It wasn’t until after I’d submitted the talk proposal to 3 conferences that I realized how big of a chunk I’d bitten off by writing in a computer vision aspect. “Participants will gain insight on impactful mindsets and actionable strategies for more effective professional development – as well as a few starter Computer Vision scripts!” Yeah…assuming they’re running Python 2.7 on a raspberry pi and are comfortable debugging their own fork of the seemingly abandoned SimpleCV repo. Smooth move, Marissa. Clearly a stellar talk. Knock ’em dead.
However, as the focus of my talk is on leveraging research-based strategies from the field of cognitive science to improve your own learning, this was obviously an excellent opportunity to give my talk takeaways a focused run-through. So before I get to the outcome of my work on the SimpleCV library, let’s establish those cognitive theory talking points.
My talk is structured around a study from several research groups(†) that identified characteristics differentiating how novices and experts approach new situations in a given domain. The shortest description the study’s authors gave for their concept was the six principles of expertise, which makes me cringe to say. As a mathematician, I tend to have a really hard time making imperative statements. What if we missed a criterion? Or worse, what if we find a counterexample??? (Overthink much, Marissa?) Regardless, the 6 characteristics they described were as follow:
- Experts notice features and meaningful patterns of information that are not noticed by novices.
- Experts have acquired a great deal of content knowledge that is organized in ways that reflect a deep understanding of their subject matter.
- Experts’ knowledge cannot be reduced to sets of isolated facts or propositions but, instead, reflects contexts of applicability: that is, the knowledge is “contextualized” on a set of circumstances.
- Experts are able to flexibly retrieve important aspects of their knowledge with little attentional effort.
- Though experts know their disciplines thoroughly, this does not guarantee that they are able to teach others.
- Experts have varying levels of flexibility in their approach to new situations.
In the talk, I unpack each characteristic with comparisons that developers across domains can relate to. However, I would like to have more concrete contrasts as a way to get some tangible ML concepts and realities across. With that in mind, I considered my experience digging into SimpleCV in the context of expertise.
Running through a quick inventory of my expertise at this stage, I can’t avoid the reality of principle 4: fluency. Expert pythonistas tend to recall appropriate libraries and commands much more quickly that I would say that I currently do. Until I can approach a problem with a list of possible libraries or solutions off-hand, fluency is my counter-example. Ok, so I fall into the novice category at this point. What now?
This is where we consider scaffolding, also known as scaling. In the field of education, scaffolding is targeted scaling that we specifically intend to remove as proficiency increases. We may scale the scope of our goal or the time-frame for success, the size of our team or the resources we intend to use. For me, it seemed appropriate to start by segmenting my scope. Rather than trying to put together a tutorial that relies on SimpleCV being functional in a wide range of environments, I first needed to get SimpleCV to run. Full stop.
Troubleshooting goal #1: Make it work
Operating at the novice level does not disqualify you. Metacognitive self-assessment is not an excuse to box yourself into a fixed category. Genuine and critical self-reflection should be interpreted in the context of ‘artisans’ vs ‘virtuosos’ (principle 6): everyone will be a novice in some domain at many points in a lifetime. That’s kind of the point of being a life-long learner – we don’t know it all! So being at the novice stage of learning does not mean you should wait to start a project until you are at the expert level. Pursuing projects for which you’re in over your head can progress you toward expert-like learning because you are contextualizing as you learn (principle 3). We rarely read through all documentation before beginning a project. Not that it’s a terrible idea, it’s just more likely to stick (or to make sense, period) if we have a concrete context for what we’re learning.
So my context for learning is clearly SimpleCV in this project, but the content of my learning is turning out to be much more than anticipated. Did I get it to work?
Heck yeah! Kind of. Calling simplecv in Python 2.7 no longer gives errors now that I’ve installed my updated fork of it on my Raspberry Pi, but there are still some bugs in trying to do image capture with it. For now, I’m going to say that this technically meet my goal #1. But I don’t feel much closer to expertise.
Now what?
Troubleshooting goal #2: Make it right
I could reassess and take stock of my growth toward expertise, which has not been insignificant. However, I think I’ll break that out into a separate post. For now, I’m playing around with my goals #2 (working through the actual functionality of the library, fixing bugs as they appear) and #3 (making progress toward Python 3 compatibility on my fork of SimpleCV). It feels like that third bite is much bigger than I’ll be able to achieve, but that’s assuming a team of size 1. In preparing for my talk itself, a certain amount of work will be on my own, but collaboration is part of the beauty of the open-source software community. In fact, discussing others’ progress toward Python 3 support and working with their forks makes that goal #3 feel more possible.
Since one of the takeaways of my talk is the power of growth mindset, I’ll close with a brief video on the topic.
Though my talk is changing as we speak, and will continue to do so even after I present it a few times, one takeaway that I will believe will continue to serve me well is this: no one is born knowing everything, so everything we want to learn is accessible – though it may take a few extra steps (and some creative risk) to get there.
Footnote:
† The study, published in April 1999 with the work of The Committee on Developments in the Science of Learning and of the National Research Council (NRC), was expanded upon in August 2000 with the work of an additional NRC committee, The Committee on Learning Research and Educational Practice.
