Complex Complex Sociotechnical Press... Sociotechnical Lecture Three Systems Systems

Stories of Complex Sociotechnical Systems: Measuring Measuring Measurement, Mechanisms, and Meaning Happiness Happiness Some motivation Some motivation Measuring emotional Measuring emotional Lipari Summer School, Summer, 2012 content content Data sets Data sets

Analysis I “Social Scientists wade into the Tweet Analysis Songs Songs Prof. Peter Dodds Blogs stream” by Greg Miller, Blogs Tweets Science, 333, 1814–1815, 2011 [15] Tweets Positivity Bias Positivity Bias Department of Mathematics & Statistics | Center for Complex Systems | “Does a Nation’s Mood Lurk in Its Songs and Vermont Advanced Computing Center | University of Vermont References I References Blogs?” by Benedict Carey New York Times, August 2009. (
)

I More here: http://www.uvm.edu/∼pdodds/research/ (
)

Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License. 1 of 83 4 of 83

Complex Complex Outline Sociotechnical Happiness: Sociotechnical Systems Systems

Measuring Measuring Measuring Happiness Happiness Happiness Some motivation Some motivation Some motivation Measuring emotional Measuring emotional content content Measuring emotional content Data sets Data sets Data sets Analysis Analysis Songs Songs Blogs Blogs Tweets Tweets

Analysis Positivity Bias Positivity Bias Songs References References Blogs Tweets

Bentham: Jefferson: Positivity Bias Socrates et al.: hedonistic . . . the pursuit of eudaimonia [8] calculus happiness References

2 of 83 6 of 83

Complex Complex Papers and so on: Sociotechnical Early drafts: Sociotechnical Systems Systems

“Temporal patterns of happiness and Measuring Measuring information in a global social network: Happiness Happiness Some motivation Some motivation Hedonometrics and Twitter” Measuring emotional Measuring emotional [7] content content Dodds et al., PLoS ONE, 2011 Data sets Data sets Much better version here: Analysis Analysis Songs Songs http://arxiv.org/abs/1101.5120 (
) Blogs Blogs Tweets Tweets I “Positivity of the English Language” Positivity Bias Positivity Bias [11] Kloumann et al., PLoS ONE, 2012 References References

I “Measuring the Happiness of Large-Scale Written Expression: Songs, Blogs, and Presidents” Dodds and Danforth, Journal of Happiness Studies, 2009 [6]

I language assessment by Mechanical Turk (labMT 1.0)

I http://www.onehappybird.com (
)

3 of 83 7 of 83 BREVIA Complex Complex Sociotechnical Emotional content includingSociotechnical the least enjoyable. Although people’s Desiring happiness—not just for boffins: Systems A Wandering Mind Is an mindsSystems were more likely to wander to pleasant topics (42.5% of samples) than to unpleasant topics I Average people routinely report being happy is what Unhappy Mind (26.5% of samples) or neutral topics (31% of sam- [12, 13, 5] ples), people were no happier when thinking about they want most in life Measuring Matthew A. Killingsworth* and Daniel T. Gilbert pleasantMeasuring topics than about their current activity (b = Happiness So how does one measure –0.52,Happiness not significant) and were considerably un- And it matters: “Happy people live longer:. . . ” Some motivation nlike other animals, human beings spend more of 22 activities adapted from the day recon- happierSome when motivation thinking about neutral topics (b = I Measuring emotional 1. happiness?alotoftimethinkingaboutwhatisnot struction method (10, 11), and a mind-wandering –7.2,MeasuringP <0.001)orunpleasanttopics(emotional b = –23.9, [5] content Ugoing on around them, contemplating question (“Are you thinking about something P <0.001)thanabouttheircurrentactivity(Fig.1,content Survey by Diener and Chan. Data sets events that happened in the past, might happen other than what you’re currently doing?”)answered bottom).Data sets Although negative moods are known 2. levels ofin other the future, or emotional will never happen at all. states? Indeed, with one of four options: no; yes, something pleas- to cause mind wandering (13), time-lag analyses Analysis “stimulus-independent thought” or “mind wan- ant; yes, something neutral; or yes, something un- stronglyAnalysis suggested that mind wandering in our Songs dering” appears to be the brain’sdefaultmode pleasant. Our analyses revealed three facts. sampleSongs was generally the cause, and not merely Blogs of operation (1–3). Although this ability is a re- First, people’smindswanderedfrequently,re- theBlogs consequence, of unhappiness (12). Tweets Just ask peoplemarkable evolutionary howhappy achievement that they allows are.gardless of what they were doing. Mind wandering Third,Tweets what people were thinking was a better people to learn, reason, and plan, it may have an occurred in 46.9% of the samples and in at least predictor of their happiness than was what they Positivity Bias [2, 4, 3] Positivity Bias I Experienceemotional sampling cost. Many philosophical and religious(Csikszentmihalyi30% of the samples taken during et every al.) activity were doing. The nature of people’sactivitiesex- References traditions teach that happiness is to be found by except making love. The frequency of mind wan- plainedReferences 4.6% of the within-person variance in hap- living in the moment, and[9] practitioners are trained dering in our real-world sample was considerably piness and 3.2% of the between-person variance in I Day reconstructionto resist mind wandering and “(Kahnemanto be here now.” higher et than al.) is typically seen in laboratory experi- happiness, but mind wandering explained 10.8% National indices of These traditions suggest that a wandering mind is ments. Surprisingly, the nature of people’sactiv- of within-person variance in happiness and 17.7% an unhappy mind. Are they right? ities had only a modest impact on whether their of between-person variance in happiness. The var- well-being: Laboratory experiments have revealed a great minds wandered and had almost no impact on the iance explained by mind wandering was largely But self-reportingdeal about the cognitive has and some neural bases ofdrawbacks: mind pleasantness of the topics to which their minds independent of the variance explained by the na- wandering (3–7), but little about its emotional wandered (12). ture of activities, suggesting that the two were in-

I Bhutan on November 20, 2010 relies onconsequences memory in everyday and life. The self-perception most reliable Second, multilevel regression revealed that peo- dependent influences on happiness. I method for investigating real-world emotion is ex- ple were less happy when their minds were wan- In conclusion, a human mind is a wandering I France perience sampling, which involves[14] contacting peo- dering than when they were not [slope (b)=–8.79, mind, and a wandering mind is an unhappy mind. I inducesple misreporting as they engage in their everyday activities and P <0.001],andthiswastrueduringallactivities, The ability to think about what is not happening I Australia asking them to report their thoughts, feelings, and is a cognitive achievement that comes at an emo- I costly actions at that moment. Unfortunately, collecting tional cost. real-time reports from large numbers of people as References and Notes they go about their daily lives is so cumbersome 1. M. E. Raichle et al., Proc. Natl. Acad. Sci. U.S.A. 98,676 and expensive that experience sampling has rarely (2001). been used to investigate the relationship between 2. K. Christoff, A. M. Gordon, J. Smallwood, R. Smith, www.sciencemag.org mind wandering and happiness and has always J. W. Schooler, Proc. Natl. Acad. Sci. U.S.A. 106,8719 8 of 83 (2009). 12 of 83 been limited to very small samples (8, 9). 3. R. L. Buckner, J. R. Andrews-Hanna, D. L. Schacter, We solved this problem by developing a Web Ann. N. Y. Acad. Sci. 1124,1(2008). application for the iPhone (Apple Incorporated, 4. J. Smallwood, J. W. Schooler, Psychol. Bull. 132,946(2006). Cupertino, California), which we used to create 5. M. F. Mason et al., Science 315,393(2007). 6. J. Smallwood, E. Beach, J. W. Schooler, T. C. Handy, Complex an unusually large database of real-time reports ComplexJ. Cogn. Neurosci. 20,458(2008). Sociotechnical of thoughts, feelings, and actions of a broad range 7.Sociotechnical R. L. Buckner, D. C. Carroll, Trends Cogn. Sci. 11,49(2007). An easy knock: Happiness, attention, and doing: Downloaded from Systems of people as they went about their daily activ- 8.Systems J. C. McVay, M. J. Kane, T. R. Kwapil, Psychon. Bull. Rev. ities. The application contacts participants through 16,857(2009). 9. M. J. Kane et al., Psychol. Sci. 18,614(2007). their iPhones at random moments during their 10. D. Kahneman, A. B. Krueger, D. A. Schkade, N. Schwarz, waking hours, presents them with questions, A. A. Stone, Science 306,1776(2004). Measuring and records their answers to a database at www. 11.Measuring A. B. Krueger, D. A. Schkade, J. Public Econ. 92,1833(2008). trackyourhappiness.org. The database currently 12. Materials and methods are available as supporting Happiness Happinessmaterial on Science Online. Some motivation contains nearly a quarter of a million samples 13.Some J. Smallwood, motivation A. Fitzgerald, L. K. Miles, L. H. Phillips, Measuring emotional from about 5000 people from 83 different coun- MeasuringEmotion 9,271(2009).emotional content tries who range in age from 18 to 88 and who 14.content We thank V. Pitiyanuvath for engineering www. trackyourhappiness.org and R. Hackman, A. Jenkins, Data sets collectively represent every one of 86 major oc- Data sets W. Mendes, A. Oswald, and T. Wilson for helpful comments. cupational categories. Analysis To find out how often people’smindswander, SupportingAnalysis Online Material Fig. 1. Mean happiness reported during each ac- www.sciencemag.org/cgi/content/full/330/6006/932/DC1 Songs Songs what topics they wander to, and how those wan- tivity (top)andwhilemindwanderingtounpleas- Materials and Methods Blogs derings affect their happiness, we analyzed samples ant topics, neutral topics, pleasant topics or not TableBlogs S1 Tweets from 2250 adults (58.8% male, 73.9% residing in mind wandering (bottom). Dashed line indicates ReferencesTweets the United States, mean age of 34 years) who were 18 May 2010; accepted 29 September 2010 Positivity Bias mean of happiness across all samples. Bubble area Positivity Bias randomly assigned to answer a happiness question indicates the frequency of occurrence. The largest 10.1126/science.1192439 (“How are you feeling right now?”)answeredona References bubble (“not mind wandering”)correspondsto References continuous sliding scale from very bad (0) to very 53.1% of the samples, and the smallest bubble Harvard University, Cambridge, MA 02138, USA. good (100), an activity question (“What are you (“praying/worshipping/meditating”)correspondsto *To whom correspondence should be addressed. E-mail: doing right now?”)answeredbyendorsingoneor 0.1% of the samples. [email protected]

932 12 NOVEMBERKillingsworth 2010 VOL 330 SCIENCE and Gilbert,www.sciencemag.org Science, 2010 [10]

Science = Orwell Policy = Brave New World

10 of 83 13 of 83

Complex Complex Sociotechnical We’d like to build an ‘hedonometer’: Sociotechnical Systems Systems

Measuring I An instrument to ‘remotely-sense’ Measuring Happiness Happiness Some motivation emotional states and levels, in real Some motivation Measuring emotional Measuring emotional content time or post hoc. content Data sets Data sets

Analysis Analysis Songs Songs Blogs Blogs Tweets Ideally: Tweets

Positivity Bias Positivity Bias References I Transparent References I Non-reactive I Fast I Complementary to I Based on written self-reported measures expression I Improvable I Uses human evaluation Some possibilities:

I Natural language processing (e.g., OpinionFinder) [16] I See story here (
) for example [slate]. 11 of 83 I Declared mood levels in blogs (e.g., Livejournal) 14 of 83 Complex Complex ANEW study Sociotechnical Analysing text: Sociotechnical Systems Systems

Measuring Measuring Happiness Happiness I ANEW = “Affective Norms for English Words” Some motivation Some motivation Measuring emotional Measuring emotional content content Data sets Data sets I Study: participants shown lists of isolated words Analysis Analysis Songs Songs Blogs ANEW Blogs I Asked to grade each word’s valence, arousal, and Lyrics for v f Tweets words k k Tweets Michael Jackson’s Billie Jean X vkfk dominance level k Positivity Bias k=1. love 8.72 1 vtext = Positivity Bias “She was more like a beauty queen References 2. mother 8.39 1 X fk References I Integer scale of 1–9 from a movie scene. 3. baby 8.22 3 k 4. beauty 7.82 1 And mother always told me, 5. truth 7.80 1 6. people 7.33 2 be careful who you love. vBillie Jean = 7.1 I N =1034 words—previously identified as bearing And be careful of what you do 7. strong 7.11 1 8. young 6.89 2 ’cause the lie becomes the truth. emotional weight 9. girl 6.87 4 v = 6.3 Billie Jean is not my lover, 10. movie 6.86 1 Thriller I Participants = College students (*cough*) She’s just a girl who claims 11. perfume 6.76 1 that I am the one. 12. queen 6.44 1 vMichael = 6.4 Results published by Bradley and Lang (1999) [1] 13. name 5.55 1 Jackson I 14. lie 2.79 1

15 of 83 18 of 83

Complex Complex ANEW study—three 1–9 scales: Sociotechnical Data sets: Sociotechnical Systems Systems

valence: Measuring Measuring Happiness Happiness Some motivation Some motivation Measuring emotional Texts: Measuring emotional content content Data sets 1. Song lyrics (1960–2007) Data sets Analysis Analysis Songs 2. Song titles (1960–2008) Songs Blogs Blogs arousal: Tweets 3. State of the Union (SOTU) Addresses (1790–2008) Tweets Positivity Bias Positivity Bias References Sources: References I hotlyrics.com (
) dominance: I freedb.com (
) I American Presidency Project: www.presidency.ucsb.edu (
).

