You are on page 1of 28

Towards Open-World Scenarios: Teaching the

Social Side of Data Science

Joseph Corneli, Dave Murray-Rust, and Benjamin Bach


University of Edinburgh

14:20–14:40, Friday, 6 April, 2018


Serendipity@AISB
1. Challenges in Teaching Data Science
Challenges in Teaching Data Science

▶ Toy datasets
▶ Real-world datasets
▶ Motivation
▶ Complexity
▶ Relevance
▶ Soft-skills
▶ Method evaluation
▶ Reflexive learning and self-monitoring
▶ Interdisciplinary audience
“Properties that are thought to reside in things are properties of
the observer.” – von Foerster
“Real world data” ≠ “real world problem”
Also in light of the previous slide: students typically have to
demonstrate mastery of simple, repeatable, skills.
2. Open world teaching
Open world teaching

▶ Universities and Society


▶ Teaching and Research
▶ Public Action
▶ Field-work and Collaboration
▶ Teaching to Develop Deeper Understanding
(Is this what “public action” looks like?)
“This is not, however, fast enough for an instrument like
IceCube, whose 5160 sensors each count at a rate of roughly 300
Hz, for a total rate of 1.5 MHz.” – John Jacobson
According to K. Burke, the “question that ultimately concerns
us most” is: “What is the nature of a symbol-using animal?”
2. Serendipity in practice
Serendipity in practice

▶ Collaboration
▶ Topics
▶ Contextualisation and interpretation
▶ Motivation (again!)
▶ Skills
▶ New models, methods, organisations, and theories
“Data science is inherently a team sport” – Stephen Kearns,
Continuum Analytics
(Here’s what I’m working on in my “day job” at present.)
“The color of the sky is green… on Mars” – Paul Pangaro
“[Shelita] Burke thinks it also has something to do with her use of
data to her advantage, like when she determined that 90 days was the
perfect time to release new music in order to keep fans engaged.”
“Back in 2015, [Imogen Heap’s] song Tiny Human became the first
song to automatically distribute payments via an Ethereum smart
contract.”
…or guacamole?

You might also like