A couple days ago I put up a poll about random numbers. As a refresher, here’s the results:
Which of the below is more random?
- 00000000000000000000 (40%, 6 Votes)
- 41592653589793238462 (27%, 4 Votes)
- 69467068271807053211 (20%, 3 Votes)
- 18033988749894848204 (7%, 1 Votes)
- 19225140363320662848 (7%, 1 Votes)
Total Voters: 15
![Loading ... Loading ...](https://westergaard.eu/wp-content/plugins/wp-polls/images/loading.gif)
All are 20 digit numbers, all are equally likely if you pick a random 20-digit number. All were picked in different ways.
The winner, 0000…000 was picked as being a series of 0s. It is obviously recognizable and probably won for this reason. Also, using the typeface on my homepage, it is significantly longer than the others – maybe people selecting this need to compensate? 😉 I’m pretty sure it won either because people thought it was fun to pick the least random-looking number or because they read this Dilbert comic. On a related note, my probability theory teacher back in university enjoyed telling how 7 was the most random number – somebody had made a survey asking people for a random number, and 7 came out the winner.
The runner-up was 4159…462, or the 20 first digits of π (pi) discounting 3.1. I on purpose put it neither at the first nor last position of the poll and not in the middle. The other such option was the winner 0000…000. For a long time, this was leading, and I honestly expected this option to win, but forgot to take into account that people on my homepage have the same bad sense of humor as me, so 0000…000 was a sure winner.
The third place, 6946…211 is the only “real” random number. It is generated by random.org from atmospheric noise. My guess is it placed relatively well due to being first and looking sufficiently random. Edit: Also, it’s the only odd number.
The two last places are 1803…204 and 1922…848. The first is the golden ratio φ discounting the initial 1.6 (or 0.6) and included in case people recognized the expansion of π. In retrospect, it probably had too many 8s to be looking random (funny fact, I avoided e = 2.718281828… exactly because it looks very non-random immediately after the comma). The second is a pseudo-random sequence generated using Java’s dumbest PRNG.
I retrospect, I should probably have included a number more than once, and included a number very similar to another number as well… Maybe also numbers with patterns of increasing difficulty to recognize…
My expectation was for 4159…462 (π) to win due to its placement (neither first, last nor middle) with 6946…211 (φ) as runner-up for the same reason albeit a slightly shittier placement (neither first nor last). Then we’d have the two actually random numbers, 1922…848 and 4159…462 with 0000…000 closing the ball. I didn’t take into account that 6946…211 got a boost from being first and looking “sufficiently random,” 0000…000 getting ahead because of a maladjusted friend-crowd 😉 and 4159…462 (possibly) taking a hit from a lot of 8s.
All in all, this was a bit of random fun. At least for me.
According to https://docs.oracle.com/javase/7/docs/api/java/util/Random.html it’s using (25214903917 x + 11) mod 2^48; I generated each digit individually, so I’d guess coincidence. I didn’t seed the PRNG, so I guess it used the timer.
367509346018261343009