1210: I'm So Random

Explain xkcd: It's 'cause you're dumb.
Jump to: navigation, search
I'm So Random
In retrospect, it's weird that as a kid I thought completely random outbursts made me seem interesting, given that from an information theory point of view, lexical white noise is just about the opposite of interesting by definition.
Title text: In retrospect, it's weird that as a kid I thought completely random outbursts made me seem interesting, given that from an information theory point of view, lexical white noise is just about the opposite of interesting by definition.

[edit] Explanation

Hairy walks up to Black Hat, utters a nonsense phrase and them proclaims that he is "so random". This is a reference to a modern phenomenon in which people make "random" statements, and somehow imagine themselves to be funny and interesting because of this. Black Hat, never one to hesitate over bringing someone down hard, replies that he is also random, proving this by pouring forth a torrential stream of numbers that nearly overcomes poor Hairy. Black Hat then resumes his posture at the computer, as if nothing has happened.

It is true that when brilliant and creative people speak passionately about a subject, they can make mental leaps and changes of context that might seem bewildering to an outsider. The conversation may even seem to be "random". However, simply vocalizing nonsense is not analogous, or even desirable; it is more likely a character trait of someone who has difficulty in following or adding to a normal human conversation.

Black Hat's "random" numbers are actually quoted from the first lines of A Million Random Digits with 100,000 Normal Deviates making it both "officially random", but also essentially not. (See also: 221: Random Number.)

A side note is that "Monkey tacos" is a phrase which contains two trochees. A trochee is a metric foot with one stressed beat and one unstressed beat; it may be a reference to or an unconscious allusion to 856: Trochee Fixation.

The title text elaborates on "random text", stating that he once would have believed that Hairy's random outbursts made him interesting despite the definition of "interesting" meaning the opposite. White noise is essentially random sounds waves which taken on mass blend into audio static essentially taking on a macroscopically uniform sound experience despite their random nature. This can be used in some sleep or relaxation therapies, which foils well with the random assault experienced in the comic.

[edit] Transcript

[Black Hat is working peacefully at his computer when an annoying little brat Hairy runs up behind him and goes.]
Hairy: MONKEY TACOS! I'm so random!
Black Hat: Me too!
[Black Hat spews an enormous torrent of random numbers, and Hairy is blasted away.]
[Black Hat returns to work in peace.]
Comment.png add a comment!



Are the numbers in the speech bubble truely random (as in is there a real pattern)? Can someone check? --Charlesisbozo (talk) 08:54, 10 May 2013 (UTC)

I was wondering that myself. I did a quick tally of the digits and for 0..9 I have frequencies of {24,9,18,18,14,17,14,8,9,14} respectively for the readily identifiable digits (YMMV, and while I counted the probable 5 behind Hairy's left ear, I didn't count the possible five behind his left knee, for example.) It doesn't seem to have fallen for the "too many 3s and 7s" trap, nor "too few 3s and 7s, because I know I'll pick them if I try to be random" one, because one is 'high' and one is 'low'. Ditto the "avoiding zero and using nine a lot", says I, vaguely half remembering something from the New Scientists a decade or two ago... While it's not a flat distribution, I'd also suspect it as 'constructed' if it was nearly equal tallies. Someone else can probably tell me if this sample of 145 is within variation limits but I'm still going on intuition.
What I was originally going to do is also go so far as to compare neighbours-on-neighbours. It appeared to me that there were two many like-like neighbours. It's not as easy as in if a grid-system (without holes, etc), but I trivially count a couple of dozen (probably more) and even some 'triples' and that 'stripe' of zeros (from top down to his right knee) is interesting.
That's a sign that it probably is random. Over 100 digits, let's say average 5 neighbours (in a hex grid the internal ones would each have 6 but the ones on the edge fewer), there must be close to 300 or more pairs of neighbours. One-tenth of those would be identical. Truly random sequences have far more identical neighbours than sequences that seem random to us. MGK (talk) 10:44, 10 May 2013 (UTC)
Yup, that's where I was heading with that fact (see "Preliminarily", below). Also, I don't have much more free time today, but if you're interested the corrected frequencies are {24,9,19,19,14,17,14,8,9,15} (I'd missed some!) and the guide to which marks I counted as which numbers is at http://i43.tinypic.com/awc602.png if anyone wants to do the more aesthetic job, like I was originally planning on doing... 10:54, 10 May 2013 (UTC)
Preliminarily, I choose to believe that Randall used a PRNG or even a noise source and stuck to it (even when patterns may have become apparent). Also that, on examining the image closely, he pasted Hairy's anti-aliased image over the top of the numbers then did a little extra editing. ;) 10:24, 10 May 2013 (UTC)
They are all copied directly from the first few lines of A Million Random Digits (and 100000 Normal Deviates) 14:26, 10 May 2013 (UTC)

