I gave this talk at SciPy'06. A few things have changed since then, but it's suprising how many haven't:
Higher productivity still gets more people into the room than correctness or reproducibility.
"How to publish" was as absent then as it is now. (Figuring out what to teach about publishing science in the 21st Century is pretty close to the top of my to-do list these days.)
I'm still an awful graphic designer.
The biggest thing I notice, though, is that there was no mention of how to teach. It would be another three years before I encountered the literature on teaching and learning, and another two years after that before I started taking it seriously.
Selling Python to Scientists
What I've Learned from Software Carpentry
Plus Ça Change...
Software engineering for science has to address three fundamental issues: (i) dealing with datasets that are large in size, number, and variations; (ii) construction of new algorithms to perform novel analyses and syntheses; and (iii) sharing of assets across wide and diverse communities.
– Emmott et al, Towards 2020 Science
The State of Play
Productivity? Tell Me More...
You Can't Sell a Language
What Are Their Problems?
Timing, Timing, Timing
☑ Version Control
☑ Data Crunching
☒ Prototyping and GUIs
Portal (n): a gateway...
CS 101: The Final Frontier