16 of 83 20 of 83

Complex Complex ANEW study words—examples Sociotechnical Data sets: Sociotechnical Systems Systems 4. Blog phrases containing “I feel...”, “I am feeling”, etc., 9 Measuring taken from wefeelfine.org ( ) (API, 2005–2010) Measuring love/paradise/triumphant Happiness
Happiness 8 Some motivation Some motivation Measuring emotional Measuring emotional glory/luxury/trophy content content 7 Data sets Data sets optimism/pancakes/church Analysis Analysis v 6 Songs Songs engine/paper/street Blogs Blogs 5 Tweets Thanks to ... Tweets derelict/neurotic/vanity Positivity Bias Kameron Harris Isabel Kloumann Catherine Bliss Positivity Bias

valence 4 References References fault/corrupt/lawsuit 3 trauma/hostage/disgusted 2 funeral/rape/suicide 1 0 50 100 150 200 frequency I Created by [1] ANEW = “Affective Norms for English Words” Jonathan Harris

17 of 83 & Sep Kamvar 21 of 83 Jonathan Harris & Sep Kamvar wefeelfine.org Complex Complex wefeelfine.org: Sociotechnical Song Lyrics—average happiness (valence) Sociotechnical Systems Systems

Measuring Measuring Happiness 6.8 Happiness Some motivation Some motivation Measuring emotional Measuring emotional content 6.7 content Data sets Data sets

Analysis 6.6 Analysis avg Songs Songs Blogs 6.5 Blogs Tweets Tweets

Positivity Bias 6.4 Positivity Bias References 6.3 References 6.2

mean valence v 6.1 6 5.9 1960 1970 1980 1990 2000 2010 year

22 of 83 26 of 83

Complex Complex More data sets: Sociotechnical Song Lyrics—average happiness of genres: Sociotechnical Systems Systems

Measuring Measuring Happiness Happiness Some motivation Some motivation Measuring emotional 7 Measuring emotional content content Data sets Data sets

Analysis Analysis

avg 6.5 5. Songs Songs Blogs Blogs Tweets Tweets

Positivity Bias 6 Positivity Bias 6. New York Times (20 years) References Gospel/Soul (6.91) References 5.5 Pop (6.69) 7. Gutenberg.org Reggae (6.40) Rock (6.27) 8. Google Books: http://ngrams.googlelabs.com/ (
) mean valence v 5 Rap/Hip−Hop (6.01) 9. ... Punk (5.61) Metal/Industrial (5.10) 4.5 1960 1970 1980 1990 2000 2010 year

23 of 83 27 of 83

Complex Complex Some numbers: Sociotechnical Happiness Word Shift Graph: Sociotechnical Systems Systems

Measuring Per word drop in valence of lyrics from 1980−2007 relative to valence of lyrics from 1960−1979:Measuring Happiness Happiness Counts Song lyrics Song titles Some motivation Some motivation 1 love ↓ Measuring emotional ↓ Key: Measuring emotional All words 58,610,849 60,867,223 content 2 lonely content 3 hate ↑ Data sets ↑ Data sets 4 pain ↓ ↑ ANEW words 3,477,575 (5.9%) 5,612,708 (9.2%) 5 baby ↓ love life Analysis baby ↓ god ↑ Analysis 6 death ↑ ↓ ↑ ∼ ∼ Songs home truth Songs Individuals 20,000 632,000 7 dead ↑ music ↓ party ↑ Blogs 8 home ↓ good ↓ sex ↑ Blogs Tweets 9 sick ↑ Tweets i Decreases in relatively Increases in relatively Counts blogs SOTU 10 fear ↑ Positivity Bias 11 hit ↑ high valence words high valence words Positivity Bias 12 hell ↑ contribute to drop contribute to increase All words 155,667,394 1,796,763 References 13 fall ↑ in average valence in average valence References 14 sin ↑ ↑ ANEW words 8,581,226 (5.5%) 61,926 (3.5%) 15 lost ↑ ↓ 16 sad ↓ hate lonely Word number pain ↑ sad ↓ Individuals ∼ 2,335,000 43 17 burn ↑ death ↑ trouble ↓ 18 lie ↑ dead ↑ loneliness ↓ 19 scared ↑ sick ↑ devil ↓ ↑ 20 afraid Increases in relatively Decreases in relatively Counts Twitter 21 music ↓ ↑ low valence words low valence words 22 life contribute to drop contribute to increase 23 ↑ All words ∼ 100 billion god in average valence in average valence 24 trouble ↓ Tweets ∼ 10 billion 25 loneliness ↓ Individuals ∼ 100 million −20 −10 0 10 Per word valence shift ∆ i

24 of 83 28 of 83 Complex Top 16 of ' 20,000 artists: Sociotechnical Text: h Words with a similar score: Systems avg Rank Artist Valence Soul/Gospel 6.9 chocolate (6.88), leisurely (6.88), lyrics [6] penthouse (6.81) 1 All-4-One 7.15 Measuring Happiness Pop lyrics [6] 6.7 dream (6.73), honey (6.73), sugar (6.74) 2 Luther Vandross 7.12 Some motivation Measuring emotional Dante’s Paradise 6.5 muffin (6.57), rabbit (6.57), smooth 3 S Club 7 7.05 content 4 K Ci & JoJo 7.04 Data sets (6.58) 5 Perry Como 7.04 Analysis Tweets, 9/9/2008 6.4 thought (6.39), face (6.39), blond (6.42) Songs 6 Diana Ross & The Supremes 7.03 Blogs to 12/31/2010 Tweets [6] 7 Buddy Holly 7.02 Rock lyrics 6.3 church (6.28), tree (6.32), air (6.34) Positivity Bias 8 Faith Evans 7.01 Enron Emails (
) 6.2 clouds (6.18), alert (6.20), computer References 9 The Beach Boys 7.01 (6.24) 10 Jon B 6.98 State of the Union 6.1 grass (6.12), idol (6.12), bottle (6.15) [6] 11 Dru Hill 6.96 Messages 12 Earth Wind & Fire 6.95 New York Times 6.0 hotel (6.00), tennis (6.02), wonder (6.03) [17] 13 Ashanti 6.95 (1987–2007) [6] 14 Otis Redding 6.93 Blogs 5.8 owl (5.80), whistle (5.81), humble (5.86) 15 Faith Hill 6.93 Dante’s Inferno 5.5 glacier (5.50), repentant (5.53), mischief 16 NSync 6.93 (5.57) Heavy Metal 5.4 lamp (5.41), elevator (5.44), truck (5.47) (criteria: ≥ 50 songs and ≥ 1000 ANEW words) lyrics [6] 29 of 83

Complex Complex Bottom 16 of ' 20,000 artists: Sociotechnical Blogs Sociotechnical Systems Systems Rank Artist Valence

1 Slayer 4.80 Measuring Measuring Happiness 6.1 Happiness 2 Misfits 4.88 Some motivation Some motivation Measuring emotional 6 Measuring emotional 3 Staind 4.93 content content 4 Slipknot 4.98 Data sets 5.9 Data sets 5 Darkthrone 4.98 Analysis Analysis Songs 5.8 Songs 6 Death 5.02 Blogs Blogs Tweets Tweets 7 Black Label Society 5.05 5.7 Positivity Bias Positivity Bias

8 Pig 5.08 valence (v) References 5.6 References 9 Voivod 5.14 5.5 10 Fear Factory 5.15 13 20 30 40 50 60 70 80 11 Iced Earth 5.16 12 Simple Plan 5.16 blogger age 13 Machine Head 5.17 14 Metallica 5.19

15 Dimmu Borgir 5.20 I Average happiness as a function of the age bloggers 16 Mudvayne 5.21 report they will turn in the year of their posting. (criteria: ≥ 50 songs and ≥ 1000 ANEW words) 30 of 83 34 of 83

Tref: born in 1960-1969 (havg=5.96) Complex Sociotechnical Blogs—Overall trend Tcomp: 14 years old (havg=5.55) Systems 1 −↑sick −↑hate −↑stupid Measuring −↑sad Happiness 5 happy +↑ Some motivation Measuring emotional 6.4 love +↑ content US Election 11/4 −↑depressed Data sets 6.3 −↑bored ↑ Analysis avg − lonely Songs h US Inauguration 1/20 ↑ 6.2 r 10 − alone Blogs ♥ ♥ −↑mad Text size: Tweets 6.1 ♥ −↑pain 0 Tref Tcomp Positivity Bias 10 +↓life 6 ♥ loved +↑ References 15 −↑upset 1 ↑ 5.9 Word rank 10 − fat ♥ fun +↑ Balance: Michael −↑dead average happiness 5.8 9/10 Jackson ↑ −169 : +69 9/10 2 − scared +↓ +↑ 9/10 9/11 20 10 −↑terrible 5.7 ↑ A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A friend + people +↑ 103 −↑confused 2005 2006 2007 2008 2009 2010 −100 0 ↓ r time − −↑ −↓ 25 Pi=1 δhavg,i −↑hurt −20 −10 0 10 20

