Scientific Software and the Open Collaborative Web

Practices vary between scientific domains but all too often the sharing of research software is done on an ad hoc basis between individuals and with little thought about the wider community. With code and computation routinely forming the backbone of many academic endeavours we need to focus on publishing *all* of the products of our research – papers, software, data and provenance. In this talk I will highlight some promising examples from across academia and discuss how software development in the sciences could benefit from the collaboration norms of a modern open source project.

Transcript

1. Scientific Software
   and the
   Open Collaborative Web
   Arfon Smith
   @arfon
   

   View Slide

2. http://www.flickr.com/photos/blachswan
   

   View Slide

3. http://www.flickr.com/photos/esoastronomy/
   

   View Slide

4. http://www.flickr.com/photos/esoastronomy/
   http://www.flickr.com/photos/jamiegilbert
   

   View Slide

5. http://amandabauer.blogspot.com/
   

   View Slide

6. View Slide

7. View Slide

8. Diffraction grating
   Telescope
   Detector
   

   View Slide

9. View Slide

10. View Slide

11. View Slide

12. View Slide

13. View Slide

14. 130 130 1 2048
    189 189 258 258
    480 562 378 378
    493 521 390 397
    851 851 247 274
    319 319 304 580
    493 511 610 636
    188 188 228 228
    > cat bad_pix_mask.txt
    

    View Slide

15. Wasteful
    

    View Slide

16. Wasteful
    2 days work
    !
    

    View Slide

17. Wasteful
    2 days work
    3 observing runs/week
    !
    

    View Slide

18. Wasteful
    2 days work
    3 observing runs/week
    52 weeks in year
    

    View Slide

19. Wasteful
    2 days work
    3 observing runs/week
    52 weeks in year
    15 year detector lifetime
    

    View Slide

20. Wasteful
    2 days work
    3 observing runs/week
    52 weeks in year
    15 year detector lifetime
    !
    2*3*52*15 = 4680 days (13 years)
    

    View Slide

21. Wasteful… but the norm
    2 days work
    3 observing runs/week
    52 weeks in year
    15 year detector lifetime
    !
    2*3*52*15 = 4680 days (13 years)
    

    View Slide

22. View Slide

23. Why?
    

    View Slide

24. We don’t know any different
    

    View Slide

25. We’re taught to focus on
    research not tools
    

    View Slide

26. We don’t act any different
    

    View Slide

27. Products of research, not
    software valued
    

    View Slide

28. Not treating software as first
    class research objects
    

    View Slide

29. View Slide

30. View Slide

31. View Slide

32. View Slide

33. “publishing a paper about
    code is basically just
    advertising”
    David Donoho
    http://www.stanford.edu/~vcs/Video.html
    

    View Slide

34. How is the Open Source
    community doing it?
    

    View Slide

35. Culture of reuse
    

    View Slide

36. Low friction collaboration
    

    View Slide

37. View Slide

38. The pull request
    

    View Slide

39. Code first, permission later
    

    View Slide

40. View Slide

41. View Slide

42. View Slide

43. View Slide

44. View Slide

45. View Slide

46. View Slide

47. “open source is…
    reproducible by necessity”
    Fernando Perez
    http://blog.fperez.org/2013/11/an-ambitious-experiment-in-data-science.html
    

    View Slide

48. Better at collaborating
    because they have to be
    

    View Slide

49. The social web
    ⋆
    ♥
    likes props
    favs
    

    View Slide

50. The social web
    The collaborative web
    

    View Slide

51. GitHub is about helping people
    build software together
    

    View Slide

52. What’s happening today?
    

    View Slide

53. Collaboration around code
    

    View Slide

54. Unidata: geosciences
    

    View Slide

55. View Slide

56. View Slide

57. View Slide

58. View Slide

59. View Slide

60. Astropy: astronomy
    

    View Slide

61. View Slide

62. View Slide

63. View Slide

64. View Slide

65. View Slide

66. Collaborative authoring
    

    View Slide

67. View Slide

68. View Slide

69. Collaborative teaching
    

    View Slide

70. View Slide

71. View Slide

72. View Slide

73. Collaborative data collection
    

    View Slide

74. View Slide

75. View Slide

76. Towards Collaborative
    Versioned Science
    

    View Slide

77. How do we make this
    behaviour the norm?
    

    View Slide

78. Incentive model
    

    View Slide

79. Credit
    

    View Slide

80. http://dx.doi.org/10.6084/m9.figshare.828487
    

    View Slide

81. http://dx.doi.org/10.6084/m9.figshare.828487
    

    View Slide

82. View Slide

83. “publishing a paper about
    code is basically just
    advertising”
    David Donoho
    http://www.stanford.edu/~vcs/Video.html
    

    View Slide

84. View Slide

85. Derive meaningful metrics
    from open contributions
    

    View Slide

86. “Academic environments of
    today do not reward tool
    builders”
    Ed Lazowska, OSTP event
    http://lazowska.cs.washington.edu/MS/MS.OSTP.pdf
    

    View Slide

87. A VISION AND STRATEGY
    FOR SOFTWARE FOR
    SCIENCE, ENGINEERING,
    AND EDUCATION
    

    View Slide

88. Establish a virtuous cycle#
    •  6%working%groups,%each%with%#
    •  3:6%faculty%from%each%ins;tu;on#
    14#
    http://lazowska.cs.washington.edu/MS/MS.OSTP.pdf
    

    View Slide

89. http://lazowska.cs.washington.edu/MS/MS.OSTP.pdf
    Software Tools, Environments, and Support#
    •  SoSware%environments%and%tools%are%crucial%
    •  Organic,#sustainable,#reusable,#extensible#
    •  Easy#to#translate#across#problem#domains#
    •  The#creaHon#and#usage#of#today’s#tools#and#sojware#
    environments#are#distracHng#from#the#science#
    •  Today's%academic%environments%do%not%reward%tool%builders#
    •  How#can#the#development,#hardening,#sustaining,#sharing,#
    and#integraHon#of#techniques#into#a#reusable#sojware#
    infrastructure#be#recognized#and#incenHvized?#
    17#
    Example%approach:%Teams%of%soSware%architects,%engineers,%
    and%researchers%who%will%produce%data%science%tools%and%will%be%
    evaluated%on%the%impact%of%these%tools%
    

    View Slide

90. What can we do today?
    

    View Slide

91. Share more often
    

    View Slide

92. If you’re going to share it then
    you better put a licence on it
    

    View Slide

93. View Slide

94. Share more often
    (no matter how small)
    

    View Slide

95. 130 130 1 2048
    189 189 258 258
    480 562 378 378
    493 521 390 397
    851 851 247 274
    319 319 304 580
    493 511 610 636
    188 188 228 228
    > cat bad_pix_mask.txt
    > git clone git@github.com:arfon/aat/pixel_masks
    

    View Slide

96. Thanks.
    arfon@github.com
    @arfon
    $
    

    View Slide