When a Random Poll Becomes a Monte Carlo Simulation

Description

The image displays a large, bold caption, 'I love Monte Carlo,' above a screenshot of an online poll. The poll asks, 'What day is ur birthday on' and provides three options for ranges of days: '1-10,' '11-20,' and '21-31.' The results show a remarkably even distribution of votes across the options: 1,186k for 1-10, 1,202k for 11-20, and 1,163k for 21-31, with a total of 3.551 votes indicated (note: there is a visual discrepancy between the 'k' in the options and the total, but the distribution is the key). The joke is a clever reference to the Monte Carlo method, a class of algorithms that relies on repeated random sampling to obtain numerical results. The near-uniform distribution of birthdays across a large population is a perfect real-world example of the statistical principles that underpin Monte Carlo simulations. For anyone in data science, scientific computing, or quantitative fields, the image is a humorous nod to seeing a fundamental mathematical concept emerge organically from seemingly trivial data

Common Tags

DataScience StatisticalAnalysis DataVisualization RelatableHumor

Comments

Anonymous ★ Top Pick A junior data scientist sees this and says, 'Cool, a uniform distribution.' A senior data scientist sees this, mutters 'PRNG seems solid,' and starts wondering if they can use the timestamps of the votes to estimate pi

Anonymous ★ Top Pick

A junior data scientist sees this and says, 'Cool, a uniform distribution.' A senior data scientist sees this, mutters 'PRNG seems solid,' and starts wondering if they can use the timestamps of the votes to estimate pi
Anonymous

We spent two sprints validating our PRNG; marketing’s “What day is your birthday?” poll hit χ² ≈ 0 before lunch - Monte Carlo-as-a-Service, powered by 3,551 unsuspecting users
Anonymous

Finally, a production environment where Monte Carlo actually converges to the expected distribution - turns out 3,551 random internet users are more reliable than our pseudo-random number generators, though I'm still waiting for someone to file a bug report about February having fewer days
Anonymous

When your production data accidentally validates a CS theory better than any academic paper ever could. 3,551 developers just proved that birthdays follow a uniform distribution - turns out the real Monte Carlo simulation was the friends we polled along the way. Now if only our load balancers distributed traffic this evenly
Anonymous

I love Monte Carlo too: run a poll to “simulate” birthdays, ignore that 21 - 31 has a bigger state space, and ship a date picker with a 4% bias - see you at the leap-year postmortem
Anonymous

Nothing says “data‑driven” like a Monte Carlo poll with 10/10/11‑day buckets and a locale‑broken “k” - PM calls it significant, we ask if anyone divided by bin width
Anonymous

Classic Monte Carlo: ideal uniform RNG in theory, Reddit-biased convergence to mid-month in prod
@Box_of_the_Fox 5y

I guess it's this: https://en.m.wikipedia.org/wiki/Monte_Carlo_method
@GTRst 5y

Actually it's a normal (Gauss) distribution
1. @unituni 5y
  
  Yep finally smbody wrote that
  
  @digital_insanity 5y
  
  And made an error Oopsie
2. @digital_insanity 5y
  
  Normal distribution has peak and 2 lows in both sides This is uniform distribution
  
  @unituni 5y
  
  Nope actually first and third options are those lows u are talking about so this is normal distribution it's just quite kinda low overall
  
  @GTRst 5y
  
  +1
  
  @digital_insanity 5y
  
  Concerning amplitude of each portion small variation may be statistically insignificant
  
  @unituni 5y
  
  So it as well may be a normal distribution))
  
  @digital_insanity 5y
  
  Engineers don't operate with "may be"s)) There are exact formulas how to distinguish distributions by variety of criteria And for all I've learned during metrology&statistics course that's not looking like Gauss)
  
  @unituni 5y
  
  That's not uniform as well)
  
  @digital_insanity 5y
  
  Take 100k+ survey and see yourself how BD are distributed, lol </thread>
  
  @unituni 5y
  
  And finally, as I found out from your russian texts, you lack not only knowledge in probability&statistics but also have problems with sense of humor and communication culture so yeah кек мда you are пиздец инженер))) btw we don't need 100k responses 3k is representative enough) u r welcome)
  