Per word average happiness shift δhavg,r (%) 35 of 83 Tref: Male (havg=5.91) Complex Complex Sociotechnical Sociotechnical Tcomp: Female (havg=5.89) Words most correlated with obesity levels in Systems Systems 1 love +↑ cities: −↑hurt −↑hate Measuring Measuring ↑ Happiness Happiness − sad Word havg rs p-value 5 +↓good Some motivation Some motivation Measuring emotional Measuring emotional −↑alone content stomach 5.40 0.37 1.98894e-07 content baby +↑ Data sets mcdonalds 5.98 0.30 2.60824e-05 Data sets loved +↑ happy +↑ Analysis hungry 3.38 0.27 0.000206297 Analysis ↑ Songs Songs r 10 − stupid Blogs wings 6.52 0.25 0.000388915 Blogs −↑guilty Text size: Tweets Tweets −↑sick ham 5.66 0.24 0.000763101 0 Tref Tcomp Positivity Bias Positivity Bias 10 heart +↑ starving 2.58 0.22 0.00272286 −↑scared References References 15 −↑lost spaghetti 0.00 0.20 0.00689403 1 ↓ Word rank 10 + music ihop 0.00 0.19 0.0100034 +↓free Balance: death −↓ noodles 0.00 0.18 0.0106139 ↑ −607 : +507 2 life + +↓ +↑ ketchup 0.00 0.18 0.0145088 20 10 family +↑ +↓christmas fat 3.24 0.18 0.0148845 cold −↓ sprite 0.00 0.17 0.0175705 103 −↑upset −100 0 ↑ cookin 0.00 0.17 0.0182976 r friend + −↑ −↓ 25 Pi=1 δhavg,i dead −↓ heartburn 0.00 0.17 0.0200551 −100 −50 0 50 100 sugar 6.74 0.15 0.0329359 δh kool-aid 0.00 0.15 0.0354226 Per word average happiness shift avg,r (%) 36 of 83 miller 5.36 0.15 0.036325 40 of 83 honey 7.44 0.15 0.0395531 candy 7.52 0.15 0.0398618 Complex Complex Twitter—living in the now: Sociotechnical Words most anti-correlated with obesity Sociotechnical Systems Systems levels in cities: Measuring Measuring 0.16 Happiness brunch 6.32 -0.41 6.37431e-09 Happiness breakfast Some motivation Some motivation Measuring emotional bar 5.82 -0.35 5.54374e-07 Measuring emotional 0.14 lunch content content Data sets banana 6.86 -0.35 5.67492e-07 Data sets 0.12 dinner Analysis barista 0.00 -0.35 7.29324e-07 Analysis Songs Songs Blogs delicious 7.92 -0.34 1.09807e-06 Blogs 0.1 Tweets dinner 7.40 -0.34 1.35413e-06 Tweets Positivity Bias Positivity Bias 0.08 coffee 7.18 -0.34 2.04145e-06 References espresso 0.00 -0.33 4.45903e-06 References 0.06 cocktails 0.00 -0.32 4.96518e-06

count fraction booze 0.00 -0.32 6.38461e-06 0.04 mimosa 0.00 -0.31 1.24472e-05 0.02 spiced 0.00 -0.31 1.52074e-05 veggie 0.00 -0.31 1.60439e-05 0 sushi 5.40 -0.31 1.71997e-05 0 2 4 6 8 10 12 14 16 18 20 22 24 wines 6.28 -0.31 1.7432e-05 hour of day (local time) tofu 0.00 -0.31 1.86278e-05 panini 0.00 -0.31 1.86719e-05 38 of 83 gnocchi 0.00 -0.30 2.51419e-05 41 of 83 clams 0.00 -0.30 2.52124e-05 caffeine 5.80 -0.30 2.6263e-05 Complex Complex Twitter—living in the now: Sociotechnical Twitter—livingcocktail in the0.00 now:-0.30 2.63227e-05 Sociotechnical Systems bento 0.00 -0.30 2.6349e-05 Systems huevos 0.00 -0.30 3.19903e-05 Measuring mojitos 0.00 -0.30 3.52542e-05 Measuring 0.07 Happiness Happiness Some motivation vegan 4.82 -0.30 3.5502e-05 Some motivation Measuring emotional Measuring emotional content content 0.06 Data sets Data sets

Analysis Analysis 0.05 Songs Songs Blogs Blogs Tweets Tweets

0.04 Positivity Bias Positivity Bias

References References 0.03 hungry starving count fraction 0.02 food 0.01 eat 0 0 2 4 6 8 10 12 14 16 18 20 22 24 hour of day (local time) Tweeting the Superbowl (
) [NY Times]

39 of 83 42 of 83 The happiest distribution:

haha 6 10 ha hahaha

hahahahaha

4 10

hahahahahahahahahahahahaha 2 Frequency 10

0 10 hahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahaha 2 4 6 10 26 100 140 Number of Letters (String Length)

labMT 1.0: language assessment by Mechanical Turk

http:/www.onehappybird.com (
) http://flowingdata.com/2011/10/27/language-communities-of-twitter/ ( ) Complex
valence word valence std dev twitter g-books nyt lyrics Sociotechnical rank rank rank rank rank Systems

1 laughter 8.50 0.93 3600 – – 1728 2 happiness 8.44 0.97 1853 2458 – 1230 3 love 8.42 1.11 25 317 328 23 Measuring 4 happy 8.30 0.99 65 1372 1313 375 Happiness 5 laughed 8.26 1.16 3334 3542 – 2332 Some motivation 6 laugh 8.22 1.37 1002 3998 4488 647 Measuring emotional content 7 laughing 8.20 1.11 1579 – – 1122 Data sets 8 excellent 8.18 1.10 1496 1756 3155 – 9 laughs 8.18 1.16 3554 – – 2856 Analysis 10 joy 8.16 1.06 988 2336 2723 809 Songs 11 successful 8.16 1.08 2176 1198 1565 – Blogs 12 win 8.12 1.08 154 3031 776 694 Tweets 13 rainbow 8.10 0.99 2726 – – 1723 14 smile 8.10 1.02 925 2666 2898 349 Positivity Bias 15 won 8.10 1.22 810 1167 439 1493 16 pleasure 8.08 0.97 1497 1526 4253 1398 References 17 smiled 8.08 1.07 – 3537 – 2248 18 rainbows 8.06 1.36 – – – 4216 19 winning 8.04 1.05 1876 – 1426 3646 20 celebration 8.02 1.53 3306 – 2762 4070 21 enjoyed 8.02 1.53 1530 2908 3502 – 22 healthy 8.02 1.06 1393 3200 3292 4619 23 music 8.02 1.12 132 875 167 374 24 celebrating 8.00 1.14 2550 – – – 25 congratulations 8.00 1.63 2246 – – – 26 weekend 8.00 1.29 317 – 833 2256 27 celebrate 7.98 1.15 1606 – 3574 2108 28 comedy 7.98 1.15 1444 – 2566 – 29 jokes 7.98 0.98 2812 – – 3808 30 rich 7.98 1.32 1625 1221 1469 890 ...... 48 of 83 Complex valence word valence std dev twitter g-books nyt lyrics Sociotechnical Twitter—overall time series: rank rank rank rank rank Systems

...... 6.4 ...... Measuring Monday 2008—2009—12/25 12/25 2010—12/25 2011— Tuesday 10193 violence 1.86 1.05 4299 1724 1238 2016 Happiness Wednesday 6.3 Thursday

avg 12/25 10194 cruel 1.84 1.15 2963 – – 1447 12/31 12/24 12/31 Friday Some motivation h 12/24 01/01 12/24 A 04/12 11/26 Saturday 10195 cry 1.84 1.28 1028 3075 – 226 02/14 02/14 Measuring emotional 6.2 07/04 05/09 Sunday 06/21 07/04 02/14 12/24 12/31 content 11/27 01/01 04/04 01/01 10196 failed 1.84 1.00 2645 1618 1276 2920 12/31 10/31 06/20 11/25 04/29 10/31 05/08 10197 sickness 1.84 1.18 4735 – – 3782 Data sets 6.1 11/24 06/19 10198 abused 1.83 1.31 – – – 4589 04/24 10199 tortured 1.82 1.42 – – – 4693 Analysis 6 Songs 10200 fatal 1.80 1.53 – 4089 – 3724 04/27 06/25 09/14 09/29 08/06 02/27 05/24 06/27 Blogs average happiness5.9 03/11 07/05 10201 killings 1.80 1.54 – – 4914 – 05/02 Tweets 10202 murdered 1.80 1.63 – – – 4796 07/23 08/0808/2309/1109/21 5.8

10203 war 1.80 1.41 468 175 291 462 S Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Positivity Bias 10204 kills 1.78 1.23 2459 – – 2857 N 700 10205 jail 1.76 1.02 1642 – 2573 1619 References 600 10206 terror 1.76 1.00 4625 4117 4048 2370 B 10207 die 1.74 1.19 418 730 2605 143 500

10208 killing 1.70 1.36 1507 4428 1672 998 400 10209 arrested 1.64 1.01 2435 4474 1435 – 10210 deaths 1.64 1.14 – – 2974 – 300

10211 raped 1.64 1.43 – – – 4528 Simpson lexical size Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

) 4

10212 torture 1.58 1.05 3175 – – 3126 7

10213 died 1.56 1.20 1223 866 208 826 10 3 C 10214 kill 1.56 1.05 798 2727 2572 430 10215 killed 1.56 1.23 1137 1603 814 1273 2 10216 cancer 1.54 1.07 946 1884 796 3802 10217 death 1.54 1.28 509 307 373 433 1

10218 murder 1.48 1.01 2762 3110 1541 1059 word count0 (x 10219 terrorism 1.48 0.91 – – 3192 – Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 10220 rape 1.44 0.79 3133 – 4115 2977 date 10221 suicide 1.30 0.84 2124 4707 3319 2107 10222 terrorist 1.30 0.91 3576 – 3026 – 49 of 83

Complex std dev word valence std dev twitter g-books nyt lyrics Sociotechnical Twitter—overall time series: rank rank rank rank rank Systems

1 fE@king 4.64 2.93 448 – – 620 2 f&&kin 3.86 2.74 1077 – – 688 3 f&&ked 3.56 2.71 1840 – – 904 Measuring 4 pussy 4.80 2.66 2019 – – 949 Happiness Some motivation 5 whiskey 5.72 2.64 – – – 2208 Measuring societal happiness through tweets: Measuring emotional 6 slut 3.57 2.63 – – – 4071 6.4 content 2008— 2009— 2010— 2011— Monday 12/25 12/25 12/25 Tuesday 7 cigarettes 3.31 2.60 – – – 3279 6.3 Wednesday Data sets avg 12/25 12/31 12/24 12/31

h 12/24 12/24 && 01/01 04/12 11/26 Thursday 8 f k 4.14 2.58 322 – – 185 02/14 02/14 Friday 6.2 07/04 05/09 06/21 07/04 02/14 Saturday 12/24 11/27 01/01 04/04 01/01 9 mortality 4.38 2.55 – 3960 – – 12/31 10/31 06/20 Sunday Analysis 11/25 04/29 10/31 05/08 6.1 10/31 11/24 10 cigarette 3.09 2.52 – – – 2678 06/19 Songs 04/24 12/31 10/31 11 motherf&&kers 2.51 2.47 – – – 1466 Blogs 6 04/27 06/25 09/14 12 churches 5.70 2.46 – 2281 – – 09/29 08/06 02/27 05/24 06/27

Tweets average happiness 5.9 03/11 07/05 05/02 13 motherf&&king 2.64 2.46 – – – 2910 07/23 08/08 08/23 09/11 09/21 5.8 Positivity Bias Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 14 capitalism 5.16 2.45 – 4648 – – date 15 porn 4.18 2.43 1801 – – – 16 summer 6.40 2.39 896 1226 721 590 References 17 beer 5.92 2.39 839 4924 3960 1413 18 execution 3.10 2.39 – 2975 – – 19 wines 6.28 2.37 – – 3316 – I Global happiness spikes = predictable rituals. 20 zombies 4.00 2.37 4708 – – – 21 aids 4.28 2.35 2983 3996 1197 – Global sadness spikes = unpredictable, exogeneous 22 capitalist 4.84 2.34 – 4694 – – I 23 revenge 3.71 2.34 – – – 2766 shocks. 24 mcdonalds 5.98 2.33 3831 – – – 25 beatles 6.44 2.33 3797 – – – 26 islam 4.68 2.33 – 4514 – – I No accidental happiness outbreaks. 27 pay 5.30 2.32 627 769 460 499 28 alcohol 5.20 2.32 2787 2617 3752 3600 29 muthaf&&kin 3.00 2.31 – – – 4107 30 christ 6.16 2.31 2509 909 4238 1526 ...... 50 of 83

