This document discusses the lack of computational skills among many scientists and proposes ways to improve this situation. It notes that most scientists focus on their research and do not engage with new technologies. Short, practical training courses targeting graduate students can help more scientists learn basic skills like shell scripting, version control, and testing. These skills allow scientists to be more productive and make new technologies more accessible, even if the skills are not publishable. The document advocates teaching a bit of computing in every course, hosting workshops, and changing incentives to better support computational reproducibility and open science.

1. Dark Matter,
Public Health,
and
Scientific
Computing
Greg Wilson

2. “Dark Matter Developers”
Scott Hanselman (March 2012)
[We] hypothesize that there is another kind
of developer than the ones we meet all the
time. We call them Dark Matter Developers.
They don't read a lot of blogs, they never write blogs,
they don't go to user groups, they don't tweet or
facebook, and you don't often see them at large
conferences... [They] aren't chasing the latest beta
or pushing...limits, they are just producing.
http://www.hanselman.com/blog/DarkMatterDevelopersTheUnseen99.aspx
2

3. I'm Not That Optimistic
● 95% of scientists have their heads down
– Doing science instead of talking about using GPU
clouds to personalize collaborative management
of reproducible peta-scale workflows
● Not because they don't
want to do all that Star
Trek stuff
● But because e-science
is out of reach
3

4. Shiny Toys
4

5. Grimy Reality
5

6. So Here We Are
You Them
6

7. Surely You're Exaggerating
1. How many graduate students write shell
scripts to analyze each new data set
instead of running those analyses by hand?
7

8. Surely You're Exaggerating
2. How many of them use version control
to keep track of their work and collaborate
with colleagues?
8

9. Surely You're Exaggerating
3. How many routinely break large problems
into pieces small enough to be
- comprehensible,
- testable, and
- reusable?
9

10. Surely You're Exaggerating
3. How many routinely break large problems
into pieces small enough to be
- comprehensible,
- testable, and
- reusable?
And how many know those are the
same things?
10

11. “Not My Problem”
Actually your biggest problem
If someone is chronically malnourished as a
child, giving them a CT scanner when they're
20 has no effect on their wellbeing.
11

12. Their Biggest Problem Too
Them Us
“I don't know much “Let's talk about
about this, but I'd brain scans.
like some clean water.” They're cool.”
12

13. We've Left the Majority Behind
13

14. We've Left the Majority Behind
Other than
Googling for things,
the majority of scientists
do not use computers
more effectively today
than they did 28 years ago.
14

15. The Goalposts
A scientist is computationally competent if she
can build, use, validate, and share software to:
● Manage and process data
● Tell if it's been processed correctly
● Find and fix problems when it hasn't been
● Keep track of what she's done
● Share work with others
● Do all of these things efficiently
15

16. A Driver's License Exam
● Developed with the Software Sustainability
Institute for the DiRAC Consortium
● Formative assessment
– Do you know what you need to know in order to
get the most out of this gear?
● Pencils ready?
Note: actual exam allows for several different
programming languages, version control systems, etc.
16

17. A Driver's License Exam
1. Check out a working copy of the exam
2. materials from Subversion.
3.
4.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
17

18. A Driver's License Exam
1. Use find and grep in a pipe to create a
2. list of all .dat files in the working copy,
3. redirect the output to a file, and add that
4. file to the repository.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
18

19. A Driver's License Exam
1. Write a shell script that runs a legacy
2. program for each parameter in a set.
3.
4.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
19

20. A Driver's License Exam
1. Edit the Makefile provided so that if any
2. .dat file changes, analyze.py is re-run to
3. create the corresponding .out file.
4.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
20

21. A Driver's License Exam
1. Write four unit tests to exercise a function
2. that calculates running sums. Explain why
3. your four tests are most likely to uncover
4. bugs in the function.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
21

22. A Driver's License Exam
1. Explain when and how a function could
2. pass your tests, but still fail on real data.
3.
4.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
22

23. A Driver's License Exam
1. Do a code review of the legacy program
2. from Q3 (about 50 lines long) and list
3. the four most important improvements
4. you would make.
5.
6.
7.
Could do it easily 1.0
Could struggle through 0.5
Wouldn't know where to start 0.0
Don't understand the question -1.0
23

