This talk from February 2012, by Bret Victor, is worth the 54 and a half minutes it takes to watch it. Its title is “inventing on principle,” and it is a fork with two tines. The first tine, and his overarching point, is that designers would do better not only solving particular problems, but having principles that guide their thinking, attentions, and expertise. Finding something that “is important, and necessary, and right, and using that to guide what you do.” This is familiar to people who build careers in academia to a large extent, since your professional profile is built around who you decide you are. For many people, who spend their time putting out fires, or solving immediate problems with specific solutions (like designers, product managers, marketers), inventing or working on principles is something of a luxury. But, as Victor notes, a useful and fruitful and ultimately a necessary step towards building a fulfilling career.
This is to say, it is both helpful to consider specific problems within a wider framework of what you think is important; and also to think about what you think is important, so that it can guide to whatever extent possible, the things you do. I have found that year-over-year, when confronted by thesis students, and students in classes, my perspective is that of a cultural institutionalist. For me, that means that while I think hard about individuals, I am more likely to seek explanations for changes in organizational form, priorities, activities, and the like in broader, higher-level phenomena. I am, more broadly, interested in an empirically-based and theoretically-driven worldview. Lately I have been encapsulating this by saying that I am interested in telling stories with data.
This exists in the music world already, in the form of notation software. Play something on your guitar, the software automagically (supposedly) will create the score for you. I think Victor is spot-on in saying that this reduces the distances between creator and product.
I hesitate to extrapolate from his strongest case to all cases. Victor’s amazing and wonderful worrydream site is a testament to how well this works, under some circumstances. I imagine that, applied to social science and statistics, we might get something a lot more disturbing. With a Victor-style statistics program, I could imagine setting the p-level tolerance to .05, and then using sliders to figure out what kinds of sample sizes would best fit our data to our theories. When we should be fitting our theories to our data.
But I love the idea of removing the barriers between statistical analysis and the underlying ‘finding out important stuff about the world that we need to know’. Right now, the Survey Documentation Analysis site at Berkeley is the best we have for social science data. Perhaps the graphical interfaces at FRED – the Fed Reserve @ St. Louis site does this best for economic data. But nothing approaches the elegance and thoughtfulness of design that Bret Victor has brought to his work. It is an impressive principle on which to be inventing.