The very surprising tunable hedonometer: Bailout of the U.S. financial system: Royal Wedding of Prince William & Catherine Middleton: Death of Osama Bin Laden: T : 7 days before and after (h =5.98) Tref: 7 days before and after (havg=6.00) Tref: 7 days before and after (havg=5.98) C ref avg A Tcomp: Monday, 2008/09/29 (havg=5.95) B Tcomp: Friday, 2011/04/29 (havg=6.04) Tcomp: Monday, 2011/05/02 (havg=5.89) 7.2 1−↑bailout 1 wedding +↑ 1 −↑dead −↑bill dead −↓ −↑death 7 A ∆havg : −↑down dont −↓ −↑killed −↑failed death −↓ +↓love 2.0 5 −↑not 5 beautiful +↑ 5 −↑not 6.8 last −↓ hate −↓ −↑kill avg 1.8 ↑ ↑ −↑died

h − no kiss + 6.6 1.6 −↑fail prince +↑ −↑killing

1 −↑fails +↓easter +↓happy 1.4 10 −↑blame 10 +↓happy 10 −↑saddam 6.4 1.2 +↓love no −↓ +↓haha −↑failure princess +↑ +↓me 6.2 1.0 −↑bad never −↓ −↑war 0.8 −↑don’t shit −↓ +↓mothers 15 −↑against 15 killed −↓ 15 −↑terrorist 6 0.6 −↑die not −↓ −↑bad 0.4 −↑rejected dress +↑ −↑pakistan 5.8 −↑depression real +↑ −↑buried 0.2 ↑ ↓ no −↓ Average happiness − crisis + me 0 20 money +↑ 20 +↓good 20 −↑hussein 5.6 +↓sunday +↓you +↓hahaha +↓fun weekend +↑ −↑terror r r r ↓ 5.4 +↓party friday +↑ + like +↓game party +↑ +↓wedding ↑ ↓ ↑ 10/01/08 01/01/09 04/01/09 07/01/09 10/01/09 01/01/10 04/01/10 07/01/10 10/01/10 01/01/11 25 weekend + 25 + chocolate 25 − terrorism +↓won +↓love −↑enemy Date old −↓ +↓win +↓lol −↑worst married +↑ −↑shot Word rank Word rank ↓ Word rank ↓ ↑ 3 1 10000 100 + great + game celebrating + 30 −↑panic 30 live +↑ 30 ill −↓ 8000 80 +↓awesome bad −↓ −↑hell 1200 2∆havg D E ↑ B C 0.8 −↑didn’t congrats +↑ celebrate + 2.5 6000 60 +↓google amazing +↑ +↓mom +↓saturday kill −↓ −↑loss 1000 4000 40 0.6 35 billion +↑ 35 ill −↓ 35 −↑mourn 2 # words 2000 20 −↑hurt nigga −↓ america +↑ % coverage ↓ Text size: ↑ Text size: +↓you Text size: 0 + win wow + T T 800 0 0 ↓ Tref Tcomp 0 ↓ Tref Tcomp 0 ↑ ref comp 0.4 0 1 2 3 0 1 2 3 10 cancer − 10 died − 10 − deaths avg ↑ ↓ ↑ 1.5 − sick + life 1st + h ↑ ↓ ↑ ∆havg ∆havg 40 − problem 40 ass − 40 1 sea + 600 101 101 10 ↑ ∆ 0.2 −↑crash hell −↓ usa + 100 ↓ ↑ ↓ Frequency + friday Balance: gorgeous + Balance: + beautiful Balance: 1 2 ↑ 2 ↑ 2 peace +↑ 80 F 10 − falling −165 : +65 10 congratulations−68 + : +168 10 −143 : +43 400 0 house +↑ +↓lol +↓fun 60 ↓ +↓ +↑ ↑ +↓ +↑ ↑ +↓ +↑ 45 3 + home 45 3 couple + 45 3 − evil 0.5 40 10 −↑killed 10 +↓friends 10 +↓good 200 −0.2 −↑fear kissed +↑ −↑attack 4 ↓ 4 ↑ 4 ↑ 20 10 miss − 10 she + 10 we + % coverage −100 0 ↑ 0 100 ↓ −100 0 ↑ − gossip −↑ −↓ killing − −↑ −↓ − breaking −↑ −↓ 0 0 0 r ↑ r ↓ Pr δh ↑ 1 2 3 4 5 6 7 8 9 0 1 2 3 0 5000 10000 15000 20000 50 Pi=1 δhavg,i − poor 50 Pi=1 δhavg,i + haha 50 i=1 avg,i − burial havg ∆havg word rank −20 −10 0 10 20 −30 −20 −10 0 10 20 30 −20 −10 0 10 20

Per word average happiness shift δhavg,r (%) Per word average happiness shift δhavg,r (%) Per word average happiness shift δhavg,r (%) Complex 0 Twitter—weekly time series: Sociotechnical ∆havg = 1.00 Systems

) −1 2009−05−21 to 2010−12−31: Measuring

6.45 avg Happiness h you Some motivation myme

(∆ no Measuring emotional lol content f not up 6.4 −2 dont knowwilldayallwenewlikelove Data sets todaygood makeseemorethink haha