The irony though, is that for a human being to be able to create truly random content, is indeed interesting. We are pattern forming machines Boxy (talk) 11:10, 10 May 2013 (UTC)

This is Black Hat. I'd personally believe he'd have an (unhackable) /dev/random stream personally available on tap for whenever he needs some significant entropy. Although I imagine he'd use the /dev/urandom one in this instance, knowing that the 'fuller' randomness wouldn't be appreciated enough...

Anyway, I found some free time, and this is the result: http://i39.tinypic.com/nm13dc.png If there's nothing better and it helps at all then anyone please feel free to tidy up (or correct?) and I naturally grant the whole Creative Commons doolally (i.e. to the extent that came with the original source material and what I can personally grant by dint of it being a derivative work by myself) to anyone with a Wiki account who thinks its worthwhile to officially upload it. Or just do it better yourself.

I decided to plump for what another few part-hidden numbers were, as well, while I was at it. Some 9s and a 4, in particular. Now all that is left uncoloured is one possible 5/possible 6 number at the knee area, one that might be a zero behind the head and a smaller fragment behind his lower leg that I imagine is either a 6 or 0, due to the hint of a curve emerging the other side. The 5 behind the left ear is now coloured, but it's possible you might disagree and think it's a 6. (However, I believe Hairy's hair was drawn on after his general bodyplan was moved into position over the numbers, and there is a possible hint of the top-stroke for the 5 emerging from behind the head's anti-aliasing.)

So, while doing this I additionally quanitified the frequencies of neighbouring numbers as I had originally intended.

# Freq ->0 ->1 ->2 ->3 ->4 ->5 ->6 ->7 ->8 ->9 ->?
0 24 20 8 16 11 11 16 11 4 10 19 4
1 9 9 4 3 3 3 7 5 1 1 9 -
2 19 15 3 12 13 11 12 8 7 5 13 1
3 21 9 2 12 18 8 8 12 10 8 6 5
4 15 11 3 14 8 6 10 10 6 8 11 -
5 17 16 9 11 7 10 10 10 7 3 11 4
6 15 10 4 10 11 12 11 9 4 4 7 -
7 10 5 1 8 9 6 8 5 3 5 3 2
8 9 8 1 5 9 8 4 4 5 3 7 -
9 18 19 8 13 7 13 10 8 3 6 12 -

(That came out better than I expected. Not used Wiki table markup for a long time...)

Not included, but analysed, is that average neighbours ranged from 4.67 for 3s to 6.00 for 8s. I'm not sure if that helps any though.

Note that it isn't perfectly symmetrical around the x=y line. I used a strict, but entirely visual, method for deciding whether A neighboured B, and sometimes it did by that measure and yet B did not really neighbour A when later assessed in return. Or vice-versa. Digit size differences and packing of nearby neighbours may have been the prime cause. Input errors also possible of course.