24. How Are We Doing?
a) How did you do?
24

25. How Are We Doing?
a) How did you do?
b) How do you think most scientists do?
25

26. How Are We Doing?
a) How did you do?
b) How do you think most scientists do?
c) Do you think a scientist could use
a peta-scale GPU provenance
cloud without these skills?
26

27. How Are We Doing?
a) How did you do?
b) How do you think most scientists do?
c) Do you think a scientist could use debug
a peta-scale GPU provenance
cloud without these skills?
27

28. How Are We Doing?
a) How did you do?
b) How do you think most scientists do?
c) Do you think a scientist could use debug
a peta-scale GPU provenance extend
cloud without these skills?
28

29. So Why Is This Your Problem?
If you're only helping the small minority
lucky enough to have acquired the skills
that mastering your shiny toy depends on...
your potential user base is many times
smaller than it could be.
29

30. It Is Therefore Obvious That...
● Put more computing courses in the curriculum!
– Except the curriculum is already full
30

31. It Is Therefore Obvious That...
● Put a little computing in every course!
– Still adds up: 5 minutes/lecture = 4 courses/degree
– First thing cut when running late
– The blind leading the blind
31

32. It Is Therefore Obvious That...
● Move all our teaching online!
– Where do I start?
– Signal vs. noise
– “It's not a MOOC, it's a MESS”
Everyone who understands something
correctly understands it the same way.
Everyone who misunderstands something
misunderstands differently.
32

33. What Has Worked
● Target graduate students
33

34. What Has Worked
● Target graduate students
● Intensive short courses (2 days to 2 weeks)
34

35. What Has Worked
● Target graduate students
● Intensive short courses (2 days to 2 weeks)
● Focus on practical skills
35

36. What Has Worked
● Target graduate students
● Intensive short courses (2 days to 2 weeks)
● Focus on practical skills
● Follow up with 6-8 weeks of once-a-week
online tutorials and help
– Still tuning this part
36

37. 37

38. What We Teach
● Unix shell
● Python
● Version control
● Testing
● Matrix programming,
image processing,
SQL, ...
38

39. What We Teach
● Unix shell
● Python
● Version control
● Testing
● Matrix programming,
image processing,
SQL, ...
39

40. “That's Merely Useful”
● None of this is publishable any longer
– Which means it's ineffective career-wise for
academic computer scientists
40

41. “That's Merely Useful”
● None of this is publishable any longer
– Which means it's ineffective career-wise for
academic computer scientists
● But it works
– Two independent assessments have validated the
impact of what we do for 1/3-2/3 of learners
– That's at least a six-fold increase in the number of
scientists who can get work done without karōshi
41

42. Majestic Equality
Anatole France (1844-1924)
“The law, in its majestic equality, forbids
the rich and poor alike to sleep under
bridges, to beg in the streets, and to
steal bread.”
42

43. Majestic Equality
Anatole France (1844-1924)
“The law, in its majestic equality, forbids
the rich and poor alike to sleep under
bridges, to beg in the streets, and to
steal bread.”
Thanks to modern computers,
every scientist can now devote her working life
to installation and configuration issues.
43

44. If you build a man a fire,
you'll keep him warm for a night.
If you set a man on fire,
you'll keep him warm for the rest of his life.
— Terry Pratchett
44

45. Host a Workshop
45

46. Teach a Workshop
● All our materials are CC-BY licensed
● We will help train you
46

47. Shine Some Light
● Include a few lines about your software stack
and working practices in your scientific papers
● Ask about them when reviewing
– Where's the repository containing the code?
– What's the coverage of your unit tests?
– How did you track provenance?
47

48. Be the Change You Want to See
Would you rather increase the
productivity of the top 5% of
scientists ten-fold?
Or double the productivity of the
other 95%?
48

49. Be the Change You Want to See
Would you rather increase the
productivity of the top 5% of
scientists ten-fold?
Or double the productivity of the
other 95%?
49

50. Be the Change You Want to See
Would you rather see your best
ideas left on the shelf because
they're out of reach?
Or raise a generation who can all
do the things we think are exciting?
50

51. http://software-carpentry.org
51