avg cant people conlast checkourpleaseshevideousrightwellhomethanksbesthahahagreat h shitill never heyschoolshowfirstbetterlivelife freehappy Analysis bad downdidnt blogfindhelpverynightbigalwaysworldyes hope diehate misswaitassstop facebookeveryonehatrymostsuremorninggiveonlineyoutuberealgirlthanklunchnicemusic 6.35 Songs damn sorryold tonightstartguyshouseamorphoneiphonehehedonetopgamewishmoneycoolbabyawesomegodsongfunwin meannothingdoesnt tomorrowfeliztalkmasocialthoughtbuysoonwitreadreadytv sonplayinternetamazingwowbirthdaysuperweekendfriends hell nigga hairfacemakingownyea carecarhigirlsgoogleeattruesleepprettyfoodhahahahapartyfriendfunny Blogs wrongbitchwithout isnt bientempomindopenlisteningmyselfteamboyfullsupportfoundalbumokaybelieveladyhighmoviehelloplayingfansheartbookkidsbeautiful lost waiting jamrockthinkingwebsiterememberwhitejoinsummermeetraintalkingipadfanemailwelcomeapplesexcutemomfamily −3 deadsick mad boredtiredgonelate least downloadfavoryayminechancestoryradiosalefridayfavoritewomenideabedexcitedsweet sadstupidsin forgetughwasnt servicestartedlmfaobuenolatestinfoyourselfplaneyesbrodealsavefilmpowerwatertogetherlearngreeneasywomanphotoairdearwonderdreamhealth1stkindlucksexygladenjoy Tweets problem missed avatarwordswalkeverybodydrivecdamigodesignsnowjobslondonchatavailablefastvisitreadingtourdogfutureworthfineplusyoungphotosshoppinggivingparkspecialcoffeehahahgamesstareatingperfect niggashungryagainstcut fire muchototallyforwardshowseventhotelboysateticketsmobiletipsdatelightstartingvideosipodpictureartagreedvdcollegehehehemamarestsaturdayinterestingdancecongratschristmassunbeachsmile kill lose sucksalonefat wouldntbye windowsdigitalvotedsouthanswermarketlotsunderstandtouchforeverbandpicsjesusamericanusashareshopkidsongsseatripdadcomputerbrotherholidaybdaychocolategiftloves war hurt fight forgotcouldntshutbilllowfuckin studyhugekingwritingyoursboughtgavechickenseeingdrinknapenglishableliving ladiescouplelovelymillionluckyloved death #fail boo bang networkexactlyhandspersonalsimplemessagequickactreleaseknowsdefinitelyimportantsoftwarecountryfootballfashionapril finishedteasistercleanawardsspringpizzaenergytravellikedsundaytrustluvdinnerhahahahahawon failworstpainpoor lie boringmissingfakesuck slow itunessportsmalesblackberryrideeyeplansstudiojuneextradressshoweramericacamerastyleconcertpictureslaptopideasbooksfreshmoviestruthfestivalkisssuccesslaugh 6.3 cry shotlossbitchesworry cannottraffichospitalshootnobodystuckblood rip shoesguidewalkingdrivingpossibleexperiencechoicexboxteacherfixpinklordflymailinternationalsisahahawifecatfinishhopefullyartistbreakfastbabestrongchildrenhahahahchildparentsprouddreamssinggoldwonderfulmotherloving broke leaving darkshout biggeststraightcommunityhumanselfshirtsensejoinedoriginalvegasinformationindiaoffercompletesantacreatefairsafepaidweddingcash uglyproblemsstresstaxdumb crap kickcourt globalcontactsecurityalrightballthoughtsbuildfitplayedsoulthinksclothesearthgivesgrowwarminterestedcreamcheeseparisbrotherscolorstars smartcakepeace cancerkilled scaredjealousworseannoying lazy hang updatedaheadespeciallybrazilplanningopeningmiamistudentstechlivesleadlaunchresearchwinelibrarynaturalbrainfavoritedstudentprayholy femalesingingqualityunited jokebeautywinnerhealthy Positivity Bias hurts brokendirty caught endsbreakingmess wonderingwashingtonsitesnovemberfolksflightfocusaahwedadviceabsolutelycampringbuyingpopularperformancestoriestechnologycookpoolcafelearningtoyrelationshiphopingimaginerespect enjoyingsky T W T F S S M T W T F S S M died attack smokefellfallingfool closedhomeworkdrunkissues professorcontentmegajapancaliforniareleasedcleargiveawaydatingclassicdisneyuniversitymarriedbikeamazonpapamagicdogsscienceborntastegrandblessappreciatealivegreatestlikesgorgeoushilariouswinsexcitingvacationaward killing fear trouble fucked quit cardspointscollectionwindowupdatesdrinkingasleepplacesyahoosonywiitradeestatetunelandeventscompletelychristiandevelopmentseptemberbeginepicaaahmateorangefishfamousislandeducationcinemapresentsunnysleepingdaddygoalpromise fantasticpositivecongratulationscelebrateexcellent horriblepissed sore doubt lamezerozit indeedcampuskitchenxoxolargethemeahaqueenspeedenglandcarsvalueeveningstarbucksgainfeedoffersteachcareerthanxfloridamumtreatgleemomsrestaurantopportunitydancingpethelpingtreegardenguitarmooncreativehumor richlaughing evil angry nastyshameafraidwasteblame bugnothin nonehumidity celebrityxp halloweentigerworldsmilkskypeviptipfranceboyfriendjeansaustraliabiggerrisedealsnightse−mailhelpsanimaltrulydaughterdollarfatherfaith honeyearn hugswinninghappiness cryingdying losingerrorspamfightingterribleconfuseddrug monsterapart stoppedtoughforce malekeepingbasketballlaunchesphonessamsungindianhollywooddecemberprofessionaldeservesimplyexclusivesoupsaladsportpiebuddylearnedfitnesssandwichsharingyumyummy easierhugangeldeliciousjoy headachebomb scarylies burnbattlerob dopepressureriskcrack createdcertainpickedoscarburgertradingunlockeddesignerplanebreadwavemarioleadersilvercousincoastmodernlikelygrowingincreaseplanetguesthusbandbirdlake cookinggoldenrosesurprisebrillianthero dies shootingtearscrashrude gundebtbullshit goodbye twit walkedbeginningfixednationinstantbellahosticon haloflyingdiscolightseuropescoreacceptmemoryprincehigherultimatelanguageanniversaryknowingraisepromoteagreedtalentfavouritehonornature entertainmentblessedheavensuccessful bloodysmokingavoid hanginggangow japanesephotoshopexpressamenpurpleactinglegsarrivedpregnantengineeraidreceivedbbqkiddingfasterwifisoccerreceivespiritsleptsurprisedriversugarartists#musicpurehonestlygirlfriendjuicediscountvalentinesincrediblefantasyromancecandybillion freedomsunshineholidaysenjoyed jailwars killerfluhaters weak difficultshotsfault droppednooo numbersvistateamsnetworkinggrownbootsupswirelessclearlyspainupcomingmeatmmmmplazapurchaseasianinterestbreathmp3playinjazzricerightsprefersolarcatsfemalesinvitelyrics musicalxmashonestbabieswinnersprizesmiles hates worriedupsetban lonely idiotlack gossipfallsdelete rentridiculoustrial storesallowcustomersunionconversationwoodsadultturkeysafetypackagemenufacesvalleygodspaintdressedhiringmountaincalmstreamdatesamericanswingslipssistersyouthmarriagepartnerplaysjusticedrinkssolutionphotographyinspirationbath presentswishesmealcookies incomecheersbrightrelaxkisses arrested crimeaccidentcrisislyingsurgerymistakeimpossiblenervous ignore expensivepiss oopsoffline recordssuggestionsbringingleadsamsterdamboat strategyvintageexerciseskillsteachersimprovedrawchipsemailsartsleadingproductivegallerydonatehorsesingeruniquememorieseasilyyogafruitbenefitsdollars genius comedy failed mistakeshorror warningaddictedemptycuts ended deliveryvotingrare surveyt−shirtpremierridingsexualchieffullytacohotelspublishedimagescharityanswersrecipepremieresoftcouponupgradegrowthexpertdiscoverfancygoalscorrectpowerfuljourneywisehelpedinspiredfabulouscaresmommygifts jokespleasure kills devil blinddrugsbills insane awkward mortgageloosestrike productionliverpoolfavelaunchedownerfabmidnightchancesitalyoptionschefbuilt teachingperformlegendmoralrecoverypotentialcultureunclebananaeasterreliefsavedspabirdsmarrysaintscomicusefulknowledgeheartssweetieadorablediamondbonus References murder failure fireddangerousnegative freezingdentist hittinggrosssosalarmargh meaningfinaleeggscasinomomentsnikemodelsrobotdebuttwinjacketmirroruniversalprovidewillingitaliannewestloungerisingbuttergoodmorninghighlyvisionmuseumanimalssharescreatingeffectivebedroomprotectmassagepianobenefitsavingprincessshinecureprogressfoodspassiongoodnessloversfriendlyprofit day of week ouch bite attendbff suite cookie strength steal guiltyfeverawful openedresultfundfactsmindsfrifarmofferingboostsolutionsdesirewishing loser burning complainscream beast punch tunedpurposesigningsaucereachedwoodslimpremiumchristwebsitesbreatheparaderoflresourcesforgiveperformingfestswimmingprayeramorepridehomesconfirmedcruisewoohoopastaprayingplentybirthfiestamillionsfilmslovermotherspartiescheerpuppyb−day virus slap niggazemergencygunsregretcoughdeletedcreepyshockghostexamscostspakistan pubthousandskindleshirtsbalancebandsactorbreastasiadegreegrillorganicactivitywritersfirefoxcurioustennishillsbiblefamedrawingrelationshipsindependentuniversegracediscoveredtheateroceantreescolorstalentedtoys thanksgivinggrandmaflowerssmiling victory cheatworriesstoletrash roughmadnessnoise closing virginracingblondecdsthaisurelyvillagereadergmailbeatlesprimehire zooplantpaintingcokeactivetoastwisdom adventurepal celebrating starvingunfortunatelycancelled tear zombie loansgripe religioncertainlycouchkeyboardvotessnowingawarenesshubbydesignscostumegradmonkeyberrylargestleadershipworldwideinvitedadvanceclothingkittytheatremommacompletedswimprayersidealhopeslovin laughs terrordisaster assholehating wasted kicked deadlinemessed bum survivebass affordculturaldecenthollandbabesswinggamingsupportingcrystalfriescaptainvisitingbucks jewelrydadscomputersdarlingpreciousopportunitiesrelaxing suicide earthquakedisappointednightmareannoyedtaxes illegalscreaming whipdammitrefuse blockedremove ughh programsbeansacademymarketscluedownloadsbrowserlistenedaustralianvirginiasoundtrackflaggr8patienceyoungershorefamiliarpaysraisedunlimitedperfumepersonalitysuprememotivationrescueabilitynovelpreparedunderstandingadvantage thankfulangels −4 violenceprison sadly grave haterhangoverbadly desperatewhorenuclearobsessed locked essentialreplieddollsuperiorwesternplaylistfishingjoiningnursetributedesignednewspaperrevealedgrant raisesdedicatedhighestrecipesalbumshawaiiresortheroesimpressedluxury funniestcelebration abuseinjury damageshitty falsejerk afghanistandire anti hardestcharges exacthousingvehiclegreekmapsgrocerybubblehaircutdimegrammydownloadedlifetimeremembereddesignersdiscoverysangcoloursteakgreaterrockinace liking cherrypartners gloryfamilies flower rainbow criminaldisease anger panic banned forgottenforced delayghetto baskettopsherselffundstournamentduck80sbatmankarmaproducedarriveactresswavestwinsrockingstunningcousinsgoodnightadvancedexpectingauntforestshared pearlstrongerconfidencechampionshipcomfortablegratefulpraiseromanticpromotionfriendship racisthomelesscheatingfloodattacksdisgustingruincomplaininglieddislike envy fee owe interactivefurniturebrazilianlegitofferedincludeddownloadingkingdomsightfinishingenterprisedeliverheadphonesdvdsexpertscloudsamericaselectronicstudiescommunicationnewlyencorecamerastelevisionfocusedsmooth olympicsprizes roses fails cried threatarrest mafia protestkicking hardly attendingchoicespeanutcoat steadycowboyscontactsproperlysodaolympiccastlequeensringsdeliveredacoustic shopsexperiencedreuniongraduateappreciatedstrawberrydreaminghelpfulwonderlandpleasedcouragechampionweekendscupcakes laughed funeralpainfulinjuredtheft beatingretarded suspectcriticalliarcomplicatedwickedkillin shocked iraq specialistsoapchilicricketreaderspacificlemonvictorprovidesproducedevelopallowscoolerglobevolunteersaintscoredgrassprotectionscotlandseasonssonsnecklace cutiecartoonsavings comforttastymiracle hated uselessstolen accused trapfallen patients junk sponsorpilotinnmusclejumpingenvironmentspicelionquestconnectedeatenpotatodawntacossolvedressescorncourtesycoolestpancakesbraveearned fortuneinspiringwealthattractivehandsome kissing cruel divorcehurtingrecessiondestroyrageenemyaddictionstressed idiots hungerstrugglemonsters lawyerbald membershipvibefreelancemastershousestrailcerealbowlingoooohearringshandyclubsbelletasteselectronicswinksingsacceptedphotographertunespromised bunnyoutdoorclever petsperfectly sweetheart rape depressionsuffering danger bitter screwedfools ignorant dustrevengedirtpimp suppliesfridgerecognizeequalelvisstrategiesemotionspolishansweredscientists invest intelligencehomemadeentertainingcutest depressed ruinedassault sucked fees terminal perspectivemelbourneexpandhintchemistrytowarddish pennypromisesneighborhooddocumentaryinnocentresourcemintsurfinventedproposalbakewonderscreationchildhoodcomicspillowsecurevanillaimpressiveshiningfridayslibertyvalentineblessing blessingssucceedparadise suffer victim cursepathetic crappyscareexhaustedargumentspitdelayed slut slip automaticdevelopingpresencecheaperkaraokefictionrealiseswissevolutionsupportssoonerraising mercedesnephewpowersrevenuetravelingnaturallyfathers adore achievecupcakeprofitsachievement depressing victims miserablehurricanestrugglingfears argue confusing bugs missesminuszombies nationscreatesnativeadultstraditionalovenelizabetheagleslivin worshiprumbrunchbroadwayactivitieseatsclaprestaurantsguaranteedpopcornvitaminseas fairyplants fraud fightsratbeg wasting gangstadelays operationbotheredfist ruins automaticallyoregonfixingcreeknursingguestsengagedresolutionwheelsupdatingknightmorningsrocketpoetrypantieshatscouplesbouncesprings niecehorses confidentsnacksunsetapprovedcaringsundays merry enemiesstealingbastard losesburnsfrustrateddeadly garbageyuck snake fliesdisagreedarknessaccidentally pregnancyallianceinstalleddeservedneighbortemplevaginatrainsbreathingtechnologiesvisitorsdestinationlotterytouchedrecommendedbakedgraphicspumpkinincrediblycookedchattingclassyhumansruby healingparentinspire intelligent cakes attackedselfish burnedignorancesuspended bustedchaos messyremovedthirsty writes personssausageatlanticcottonjellyfarmersenvironmentalcraftwhalegoodscleanedexhibition varietybelievesgainsbreezetripsawsomerabbitfireworksparks magical excitement 8 fatal failing anxietyyelling bulletcage forgettingexcuseshatin oliverainsvocalsofaskirtdestinyseedeaglewwwembraceagreementselectedinstantly maturechampvolunteerscrownmercytoursspiritualtreats superstarimproved superman hooray earnings kissed violent losers missin lightingmagazinescitizensrememberingimportance growsease brainsvalid platinumflavorconcertsromecharm brutal canceleddull burnt offense limits transportinvestingcertificatevisitedemploymentresolvewaterspurchasedparticipatevisitsreceptionhumblerelevantdiningtastingshowers shinychampagnemedalhealimaginationnoble genuinegardensreward successfullydiamondsfinest laughter unemployment severecombat mob stereojennylawnstarringgrantedleafrockedsupportersproteinstylesconversationsactorsjamaicaexploreceremonyroastbellsbakinggreet pies carnivalhonoredamazed treasure killersslave thug lowest poop yell slam owns recommendationsupsideappropriateprofessionalsrises increaseddewsavesfuturesrides skillcurrencypalacesignificantimprovementguaranteeworthyproductivity bridedaughterssaturdays torture ashamedache infection defeat deny retire shadegeneratethemesexperiencesrelationspartnershipseedswonderedgainedbakerblanketshelterislandsskatingattractneat shakespeare juicy butterfly poverty cigarette cigarettespoliticians obsession heating combinationprovidinginformedfeedsproposedpracticalcowboylogicintroducedspotlightassistgatheringdeterminedpotatoesdonationsunderstoodhavenapprovecandle encourageautumnsafely valuableaccomplisheddessertcuddle pleasantloyal poison cheated offensivetrapped ignoredsyndrometornadodiss rival sink christiansgrooveelectricityoakinsighttaleegyptswedencommitmentpaintedfeeding backyarddrumsreceivingpoemdiplomatouchinghired dancerdonatedcorrectlyapproval appreciationindependenceblissluckily glorious sweetest destroyed injuriesdisableddevilslawsuit warned witch cracked cambridgeextendalumnibelievingrestore vehiclesfairlydevelopedcommittedmaximumlambphilosophyinitiativesupportedrecoversparks mountainsimprovinginvitationreliableloyalty kindness attraction crashedstrokemisery dumpexplosion deficitscandal weapon riskscreep introductionrecoveringreceivesgarlicdictionaryprovided increasesportraitscientificvelvetperformancescircus surprisingvaluessurprisinglyefficientsurvivedgardeningearning perfection faithful outstanding unhappy bleedingaddict annoy suckersymptoms relateshouldersensurebedsvenusathletesreputationlipstickperformedarchitectureincreasingadoption believed skiescomfy weapons penalty greedyambulanceagingstink runawaybeggingdisappear armed beliefsheepexpressionapplyingauthorsconnectionssparkexpansionbelongsdemocracykineducationalpossibilitytalessleeps humanitycolours salarymusicianssatisfieddivine honesty deaths abortionbastards cries arguingdeclinefoughtpity confusionfuckerunfair acid dim riderbeerscheekallowingscientistdrawshorizonaimsdemocraticarchitectadopt musclesaccuratecostumesorchestracapable dove belovedpeacefulachievedmillionaire terrorist crimes denied smashedstruck preventionclimbingchoir neon overcomecoinorganizedheritagewishedromeomuffin wildlifescholarshipsunrisemamma fruits kitten rejectedrobbed cocainedeaf buriedweaknesscomplaintsunable damned tornoppositionhospitals cracks newspapersoverseastransportationengageindianswheatexistsdepositfloating progressiveimpresssingersacceptableponygrants greatlyabroadmusician eternal awarded harm errorsbreakdownriot conflict conspiracy pale alleged coloredreflectionmutualtourist pioneerarrivalacquiresketch unitegrin dynamic glow raped bankruptcytragedy threatssinsfoul anxiouswreck foolish dizzy principlesconstitutioncheeks exercisesethics lime mentorqualified gentlebloom tsunami bleed disorder choke unlikely stranded goddamn harbor collectiveemergingreasonablelanguagesprospecteden teachescompassionharmonyweddings dances butterflies sicknesspunishment complaint shoutingparanoid swinecrowdedunnecessaryfarewell skull engineersbridgesintegratedelectedbentleyentersestablishedstable collegesinfiniterays landscapedancersmansionpopularitygentlemen relaxedpromoted murdered corruption psychobeaten bombs jealousy rats insanity recklesseliminatecritics demands recognizedchallengingbuyerspiritswinescolleaguestouchescontributeliteratureassetcommunicatepromising gentlemantrusted passionate extraordinary delight robbery satan thief lynch whack suspected creeping preparationfloat nurses doughlifteduptownleapsalvationtraditiontenderathleticrecognitionsymphonyprotected switzerlandunderstandssilk brightergraduatedsatisfaction grandmother terrorism tragic nightmares drowntension regrets invasiontroubles controversyfewersucka defensive essencerevolutionaryjingleregardsatmosphereequallyinstrumental candlesfountainsupporterdreamedvoyageelegant summertimegenerous optimistic disrespectunfortunatedemon criticismdoomslappedcryin screams aggressiveartificial brew wavingstronglyphd classicalvocalsrhythmdine rivers librarieseducated miracles illness inflation filthy insecure gangster rot disappearedstab haunted downs acquiredtouristsspringfieldmellowpensionprotecting remedyguidanceassets forgiveness terrific sadnessmotherfucker destructionhatred losses threatened collapse suspiciouscrashingopposedrefusedstingdrunken bailoutbattleschronic prospectsporchpracticallyexclusivelyproceedsoccasionreflectchess softly dedicationconvenient cozy attracted whores denial advantagesefficiencypossibilitiessensation sailingfavourunity feast abandoned arguments savage brat baddestghosts principle boulevardcopper effectivelyriversidesanity rhymeseligibledoggy photographsacceptancedaylightsail equalitywealthy enthusiasm celebrated frustration rotten burden bury convicted swollenabsencesnakessentencednothings killa mineraltravelerssentimentalchickenstastedchorusinterests preferred integrity paintings melody suffered damaged graves capturedexpenses tobaccocrushed aircraftvisible patientlyquickercontributionsinstrumentprovedstreamsfarmsresolvedsungresidenceboatscompensation invention photographintellectual encouraging heartbreak woundednigger guiltthugs dread protesters participants consciousyachtcropsandsmagnetic scoring relative novelsfarmeroutdoors riches grief defeated doubtsscars integralinstrumentsinvestmentsromansmirrorsv−day sailor functional recoveredtheatershallelujah caredvalued shines smiled drowning screamed shatteredcrudedemandinghangs beings studied aboardedisonflowingscent wage desires heavens thieves worthless faggot dispute lacking vainslippedvoodoo racial concepts whistlecheddar balancedpoeteager infantearns championshipsgentlyheavenlytriumph honour hopelessdemons bullets vulnerable fooledtore madamenomination sciencesgravy literaryinteractionrelativessurvivorsheartbeat dearest pollution suspension mistaken darker outlets aviation drawingsrestoredambitiousuniversities oxygenmanners artistic difficulties restrictions bore numerous clownshopedexistedfarmingherb intimate satisfy affection slavery disturbed tearingvoid restless survivingarchivesincreasinglyparticipation pearls thrill remarkablegrandfathercherish sunlight wound scar critic towns dynamicsproposalsethicallegitimate relation conservation rhyme eternitydetermination dancedprosperity rejection suspects dusty nearest providenceacknowledgeobserve imaginedownershipcontribution suitableprivilegeguitars sincere companion abused arrestswounds achingwoe forbiddendifficultyraging oppose absent beliefs fluid enabletransformation virtue improvementssparkleoceansleisurewarmth sorrow bombing execution crookedwithdrawalrustyfragile precisesensesprofession logicalgenerationspoems amour secured dignity employed diseases tumor shotgun shove expense trinityrailway illustrated believer cosmic proceed qualities 7 discrimination greed despisepistol wrath snatch revival illustrationbeam civilization sweetergladly iraqi landmark consciousness traveler radiation allegations consciencelikewise commonwealth encouraged slaves sinking woof produces definite firefighters museums fulfill troubled cursed fury foe stomp opponentslipping approachesacres observationfoundedbedtimeprayedcurves expressionssimplicitysculpture forgiven prisonercemetery inferior preferencerighteous preserveshoresnephewsagricultureelder visions sweetness damagesvicious screamin stain bacteria gutter brooksolving automobile metropolitan immortal stability disgrace incorrect injectioncondolencesmisunderstood strain sanctions avoided significantly behold hearted feather strollcharms performers prisoners neglectedchoking phony mold roughly glanceconsensusforthcoming developmentsalmightyvolumes poets specialty respected rainbows bruised dismissed frown separation sophisticatedcontinuousadvances foxyrevenuessixteen tremendousmankindsavior accuracy glowing missile uncertainty denying advocates illustrations dreamer devoted moonlight violationdespair spite darkest sights extending rational splendid motherfuckers prejudice suckers branches merit slaughter bruise muthafuckaurine snitch fireplacecontributedconsiderationcathedral nieces x−mas demise havoc disorders flee theories phenomenon freely violatedloneliness muthafuckin feared opponents hustler diagnosis audiencesmendseventeen innocencespontaneousfurnished completion holocaust seized baghdad foundations jewels agony separated stains resigned crimson restoration contributingpainterwages groovy declined sinner thirstuncertain transformedoccasionsprofound altogethershopper neglect violations restrictedhaunt executed composition lyrical dearly everlasting fantasies infections complained resignation litigation nixon representation treaty climbedvoluntaryenabled traditions cooperation widowsleeplesscoffin inmates sirens plead advocacy experiments justifiedcradletraveled drowned motherfucking hostile hauntingisolated erased yields purely vengeance blinded monetaryattachmentsignificanceshells avidprominent composer respects ecstasy frightened suspicion limitations punks negotiatethronecrops reasoningclearer functioning corpsebetrayed tomb outdated continentdimesoutcomesnarrativeassured desired effectiveness penalties fades metalsprospectivegoodman poetic princescultures salaries weep refugees misspelled strapped cattle surplusemploy antiques velocityroam calcium allies praised −5 tortured heartache slug quantities arise sufficient therapeutic unconscious composed architecturaldevotion scholar heartbreaker lawsuits abnormal pianist succeeded killings desperation scarred eliminateddeception villages stained influences preservation raindrops favorable distress fiend servants shiver reindeer lullaby thorn centuries screenplay grandson mourn impose 1970scomprehend honorary scholars argues upward expedition1980s depositslifelong pimps carriage1960s distinction starry conflicts gangstashelpless neighborhoodsinteractions obtained favored liberation condemned fright complications objectives tensions accusations reasonably idealsheartfelt jukebox decay associations philosophical infants cowards weeping unclear agreements sails saddened isolation distinctgaze saviour prevail defect dagger starlight rotting gloom reluctant maintainedagricultural generosity burial sinners casket slumdog acknowledged technological greeks competent abundant peacefully inadequate meanings allright victories normalized word frequency missiles lonesome cavity deficiency telecommunications majesty elevated motherfuckin argued distinguishedguiding nearerwritings eighteen rudolph liberties criticized incorrectly combinations accompanied continuity discontinued pressures indictedhussein begged shatter respective remorse emptiness deceive stimulation bookstores obituary turmoil confrontationexcluded successor conceived certainty forests 6 jaded varieties accuratelycooperative grandchildren sorrows bleeds unholy foes rags applicable justification sympathetic untruetaxation novelist pleaded historian successionaestheticdistinctive supermarkets virtues thrills prosecution forsaken admiration fists betray desirable adaptedunwind respectively belonging adored decreased privileges accompanying bled thorns initiated sufficiently eternally caress avg confined chained considerations proceededtreasurer succeeding saddam defendant exercised proteins defects sorely artillery readily belonged noose tumors fled acquainted considerableassembled subjected equilibrium cherished mourns attained fiends slugs dividends trembling granddaughter sailed

