Tuesday, February 2, 2016

Why I don't want to share your screen -- OR -- What I learned from stackoverflow

I know it sounds arrogant, but I don't want to share your screen to sort out a Python programming problem. I have two reasons and I think one of them is a good one.

It's both pedagogical and personal. 

Personally, I'm often left breathless by demos. Watching the cursor fly around the screen is -- well -- dizzying. What was I supposed to be watching? Who's IM messages are popping up? What meeting reminders are you ignoring?

It may seem helpful to wave the cursor around, and show me your whole desktop world. And for some people, the discussion may actually be helpful. Sometimes they have an epiphany while they're explaining stuff to me. That's good. For me, it's bewildering. Sorry. I'm only going to read the visible fragments of your emails in the background window.

From a pedagogical perspective, there's this point:

I think that it's very important to learn how to focus on the details that matter.

This breaks down into several related skills:

  1. I think everyone needs to be able to copy and paste text. Screenshot images are hard to work with. On Stack Overflow, a 4-space indent is mandatory. It's not hard. A surprising number of programmers struggle with it.
  2. Articulate the actual problem. "Doesn't work" really is not sensible. I think it's important to insist on a concrete statement of the problem. Asking me to deduce it while looking at your screen isn't building any of your skills. 
  3. Find the relevant portion of the Python traceback. Yes, that's hard. But it's part of coding. Asking me to read the traceback doesn't build your skills.
  4. Find the relevant portions of the code that's broken. Again, when I pinpoint the line of code from reading the traceback, your skills haven't grown. I'm well aware that it's confusing when there's a long traceback from a framework that only seems to include your module 6 levels in. If you aspire to mastering code, that has to be part of your aspiration.
  5. Hypothesize a root cause. This is perhaps the hardest skill. The confirmation bias problem leads many people to write wrong code and complain that it's "broken" in a vague way. During screen sharing they scroll past their assumptions as if they're always correct. I have sympathy. But, it's essential to understand the semantics of alanguage. More importantly, it's essential to learn to judge where our assumptions might deviate from reality. Overcoming confirmation bias is hard. Maybe a long conversation is the only way to realize this; I hope not.
  6. Experiment. Python offers the >>> prompt at which you can experiment. Use it. This is the best way to explore your assumptions and see what the actual language semantics are.
Maybe I'm just being hypersensitive, but there's little to really talk about. If we could focus on the relevant code, perhaps through copy-and-paste, I can help. Otherwise, I feel like I'm just watching helplessly while an amusement park ride spins me around for a while, leaving me dizzy and confused. And not having offered any concrete help.