full transcript

From the Ted Talk by Sebastian Wernicke: Lies, damned lies and statistics (about TEDTalks)

Unscramble the Blue Letters


If you go on the TED wtebise, you can currently find there over a full week of TEDTalk videos, over 1.3 million words of transcripts and mlonilis of user rtgians. And that's a huge amount of data. And it got me wondering: If you took all this data and put it through statistical analysis, could you reersve engineer a TEDTalk? Could you create the ultimate TEDTalk? (Laughter) (Applause) And also, could you craete the worst possible TEDTalk that they would still let you get away with?

To find this out, I looked at three things: I looked at the topic that you should chosoe, I lkeood at how you should deliver it and the visuals onstage. Now, with the topic: There's a whole range of topics you can choose, but you should choose wisely, because your topic strongly correlates with how users will react to your talk. Now, to make this more concrete, let's look at the list of top 10 wrdos that statistically stick out in the most favorite TEDTalks and in the least favorite TEDTalks. So if you came here to talk about how French coffee will spread happiness in our bnaris, that's a go. (luhteagr) (Applause) Whereas, if you wanted to talk about your project inlivovng oxygen, grlis, aircraft — actually, I would like to hear that talk, (Laughter) but statistics say it's not so good. Oh, well. If you generalize this, the most favorite tlkdteas are those that feature topics we can connect with, both easily and deeply, such as happiness, our own body, food, eimtonos. And the more technical topics, such as architecture, materials and, strangely enough, men, those are not good topics to talk about.

Open Cloze


If you go on the TED _______, you can currently find there over a full week of TEDTalk videos, over 1.3 million words of transcripts and ________ of user _______. And that's a huge amount of data. And it got me wondering: If you took all this data and put it through statistical analysis, could you _______ engineer a TEDTalk? Could you create the ultimate TEDTalk? (Laughter) (Applause) And also, could you ______ the worst possible TEDTalk that they would still let you get away with?

To find this out, I looked at three things: I looked at the topic that you should ______, I ______ at how you should deliver it and the visuals onstage. Now, with the topic: There's a whole range of topics you can choose, but you should choose wisely, because your topic strongly correlates with how users will react to your talk. Now, to make this more concrete, let's look at the list of top 10 _____ that statistically stick out in the most favorite TEDTalks and in the least favorite TEDTalks. So if you came here to talk about how French coffee will spread happiness in our ______, that's a go. (________) (Applause) Whereas, if you wanted to talk about your project _________ oxygen, _____, aircraft — actually, I would like to hear that talk, (Laughter) but statistics say it's not so good. Oh, well. If you generalize this, the most favorite ________ are those that feature topics we can connect with, both easily and deeply, such as happiness, our own body, food, ________. And the more technical topics, such as architecture, materials and, strangely enough, men, those are not good topics to talk about.

Solution


  1. choose
  2. words
  3. reverse
  4. create
  5. involving
  6. looked
  7. laughter
  8. girls
  9. emotions
  10. millions
  11. tedtalks
  12. brains
  13. ratings
  14. website

Original Text


If you go on the TED website, you can currently find there over a full week of TEDTalk videos, over 1.3 million words of transcripts and millions of user ratings. And that's a huge amount of data. And it got me wondering: If you took all this data and put it through statistical analysis, could you reverse engineer a TEDTalk? Could you create the ultimate TEDTalk? (Laughter) (Applause) And also, could you create the worst possible TEDTalk that they would still let you get away with?

To find this out, I looked at three things: I looked at the topic that you should choose, I looked at how you should deliver it and the visuals onstage. Now, with the topic: There's a whole range of topics you can choose, but you should choose wisely, because your topic strongly correlates with how users will react to your talk. Now, to make this more concrete, let's look at the list of top 10 words that statistically stick out in the most favorite TEDTalks and in the least favorite TEDTalks. So if you came here to talk about how French coffee will spread happiness in our brains, that's a go. (Laughter) (Applause) Whereas, if you wanted to talk about your project involving oxygen, girls, aircraft — actually, I would like to hear that talk, (Laughter) but statistics say it's not so good. Oh, well. If you generalize this, the most favorite TEDTalks are those that feature topics we can connect with, both easily and deeply, such as happiness, our own body, food, emotions. And the more technical topics, such as architecture, materials and, strangely enough, men, those are not good topics to talk about.

Frequently Occurring Word Combinations


ngrams of length 2

collocation frequency
favorite tedtalks 6


Important Words


  1. aircraft
  2. amount
  3. analysis
  4. applause
  5. architecture
  6. body
  7. brains
  8. choose
  9. coffee
  10. concrete
  11. connect
  12. correlates
  13. create
  14. data
  15. deeply
  16. deliver
  17. easily
  18. emotions
  19. engineer
  20. favorite
  21. feature
  22. find
  23. food
  24. french
  25. full
  26. generalize
  27. girls
  28. good
  29. happiness
  30. hear
  31. huge
  32. involving
  33. laughter
  34. list
  35. looked
  36. materials
  37. men
  38. million
  39. millions
  40. onstage
  41. oxygen
  42. project
  43. put
  44. range
  45. ratings
  46. react
  47. reverse
  48. spread
  49. statistical
  50. statistically
  51. statistics
  52. stick
  53. strangely
  54. strongly
  55. talk
  56. technical
  57. ted
  58. tedtalk
  59. tedtalks
  60. top
  61. topic
  62. topics
  63. transcripts
  64. ultimate
  65. user
  66. users
  67. videos
  68. visuals
  69. wanted
  70. website
  71. week
  72. wisely
  73. words
  74. worst