servings arises 10 h teardrops ceased preserved adoring satisfactory

tenderness comforted unkind

honky

5 indictment

tremble darkened log −6 4

T W T F S S M T W T F S S M Hour 13, 2010 day of week −7 −4 −2 0 2 4 55 of 83 Average happiness havg - 5

Complex Complex Tref: Tuesdays (havg=6.03) Sociotechnical Sociotechnical Tcomp: Saturdays (havg=6.06) 0.4 Systems A Systems 1 love +↑ no −↓ 0.2 haha +↑ party +↑ 5 fun +↑ Measuring Measuring saturday +↑ 0 +↓new Happiness Happiness ↑ weekend + Some motivation Some motivation not −↓ 10 happy +↑ Measuring emotional −0.2 Measuring emotional dont −↓ content content ↑ − last Data sets Data sets hahaha +↑ Happy :) :( Sad −↑bored −0.4 Tea Party ! Afghanistan 15 −↑drunk Analysis (amb) avg Analysis live +↑ h die −↓ Songs Songs friends +↑ Blogs −0.6 Blogs game +↑ 20 con −↓ Tweets Tweets movie +↑ cant −↓ Positivity Bias −0.8 Positivity Bias r −↑fight birthday +↑ 25 +↓google References References great +↑ −1 sunday +↑ family +↑ Word rank beautiful +↑ 30 beach +↑ −1.2 home +↑ 09/09/08 12/01/08 03/01/09 06/01/09 09/01/09 +↓lunch date sick −↓ shopping +↑ B 35 playing +↑ −1 −↑don’t amazing +↑ Text size: Tref Tcomp 100 bad −↓ −2 awesome +↑ ↓ 40 1 homework − 10 wedding +↑ −3 −↑hangover rel freq

Balance: 10 102 −↑miss +↓free −87 : +187 −4 ↓ +↓ +↑ log 45 3 shit − 10 court −↓ nice +↑ −5 4 ↑ 09/09/08 12/01/08 03/01/09 06/01/09 09/01/09 10 won + 0 100 +↓school ↑ ↓ r − − date 50 Pi=1 δhavg,i movies +↑ 56 of 83 59 of 83 −10 −5 0 5 10

Per word average happiness shift δhavg,r (%)

Complex

The daily unravelling of the human mind: A Tiger Woods B BP Sociotechnical 0.2 0.2 Systems 0 0 −0.2 −0.2 (amb) avg (amb) avg −0.4 −0.4 h h −0.6 −0.6 Measuring −0.8 −0.8 Happiness −1 −1 6.15 10/01/09 12/01/09 02/01/10 04/01/10 06/01/10 01/01/10 03/01/10 05/01/10 07/01/10 0.06 Some motivation 2009−05−21 to 2010−12−31: date date Measuring emotional 6.1 0.05 content −3 −3 0.04 Data sets rel freq rel freq avg −4 −4 10 10 h 6.05 0.03 Analysis log log count (%) 0.02 −5 −5 Songs 09/01/09 12/01/09 03/01/10 05/31/10 12/01/09 03/01/10 05/31/10 6 Blogs 0.01 date date Tweets 0 4 8 12 16 20 24 4 8 12 16 20 24 0 T : All Tweets (h =6.06) T : All Tweets (h =6.04) 0 2 4 6 8 10 12 14 16 18 20 22 24 C ref avg D ref avg hour of day (local time) hour of day (local time) Tcomp: Tiger Woods (havg=5.74) Tcomp:BP(havg=5.57) Positivity Bias 6.15 1 −↑accident 1 −↑disaster References 2010 −↑crash no −↓

avg −↑injured −↑down h 6.1 no −↓ −↑shut 5 −↑scandal 5 −↑kill +↓love +↓love 6.05 −↑hospital −↑fake −↑cheating −↑stop ↑ ↓ r wife + r + me 6 10 −↑alleged 10 +↓haha car +↑ Text size: −↑not Text size: +↓me +↓lol 0 Tre f Tc omp 0 Tre f Tc omp 5.95 10 santa +↑ 10 ill −↓

Average happiness −↑cheated +↓hahaha Monday Tuesday Wednesday Thursday Friday Saturday Sunday Monday 15 1 ill −↓ 15 1 dont −↓ 10 10 5.9 −↑cheat −↑died Word rank Word rank midnight noon midnight noon midnight noon midnight noon midnight noon midnight noon midnight noon midnight noon midnight sex +↑ Balance: billion +↑ Balance: 2 2 10 +↓haha 10 con −↓ Local time ↑ −188 : +88 ↑ −167 : +67 − alone +↓ +↑ − blame +↓ +↑ 20 3 −↑not 20 3 −↑criminal 10 +↓happy 10 +↓happy −↑breaking −↑damage 4 4 10 −↑divorce 10 −↑costs −100 0 movie +↑ ↑ ↓ −100 0 −↑fails ↑ ↓ r ↓ − − r ↑ − − 25 Pi =1δh av g, i + good 25 Pi =1δh av g, i − crisis −20 0 20 −20 −10 0 10 20 60 of 83

