I've been forcing myself to learn enough R and ggplot to make nice looking plots. It's taking me painfully long to learn to do even the most basic things, and +Carlos Scheidegger has been a big help !
What makes learning ggplot hard is grasping its underlying design philosophy. There's a notion of layers in a plot that I'm still getting used to. Once you get the hang of it it's quite elegant, and makes modifying plots very easy -- once you get the hang of it.
All of which makes me a little more sympathetic to people who struggle with using LaTeX. LaTeX has many of the same unnatural design principles, starting with the lack of a WYSIWIG interface (and no, LyX doesn't really count). It's an incredibly powerful interface (like ggplot), but it's darned hard to do simple things.
In fact, I've been wishing for a "Word equivalent" for ggplot2. Just like overleaf.com is now trying to make LaTeX easier for the general public, I would love to see some kind of interactive interface to ggplot2 that can generate the code automatically for standard tasks.
Thursday, April 30, 2015
Tuesday, April 28, 2015
I've had an intense and exciting day and a half here at "politics camp", otherwise known as LISPI. Many thanks to the CRA for organizing this event. It's been an eye-opener in many ways. I've come away from this meeting with a new appreciation for the kind of lobbying efforts CS folks undertake in DC to make sure we can continue to do research and live in our little bubbles.
One of the "homework assignments" that we were given yesterday for today morning was to practice a 3-4 minute pitch to a staffer for a representative/Senator on a topic of our choosing (about the importance of NSF funding, about a specific issue we think is important, etc). Our "panel" consisted of former Hill staffers who've been on both sides of the pitch table, and each of us had to do a "Shark Tank"-like pitch to them.
It was surprisingly nerve-wracking. I had prepared a pitch on recidivism and the importance of fairness when using data-driven predictors, and having to reduce it down to a few minutes of intelligible speech took some effort. Luckily, the panelists were very constructive in their criticism.
Outside the presentations and discussions, there are some serious issues bubbling up in DC right now that kept cropping up during the workshop.
Firstly, the 2007 America COMPETES Act is up for reauthorization, and the House Science committee is attempting to take an axe to NSF funding for social sciences and geosciences. In a move that would have made the British Raj (and all computer scientists) proud, they dangled a classic divide-and-rule trick over the head of the NSF by increasing the budget of CISE, engineering and MPS as a "compensation". This level of detailed guidance for budgeting is apparently unprecedented and the NSF (and the CRA) is fighting it.
Secondly, there's a big debate going on about backdoor encryption and whether it's technically possible and/or desirable to allow government backdoors into encryption on devices like iPhones etc. I'm not particularly competent to weigh in on these issues, but there were a lot of security folks at the workshop who brought this up as their major concern during our morning pitch session.
In any case, it's away from the US and onto Canada, for SDM 2015 in Vancouver.
Monday, April 27, 2015
I'm at the CRA-organized Leadership in Science Policy Institute, otherwise known as LISPI. After a day of intense talks, I feel like day three of a conference: which is to say, totally drained and overwhelmed by the amount of information coming my way. Who knew that science sausage was so hard to make !
Here are some things I've learnt so far:
Here are some things I've learnt so far:
- Physics is our feared and admired
- I'd be glad to have half as much energy as Ed Lazowska when I get to his level.
- The CRA continues to be truly awesome
- I feel like I'm in an episode of the West Wing sometimes.
- The budget process seems far more structured and less crisis-ridden than the media makes it out to be.
And for all other details of the meeting, I will refer you to the Chatham House Rule.