  @digital_insanity 5y
  
  Тоесть все свелось к: "Это ж просто мем, ты не умеешь в юмор" Апелляции к культуре общения (будто комуто не насрать на нее) и телепатия по поводу знаний в статистике "Ну такто там будет равномерное распределение, но в этом то случае мы видим же что чиселки разные на картинке!!" А, ну и да, продолжать дальше отвечать на англ человеку знающему русский, это вообще за гранью)
@unituni 5y

Anyway well we are talking about three columns so that's kind of incorrect a little bit in terms of analysis)
1. @unituni 5y
  
  .
@napLord 5y

Каким образом число родившихся людей по дням месяца у вас нормальное распределение? ебаный стыд. Может еще число родившихся людей по дню года у вас нормальное распределение?
1. @digital_insanity 5y
  
  Да смотрю тут пиздец инженеры собрались
  
  @unituni 5y
  
  /
@digital_insanity 5y

Определяют распределения по 3 точкам и с уверенностью находят нормальное, хотя тут его нет и не может быть
@digital_insanity 5y

Кек мда
@unituni 5y

Well we were discussing these three columns on the meme and not the function of human birthdays (f(date)) so u r taking it too serious)) if we would like to analyze all 365 days for sure it will not be neither uniform nor normal distribution, but closer to the second one because there is some low peaks on the end of the spring-summer, likely a composition of a few normal distributions)
@unituni 5y

Those peaks locations through the year depends on tons of different factors including possibility of unexpected ones like pandemic) And about the dates distr. on meme - it's tricky and yeah, in general it will be uniform distribution but in this particular case those three columns can form (if we can say so lol, as I mentioned before it's kinda incorrect approach) low-peaked normal distribution (we still have those 3.4% of difference, as well as they may not so we cannot tell exactly because we don't know a lot of details)
@unituni 5y

well i know google translate service)
@digital_insanity 5y

Ждем-с что будет дальше
@unituni 5y

so your only defense is that it can be statistical error? well, i`ll be waiting for your proof using those formulas you were talking about) Ive said enough and your message just confirms all that i said above)
1. @RiedleroD 5y
  
  my god, calm down
  
  @unituni 5y
  
  y u dont say same thing to the guy who thinks its cool to demonstrate his own ego by writing offensive phrases
  
  @RiedleroD 5y
  
  …I can't speak russian man. Even so, it's the internet. Don't get offended, that's just pointless.
@unituni 5y

i cant too
@unituni 5y

but still it looks like he (or she) expected that i won't find out)
1. @RiedleroD 5y
  
  find out what?
@unituni 5y

that he says about people who got in the discussion that they`re shitty engineers
1. @RiedleroD 5y
  
  so what?
@unituni 5y

well i tried to explain my point of view so that was kinda disrespectful but well at least i feel my satisfaction in the end thankfully to my knowledge not the curses)
1. @RiedleroD 5y
  
  ok so I just translated his texts with google translate, and maybe I'm missing something, but from my point of view, you're just being obnoxous
@o6m0o9m 5y

Omg, it was just a joke Still you are not right At first, it's discrete so there are no "normal" or "uniform" distributions in here At second, it's even not discrete-uniform in the limit (there are less 29-31 days in months)
1. @unituni 5y
  
  yep exactly!
  
  @RiedleroD 5y
  
  then where's the problem?
  
  @unituni 5y
  
  ''Yes, I look here asshole engineers gathered''
  
  @RiedleroD 5y
  
  translates to "yes, I look here fucking engineers gathered" on my machine but even if he called you an asshole… get over it, geeze
2. Deleted Account 5y
  
  Your first point isn't really relevant but your second one is true
  
  @o6m0o9m 5y
  
  It's matter only in a context of being "statistically correct" which was mentioned earlier
@unituni 5y

here)
@unituni 5y

it was his message
@unituni 5y

well i already got) the last ten minutes i was exaplaining to you why i answered him but u still dont see the problem
@RiedleroD 5y

no
@unituni 5y

and i will get over it too))
1. @RiedleroD 5y
  
  then don't complain please
@unituni 5y

i wasn't going but ok
1. @RiedleroD 5y
  
  you were the whole time but ok
@unituni 5y

what?
@gammapopolam 5y

привет я россия

Use J and K for navigation

When a Random Poll Becomes a Monte Carlo Simulation

Description

Categories

Common Tags

Comments

More like this

Statisticians Be Like: This May or May Not Happen

Boy Born on a Tuesday: 66.6% vs 51.8% Probability Standoff

Professor's Shirt Trolls Class: Extrapolate From Incomplete Data