Per word average happiness shift δhav g, r (%) Per word average happiness shift δhav g, r (%) Complex Complex

A Pope B Israel Sociotechnical Happiness in Manhattan (just for fun): Sociotechnical 0.2 0.2 Systems Systems 0 0 −0.2 −0.2 (amb) avg (amb) avg −0.4 −0.4 h h −0.6 −0.6 Measuring Measuring −0.8 −0.8 Happiness Happiness −1 −1 02/01/10 04/01/10 06/01/10 08/01/10 11/01/08 01/01/09 03/01/09 05/01/09 Some motivation Some motivation date date Measuring emotional Measuring emotional content content −3 −3 Data sets Data sets rel freq rel freq −4 −4 10 10 Analysis Analysis log log −5 −5 Songs Songs 03/01/10 05/31/10 09/09/08 12/01/08 03/01/09 date date Blogs Blogs Tweets Tweets T : All Tweets (h =6.04) T : All Tweets (h =6.05) C ref avg D ref avg Tcomp: Pope (havg=5.40) Tcomp: Israel (havg=5.30) Positivity Bias Positivity Bias 1 −↑abuse 1 −↑war References References −↑scandal −↑fired sex +↑ −↑against −↑victims −↑killed 5 −↑arrest 5 −↑attacks no −↓ −↑attack −↑against −↑offensive +↓love −↑fire ↑ ↑ r − arrested r − crimes 10 child +↑ 10 peace +↑ −↑not Text size: −↑no Text size: −↑resignation −↑not 0 Tre f Tc omp 0 Tre f Tc omp 10 +↓me 10 −↑die +↓haha −↑killing 15 1 −↑crisis 15 1 −↑stop 10 +↓lol 10 +↓love Word rank Word rank dont −↓ Balance: −↑bombs Balance: 2 2 10 −↑crimes 10 −↑fighting ↑ −157 : +57 ↑ −131 : +31 − attacks +↓ +↑ − kill +↓ +↑ 20 3 −↑accused 20 3 −↑kills 10 −↑deaf 10 −↑conflict See Blog post on onehappybird ( ) −↑down −↑death
4 4 10 +↓hahaha 10 −↑terrorist −100 0 −↑evil ↑ ↓ −100 0−↑weapons ↑ ↓ r ↑ − − r ↑ − − 25 Pi =1δh av g, i − rape 25 Pi =1δh av g,− i bombing −50 0 50 −10 0 10 61 of 83 64 of 83

Per word average happiness shift δhav g, r (%) Per word average happiness shift δhav g, r (%)

13

(amb) (amb) Complex Word havg Total Tweets Total ANEW Word havg Total Tweets Total ANEW 1. love +1.42 46,687,476 (6) 85,269,499 (5) 51. me -0.06 144,342,098 (4) 88,088,051 (4) Twitter—location: Sociotechnical 2. happy +1.32 16,541,968 (13) 32,442,529 (8) 52. ? -0.07 2,333,283 (53) 674,679 (69) Systems 3. win +1.26 7,981,856 (26) 14,640,728 (20) 53. commute -0.09 90,126 (94) 90,092 (92) 4. kiss +1.21 1,697,405 (59) 3,162,330 (48) 54. gay -0.09 2,727,309 (47) 1,697,177 (57) 5. cash +1.21 1,279,236 (63) 2,468,496 (51) 55. right -0.10 19,166,480 (10) 15,850,283 (19) 6. vacation +1.11 934,501 (67) 1,783,270 (56) 56. school -0.11 9,264,217 (24) 6,924,193 (34) 7. Christmas +1.03 4,887,968 (35) 10,645,630 (25) 57. Republican -0.13 229,773 (86) 188,338 (85) Measuring 8. God +0.95 8,576,364 (25) 17,867,768 (16) 58. they -0.16 27,442,360 (8) 27,150,189 (11) Happiness 9. party +0.93 6,438,886 (29) 12,090,597 (23) 59. winter -0.19 1,255,945 (64) 1,217,225 (64) 10. sex +0.89 3,551,767 (39) 7,087,972 (31) 60. lose -0.19 2,056,468 (55) 2,091,540 (53) Some motivation 11. Valentine +0.85 247,288 (84) 464,914 (75) 61. Jon Stewart -0.20 52,084 (97) 33,086 (96) Measuring emotional 12. family +0.79 5,014,816 (32) 10,629,361 (26) 62. gas -0.22 1,022,879 (65) 812,029 (68) content 13. sun +0.65 2,385,348 (52) 4,602,627 (44) 63. no -0.22 95,129,093 (5) 38,894,616 (6) Data sets 14. life +0.50 14,006,454 (17) 27,770,768 (10) 64. Democrat -0.23 93,193 (93) 75,450 (93) 15. hope +0.48 11,833,337 (18) 22,952,366 (13) 65. left -0.27 4,893,634 (34) 4,611,878 (43) 16. heaven +0.43 741,878 (71) 1,485,702 (59) 66. Senate -0.29 447,732 (78) 316,835 (80) Analysis 17. :) +0.42 10,470,483 (20) 6,787,678 (35) 67. election -0.30 560,184 (75) 375,055 (78) Songs 18. income +0.36 510,425 (76) 418,161 (77) 68. Sarah Palin -0.34 225,577 (87) 150,096 (88) Blogs 19. friends +0.33 7,669,719 (27) 7,541,106 (29) 69. Obama -0.35 2,981,150 (44) 1,998,326 (54) 20. snow +0.32 2,596,165 (49) 5,011,785 (40) 70. economy -0.36 608,878 (73) 460,834 (76) Tweets 21. :-) +0.32 1,680,165 (60) 1,102,512 (67) 71. Congress -0.36 391,510 (79) 279,695 (81) 22. night +0.29 17,089,505 (12) 17,606,796 (17) 72. drugs -0.39 509,606 (77) 469,091 (74) Positivity Bias 23. vegan +0.28 183,889 (90) 178,676 (86) 73. Muslim -0.42 215,300 (88) 146,506 (89) 24. Jesus +0.27 2,027,720 (56) 1,673,992 (58) 74. George Bush -0.43 32,341 (98) 23,102 (98) 25. girl +0.25 10,070,132 (22) 19,886,691 (14) 75. climate -0.44 364,177 (80) 229,129 (83) References 26. USA +0.23 2,157,172 (54) 1,204,585 (65) 76. Pope -0.51 152,320 (91) 135,955 (90) 27. you +0.22 173,276,993 (3) 145,464,084 (2) 77. oil -0.53 1,377,355 (62) 1,148,990 (66) 28. our +0.21 14,062,465 (16) 14,437,899 (21) 78. I feel -0.54 5,173,513 (31) 4,702,352 (42) 29. ;) +0.20 2,618,940 (48) 1,475,221 (60) 79. Glenn Beck -0.54 113,991 (92) 101,090 (91) 30. health +0.20 2,575,543 (50) 4,950,202 (41) 80. Islam -0.54 187,223 (89) 70,311 (94) 31. tomorrow +0.20 10,379,637 (21) 8,899,406 (28) 81. :-( -0.65 341,141 (81) 244,215 (82) 32. ! +0.16 3,463,257 (40) 1,385,072 (62) 82. :( -0.70 2,907,145 (45) 1,891,225 (55) 33. summer +0.13 2,998,785 (43) 2,554,459 (50) 83. flu -0.75 901,403 (68) 639,000 (70) 34. we +0.13 39,132,934 (7) 34,513,587 (7) 84. rain -0.78 3,233,464 (41) 5,959,903 (38) 35. today +0.13 25,588,506 (9) 23,619,518 (12) 85. BP -0.78 582,167 (74) 326,100 (79) 36. man +0.12 15,856,341 (14) 29,558,118 (9) 86. mosque -0.79 69,812 (95) 46,736 (95) 37. woman +0.10 2,543,036 (51) 5,603,347 (39) 87. dark -0.95 1,577,553 (61) 3,233,911 (47) 38. Stephen Colbert +0.10 23,778 (99) 14,697 (99) 88. Lehman Brothers -1.08 8,500 (100) 4,280 (100) 39. ;-) +0.10 943,413 (66) 516,171 (73) 89. Goldman Sachs -1.08 52,703 (96) 30,769 (97) 40. RT +0.06 339,055,724 (1) 142,219,359 (3) 90. Afghanistan -1.15 273,519 (83) 172,637 (87) 41. coffee +0.04 2,800,972 (46) 2,399,867 (52) 91. Iraq -1.37 238,931 (85) 213,425 (84) 42. church +0.03 1,812,251 (58) 3,452,171 (45) 92. cold -1.39 3,670,447 (36) 7,015,518 (32) 43. work +0.02 18,415,618 (11) 16,191,802 (18) 93. gun -1.81 680,903 (72) 1,263,217 (63) 44. I +0.02 307,960,343 (2) 282,865,043 (1) 94. hate -2.43 9,652,881 (23) 18,158,870 (15) 45. yes +0.02 11,593,356 (19) 7,499,840 (30) 95. hell -2.49 6,266,162 (30) 11,056,735 (24) 46. them 0.00 15,352,295 (15) 14,398,889 (22) 96. sick -2.55 3,576,058 (37) 6,783,395 (36) 47. hot -0.01 7,122,144 (28) 6,286,163 (37) 97. sad -2.56 3,563,745 (38) 6,951,686 (33) 48. boy -0.01 4,933,333 (33) 9,670,512 (27) 98. war -2.63 1,955,901 (57) 3,417,588 (46) 65 of 83 49. yesterday -0.01 3,077,761 (42) 2,852,623 (49) 99. depressed -2.64 280,872 (82) 541,394 (72) 50. Michael Jackson -0.02 825,979 (70) 571,442 (71) 100. headache -2.83 856,600 (69) 1,446,064 (61)

(amb) TABLE III: A selection of 100 keywords and text elements ordered by average ambient happiness havg . We define ambient happiness of a keyword as the average happiness of tweets containing that keyword, relative to the overall happiness of tweets, havg 6.37. The number of tweets and total number of ANEW study words are listed in the third and fourth columns, with Complex Complex ￿ the ranking of the keyword according to these quantities shown in brackets. Note that all pattern matches with tweets were Sociotechnical Sociotechnical case-insensitive. Twitter—location: Systems Systems

invite the reader to explore the tables beyond the obser- three top ranked words are ‘love’ (havg=+1.42), ‘hap- vations we detail here. We begin with the highest py’ (havg=+1.32), and ‘win’ (havg=+1.22), and the Tref: NY (havg=6.32) (amb) and lowest rankings of ambient happiness h , for last five, in reverse order, are ‘headache’ (havg= 2.55), avg − Measuring Tcomp: CA (havg=6.38) Measuring our list, finding them to be reassuring sensible. The ‘depressed’ (havg= 2.56), ‘war’ (havg= 2.63), ‘sad’ − − Happiness Happiness 1 mad −↓ Text element and context correlate in happiness Some motivation ↑ Some motivation Measuring emotional fun + Measuring emotional scores: content beach +↑ content Data sets home +↑ Data sets 5 hell −↓ I Compare ambient happiness with text element Analysis dead −↓ Analysis Songs free +↑ Songs happiness. Blogs hate −↓ Blogs Tweets +↓wit Tweets ↑ I Spearman correlation coefficient: Positivity Bias r 10 car + Positivity Bias −10 +↓snow Text size: r ' 0.79, p-value < 10 . −↑time s References 0 Tref Tcomp References 10 cold −↓ An on-average result: says nothing about any −↑rain I 15 fall −↓ 1 ↓ individual sentence. Word rank 10 + music +↓sex Balance: bus −↓ I Extra random piece: stemming fails. ↓ −79 : +179 2 cut − +↓ +↑ 20 10 −↑bomb −↑ambulance square −↓ 3 10 hit −↓ 0 100 accident −↓ r −↑ −↓ 25 Pi=1 δhavg,i −↑fire −20 −10 0 10 20

Per word average happiness shift δhavg,r (%) 63 of 83 66 of 83 Complex Complex Twitter—popularity based on follower count: Sociotechnical Twitter—interactions: Sociotechnical Systems Systems

09/09/08 − 06/30/09 09/09/08 − 06/30/09 9 10 10 10 Measuring Measuring 8 10 Happiness Happiness Some motivation Some motivation 7 8 10 10 Measuring emotional Measuring emotional content content

6 # Words # ANEW 10 Data sets Data sets

6 5 10 10 0 1 2 3 4 5 0 1 2 3 4 5 Analysis Analysis 10 10 10 10 10 10 10 10 10 10 10 10 Songs Songs Blogs Blogs 6.55 Tweets Tweets 60 ) Positivity Bias Positivity Bias 6.5 40 )

(amb) S References References N avg 6.45 20 h

6.4 0

−20 6.35

Happiness ( −40 6.3

Ambient Diversity ( −60 Background: 336.96

6.25 0 1 2 3 4 5 −80 0 1 2 3 4 5 10 10 10 10 10 10 10 10 10 10 10 10 Follower Count Follower Count I Decay in happiness correlation in social network.