Taking into account the differing frequencies of the (known) numbers, I came up with following table of "actual / theoretical" pairing frequency ratios:

# Dif ->0 ->1 ->2 ->3 ->4 ->5 ->6 ->7 ->8 ->9
0 1.01 1.08 1.02 0.63 0.89 1.14 0.89 0.48 1.35 1.28
1 1.21 1.44 0.51 0.46 0.65 1.33 1.08 0.32 0.36 1.61
2 0.96 0.51 0.97 0.95 1.12 1.08 0.82 1.07 0.85 1.1
3 0.52 0.31 0.87 1.19 0.74 0.65 1.11 1.38 1.23 0.46
4 0.89 0.65 1.43 0.74 0.78 1.14 1.29 1.16 1.72 1.18
5 1.14 1.71 0.99 0.57 1.14 1.01 1.14 1.2 0.57 1.04
6 0.81 0.86 1.02 1.02 1.55 1.25 1.16 0.78 0.86 0.75
7 0.61 0.32 1.22 1.25 1.16 1.37 0.97 0.87 1.61 0.48
8 1.08 0.36 0.85 1.38 1.72 0.76 0.86 1.61 1.08 1.26
9 1.28 1.44 1.1 0.54 1.4 0.95 0.86 0.48 1.08 1.08

The highest values are 1.72 more frequent than ought to be by chance (4<->8, with others not far behind), the lowest is 0.31 what should have occured by chance (3->1, 1<->7 next, 1<->8 then 1->3), and it seems to be an unremarkable progression, end-to-end with no surprising leaps and jumps that grossly disobey any 'meta-frequency' distribution expectations. Note that the 0<->0 value (which stood out on visual inspection) is 1.01, with the median being 1.02, on what should have normalised somewhere around 1.00 anyway. I find the higher frequency not too large for belief, and the lower can be explained by disconnectedness (hole and edge-effect, which wouldn't have occured on a larger, or infinite, array without gaps) but I really should have quantified "missing/unknown neighbours" (after actually excluding the remaining unknowns from analysis), perhaps something like weighting each neighbour's significance according to rarity for the original number to have neighbours, rather than just straight tallying. Too late now without redoing the count from scratch. I'd also considered weighting every instance against every other by inverse-square of distance, or similar, to be somewhat immune from the larger effects, but I'll leave that as an exercise for someone else who wishes to look into it...

Does someone want to calculate P for all of this, anyway? ;)

Finally, I also attempted to discover any embedded steganography. Odd numbers vs even numbers, for a start, but then looking for how "XKCD" or smiley faces or heart-shapes could be marked down in patterns. There are several non-linear sequences of sequential numbers, I noted (can't find anything longer than 4..8 now that I look for them again), but nothing stands out particularly as being above and beyond chance. Yet something might still exist that is far simpler but I managed to overlook it... Unless Black Hat(/Randall) has been so ub3r-1447 as choose such 'randomness' as to encode something into the derivative data!

Anyway, that's it. HTH, HAND, and I'm not spending any more time on this analysis from now on. Probably... 15:54, 10 May 2013 (UTC)

"I'm so Random"

Is it just me, or does this sound like those people at parties who drink a tiny bit, and then spend the whole party saying, "oh my Julia I'm soooo drunk right now!" 11:13, 10 May 2013 (UTC)

I would argue that if Google finds the phrase, it's not random enough. Monkey tacos fails miserably by that measure. It took me several tries [1], but I came up with platypus vindaloo. You have to google with quotes around it to get no matches. Many pages have those words, but none have the phrase. tbc (talk) 11:34, 10 May 2013 (UTC) (Practicing structured procrastination. I have code to write! But this is so fun.)

