nofollow">http://www.rexswain.com/benford.html</a>
ruby_on_railsabout 14 years ago
Its a cool idea but I would recommend A/B testing different GUI for picking numbers for removing some of the possible biases. I personally picked 7, most likely because the human eye moves from top left to bottom right so I scanned what the choices were then picked the number near where my eye last looked.<p>Also, I would love to see a heat map/click map of that page.
TheSOB88about 14 years ago
I did not understand what to do. I thought it was going to try to guess it. I didn't understand that the 1 was the answer I was giving. Sadly.
J3L2404about 14 years ago
Of the different interfaces, the prompt in the circle looked the best, although the mysterious black circle was OK.
insightabout 14 years ago
Pls add : 42