I ρ = Spearman’s correlation coefficient. I Dunbar’s number ' 150. 67 of 83 70 of 83

Complex 2 Tref: ≤ 10 followers (havg=6.29) Positive bias in the English language: Sociotechnical 3 Systems Tcomp: ≥ 10 followers (havg=6.44)

1 free +↑ +↓home 0.15 Measuring hate −↓ Happiness sick −↓ Some motivation 5 bored −↓ 0.125 Measuring emotional ↑ content good + Data sets +↓bed people +↑ Analysis hell −↓ 0.1 Songs Blogs r 10 stupid −↓ Tweets love +↑ ↓ Text size: Positivity Bias cold − N 0.075 0 10 +↓sleep Tre f Tc omp References social +↑ 15 success +↑

Word rank 0.05 1 ↓ 10 sad − +↓god money +↑ Balance: +↓christmas −93 : +193 2 0.025 20 10 headache +−↓↓ +↑ −↑news lost −↓ 3 0 10 hungry −↓ 0 100 ↓ −↑ −↓ 1 2 3 4 5 6 7 8 9 r war − 25 Pi =1δh av g, i −↑failure h avg −20 −10 0 10 20

Per word average happiness shift δhav g, r (%) 71 of 83

Complex 0.16 0.16 Sociotechnical 100 100 Systems 75 A. Twitter 75 B. Books 50 50 14000 25 25 6.6 0.12 percentile 0.12 percentile 0 0 1 3 5 7 9 1 3 5 7 9 Measuring havg havg Happiness 6.5 12000 100 100 Some motivation positive positive 0.08 75 0.08 75 Measuring emotional 50 50 content negative 25 25 negative Data sets

6.4 10000 percentage percentage 0 0 0.04 0 1 2 3 4 0.04 0 1 2 3 4 Analysis |havg − 5| |havg − 5|

P Songs 6.3 8000 Blogs Tweets 0 0 6.2 6000 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 Positivity Bias 0.16 0.16 100 100 References 75 C. New York Times 75 D. Music Lyrics 6.1 4000 50 50 1−1000 average valence v 25 25 0.12 percentile 0.12 percentile 0 0 1001−2000 1 3 5 7 9 1 3 5 7 9 6 2000 Normalized frequency 2001−3000

average follower number havg havg 100 100 3001−4000 positive 0.08 75 0.08 75 positive 4001−5000 5.9 0 50 50 negative 1 45 90 135 180 225 270 315 25 negative 25 percentage 0 percentage 0 day number 0.04 0 1 2 3 4 0.04 0 1 2 3 4 |havg − 5| |havg − 5|

0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 Average happiness havg 72 of 83 Complex Complex Sociotechnical Sociotechnical New York Times Systems Systems 1 1 A. Twitter B. Books 1 not the he i we new no ms million 1000 1000 war tax loving Measuring Measuring funeral union womenparty love hospitalleast somethinglook give won Happiness Happiness 500 error small incorrectly Some motivation 2000 2000 Some motivation action car parentspeacewin Measuring emotional Measuring emotional cancer debt trial army officerscolumbia london r 1000 content content attack condolences test goal Data sets Data sets prison mourn low maybe figurenetwork visit 3000 3000 violence declined executives gold Analysis r Analysis 1500 lose pass winning jeffrey soccersongs Songs Songs murderkilling attacks alone requireissued partieswonderfulsuccessful 4000 4000 strike Blogs Blogs twicesites girl Tweets Tweets 2000 damageemergencydisputeheavyboxing highly song failure schedule 5000 5000 o Positivity Bias 0 1 2 3 0 1 2 3 Positivity Bias abuse inflation fee ends india truly perfect 2500 arrest unusual advisertaught jail References References map deals comedy injury waste warned federation attend super offensive prosecutorseriously wonder 1 1 3000 celebration C. New York D. Music deaths biographyliterary terrorist danger hardlysharply hits shape encouragewins Times Lyrics wounded erroneous funny 1000 1000 excellent ages bobbypilot 3500 rivaldefendants inspiration fearsindictment winnersenjoyed unemployment administrator compensation 2000 2000 disaster scandal raton weak slowly shoulder folk thankshero Word Usage Frequency Rank Word Usage Frequency Rank unique 4000 sadness purchased rape recession unclear 3000 3000 sick competing prefercooking cocaine offense investigations employersprospects victories 4500 lawsuits consequences pleasure inmates billings chip saddamdamages fled pres impressiontraveledimprovingmotherssucceeded 4000 4000 wars 5000 5000 5000 1 2 3 4 5 6 7 8 9 0 1 2 3 0 1 2 3 Average happiness havg Standard deviation of happiness hσ 73 of 83 76 of 83

Complex Complex Sociotechnical Random other things (now and next): Sociotechnical Systems Systems 1 1 A. Twitter B. Books

1000 1000 Measuring Measuring Happiness Happiness 2000 2000 Some motivation Some motivation Measuring emotional I Gross National Happiness Index, hedonometer.org Measuring emotional content content Data sets Data sets 3000 3000 (in development)

r Analysis Analysis Songs I Prediction . . . Songs 4000 4000 Blogs Blogs Tweets I Scores for letters, phonemes, as a function of tense. Tweets 5000 5000 Positivity Bias Positivity Bias 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 I Fifteen additional languages being scored on References Mechanical Turk References 1 1 C. New York D. Music I How does happiness vary with proximity to nature? Times Lyrics 1000 1000 to Walmart?

2000 2000 I Emotional contagion. Word Usage Frequency Rank

3000 3000 I Quantifying metaphor and narrative and stories . . .

4000 4000

5000 5000 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 74 of 83 77 of 83 Average happiness havg

Complex Complex Sociotechnical References I Sociotechnical Books Systems Systems

1 are his with i these one us own work Measuring Measuring re 500 st along thing near lord Happiness [1] M. Bradley and P. Lang. Happiness experienceenough army Some motivation Some motivation etc pay Affective norms for english words (anew): Stimuli, types training literatureindia Measuring emotional Measuring emotional historical content content 1000 understandingadministration sitele Data sets instruction manual and affective ratings. Data sets agreementheavysouthern showed decemberafrica summer san Analysis Technical report c-1, University of Florida, Analysis 1500 sight fish winter search coast stories poetry Songs Songs Blogs Blogs medium Gainesville, FL, 1999. pdf (
) pre les height bar god's Tweets Tweets 2000 he'sregional naturally escape Positivity Bias Positivity Bias quotedneck [2] T. Conner Christensen, L. Feldman Barrett, mid windows bible churches 2500 maintenanceconsumption References References oral snow smile extracopper E. Bliss-Moreau, K. Lebo, and C. Kaschub. inch rooms sounds alcohol 3000 christiansradiation A practical guide to experience-sampling williamscreative trip bonds ann reflection telephonesocialist recall procedures. ne wet 3500 scottmeanwhileparticipantsmistake organ busy priests Journal of Happiness Studies, 4:53–78, 2003.

Word Usage Frequency Rank ha cat pregnancy romanticexcited 4000 municipal disappeared lords es aids mortality mineralatoms [3] M. Csikszentmihalyi. seventhtopics acknowledgedtheologicalrelativesconquest 4500 covers islam Flow. sooneraimedloyalty burst definitions devotion capitalism Harper & Row, New York, 1990. 5000 0 0.5 1 1.5 2 2.5 3 Standarddeviationofhappiness hσ 75 of 83 78 of 83 Complex Complex References II Sociotechnical References V Sociotechnical Systems Systems

[13] S. Lyubomirsky. [4] M. Csikszentmihalyi, R. Larson, and S. Prescott. Measuring Measuring Happiness The How of Happiness. Happiness The ecology of adolescent activity and experience. Some motivation Some motivation Measuring emotional The Penguin Press, New York, 2007. Measuring emotional Journal of Youth and Adolescence, 6:281–294, content content Data sets Data sets 1977. Analysis [14] C. Martinelli and S. W. Parker. Analysis Songs Songs [5] E. Diener and M. Y. Chan. Blogs Deception and misreporting in a social program. Blogs Happy people live longer: Subjective well-being Tweets forthcoming in Journal of the European Economic Tweets Positivity Bias Positivity Bias contributes to health and longevity. Association, 2007. pdf ( ) References
References Applied Psychhology: Health and Well-Being, [15] G. Miller. 3:1–43, 2011. pdf ( )
Social Scientists wade into the Tweet stream. [6] P. S. Dodds and C. M. Danforth. Science Magazine, 333:1814–1815, 2011. pdf (
) Measuring the happiness of large-scale written [16] G. Mishne and M. de Rijke. expression: Songs, blogs, and presidents. Capturing global mood levels using blog posts. Journal of Happiness Studies, 2009. AAAI 2006 Spring Symposium on Computational doi:10.1007/s10902-009-9150-9. pdf ( )
Approaches to Analysing Weblogs, 2005.

79 of 83 82 of 83

Complex Complex References III Sociotechnical References VI Sociotechnical Systems Systems [7] P. S. Dodds, K. D. Harris, I. M. Kloumann, C. A. Bliss, and C. M. Danforth. Measuring Measuring Happiness Happiness Temporal patterns of happiness and information in a Some motivation Some motivation Measuring emotional Measuring emotional global social network: Hedonometrics and Twitter. content content Data sets Data sets

PLoS ONE, 6:e26752, 2011. Analysis Analysis Draft version available at Songs Songs Blogs [17] E. Sandhaus. Blogs http://arxiv.org/abs/1101.5120v4. Tweets Tweets Positivity Bias The New York Times Annotated Corpus. Positivity Bias Accessed October 24, 2011. pdf (
) References Linguistic Data Consortium, Philadelphia, 2008. References [8] W. T. Jones. The Classical Mind. Harcourt, Brace, Jovanovich, New York, 1970. [9] D. Kahneman, A. B. Krueger, D. A. Schkade, N. Schwarz, and A. A. Stone. A survey method for characterizing daily life experience: The day reconstruction method. Science, 306(5702):1776–1780, 2004. pdf (
) 80 of 83 83 of 83

Complex References IV Sociotechnical Systems

Measuring [10] M. A. Killingsworth and D. T. Gilbert. Happiness Some motivation A wondering mind is an unhappy mind. Measuring emotional content Science Magazine, 330:932, 2010. pdf (
) Data sets Analysis [11] I. M. Kloumann, C. M. Danforth, K. D. Harris, C. A. Songs Blogs Bliss, and P. S. Dodds. Tweets Positivity of the English language. Positivity Bias PLoS ONE, 7:e29484, 2012. References Draft version available at http://arxiv.org/abs/1108.5192. Accessed October 24, 2011. pdf (
) [12] R. Layard. Happiness. The Penguin Press, London, 2005.

81 of 83