[1] kitten cupcakes, kitten falafel, snail falafel, snail corn, aardvark corn, aardvark baklava, aardvark vindaloo

Of course, now that you've written it here, that is no longer the case. Although surprisingly, the only result that came up was an unrelated Italian blog with a generic link back to this wiki. For the time being, however, you are the proud owner of a Googlewhack. --H (talk) 18:08, 10 May 2013 (UTC)
Haha. I don't think any googlewhacks have been discovered in years. A true googlewhack must be found without quotes. tbc (talk) 16:32, 11 May 2013 (UTC)
if you count antewhacks, then zwoddery aphasia counts (talk) (please sign your comments with ~~~~)
How about this, without quotes: "w10: brcms_c_radio_timer: dead chip". Got this message from my netbook, which I failed to handle properly a few times in the last few years. 02:37, 6 April 2014 (UTC)
"lexical white noise"

Interesting, since Black Hat speaks only digits! No letters, words, diacritical marks, silly unicode characters, or Fred Flintstone outbursts. The blast of numbers seems to come out in a 2D bubble, yet speech is strictly sequential (each character can have at most two neighbors). The noise is uninteresting at best and likely unwanted, but not unnecessary (sic). It did seem to do the trick. The pesky kid was squelched, Black Hat resumed his work (or whatever).Galois (talk) 14:45, 10 May 2013 (UTC) In fact randomness contains more information as it can't be compressed; I have no idea what Randall means by saying that in information theory randomness is uninteresting.Yehoshua2 (talk) 17:11, 10 May 2013 (UTC) ...'mean and []bullying'? I find this an appropriate response/retribution towards all those proles who think that being random is a viable source of humour. Greyson (talk) 23:01, 10 May 2013 (UTC)

Odd things that have results
Endothermic Zymurgy
Syzygy Chimneysweep
Thermobaric CheeseGrater (???) (talk) (please sign your comments with ~~~~)
Is it random

GOOOGLE is still prove (every law is wrong):

Search 1954678 About 57,600 results (0.09 seconds)

Search 2954678 About 83,300 results (0.10 seconds)

Search 3954678 About 497,000 results (0.24 seconds)

Search 4954678 About 451,000 results (0.24 seconds)

Search 5954678 About 359,000 results (0.25 seconds)

Search 6954678 About 46,300 results (0.40 seconds)

Search 7954678 About 348,000 results (0.24 seconds)

Search 8954678 About 45,400 results (0.24 seconds)

Search 9954678 About 287,000 results (0.23 seconds)

So this law is not valid according to my investigations above: [1]

And I am also random... --Dgbrt (talk) 23:30, 10 May 2013 (UTC)

Just for understanding: I should get get a peak at 1xxxxxx and the lowest result at 9xxxxxx. Just changing the first number.
According to Benford it should be look much different, and my first test on Gooo a few weeks ago did match. Maybe my 7-digit test is not random or it's just a mystery...
--Dgbrt (talk) 23:47, 10 May 2013 (UTC)
Maybe the second digit z9xxxxx is my problem. Still strange, Ask Benford...
--Dgbrt (talk) 23:56, 10 May 2013 (UTC)
Seven digit is too low. Telephone numbers may have 7 digits (up to 10 actually) and they are not random (at least not in a way Benford's law need). And multiple entries in that google search ARE related to telephone numbers. -- Hkmaly (talk) 09:02, 11 May 2013 (UTC)

You sure "tacos" is a trochee? Seems more like a spondee to me. Stevage (talk) 01:44, 13 May 2013 (UTC)

That's some stealth nerd sniping there, Randall. I sincerely did not expect to read someone suggesting that he could've encoded data in the derivative data. Even after reading 1210 comics. Also, it's been two of those in the last three comics. Damn. 04:18, 8 March 2014 (UTC)
Personal tools


It seems you are using noscript, which is stopping our project wonderful ads from working. Explain xkcd uses ads to pay for bandwidth, and we manually approve all our advertisers, and our ads are restricted to unobtrusive images and slow animated GIFs. If you found this site helpful, please consider whitelisting us.

Want to advertise with us, or donate to us with Paypal or Bitcoin?