Python Education Summit 2017 Notes

The Python Education Summit is happening on May 18, 2017 in Portland, OR. I am writing some notes as we go through the day. Apologies to all of the great things I've missed in these notes (some of the sessions run simultaneously, and I can, unfortunately, only be in one place at a time).

Keynote—Luciano Ramalho

Coming with a long history of development and teaching experience, Ramalho discussed his personal history learning to program as well as some of the exciting tools and methods of teaching programming in classrooms today. Ramalho humorously recounted his dismal Study Hall experience when he first moved to the US, and how the installation of Apple ][ machines in the library liberated him from the doldrums of sitting quietly in a classroom for an hour.

Ramalho asked, appropriately, "Why did only two of us in the whole classroom take the opportunity to teach ourselves to program?" He discussed with the audience, which led to lots of ideas about fear of failure or not wanting to be different from the rest of the group.

I think those are all valid points, but I wonder if it was more the result of a lack of vision: Those kids had no way to understand the valuable benefits they would get from learning to code. This is something I think many people deal with today: They don't really know what is involved with writing code, and they have no idea of how learning to code could be beneficial. It's a reflection of the poor understanding that people have of coding and software in general.

It seems to me unfair to expect students to be able to realize the benefit of learning something they have no idea about. It's exceptional when students can lead themselves to that point, but it's totally understandable why a student would not be able to get there without help. This is why it's so important to begin education about coding and software early in our educational careers.

Charlotte Chang—Climbing Rocks and Coding Blocks

Chang dicussed the challenges of learning to code and the ways in which these challenges can feel intimidating, scary, and dangerous. She made several great analogies to rock climbing (if you climb, you expect falls; if you code, expect failures) and generally asserted the idea that learning to code is a risky endeavor.

Chang's examples and experiences definitely echo with me both personally and in terms of what I see in my students. There are many examples of students defeating themselves before they ever really get going due to the fear of failure, feelings of inadequacy, and generally discouraging experiences of frustration and exasperation.

Jessica Ingrassellino—Teaching a Diverse Population

Ingrassellino provided an impromptu talk filling in for a speaker whose Visa was denied and, therefore, was prevented from attending the conference.

Ingrassellino provided a helpful talk discussing ways in which she accommodates students with diverse needs and abilities in the classroom. She stressed the need to create a safe space where students could fail without fear, but also to modulate content to better reach students. She did a good job underlining the basics of differentiated pedagogy and concerning ourselves with the social well-being of the classroom.

Lightning Talks

I am not going to provide full summaries of the Lightning Talks (they go too quickly!), but here are some takeaways:

• PyCon UK has a great Education track!
• Logo is still relevant, programming with graphics is fun and good.
• Processing (the visual programming framework) with the Python module is cool.
• Helping students learn to ask questions is crucial.
• Answers that students discover themselves are much more meaningful and resilient.
• Show students what success looks like.
• Show students what failure looks like.
• Adafruit is a cool company that makes systems on which you can write Python.
• Adafruit has new, tiny, cheap chips that you can use to build cool things.
• Programming is a way of thinking. Read more about it here.
• Visual Bash.

Andrew M. Dawes—Teaching Quantum Mechanics with Python

Dawes discussed using Jupyter and QuTip to teach Quantum Mechanics to undergrads. Robust tools allow students to work through material in an interactive way, and give the students ways to do real quantum mechanics work using the same tools used by professional researchers.

Students are not required to have any prior Computer Science or programming before taking this course. They have been exposed to other Math-based tools (like Maple), so the interface and code is not totally foreign, but it is a substantial amount of new stuff. Students work together to get through challenges. Class time includes pair programming work.

Jupyter Notebook allows for a much more smooth integration of the advanced QuTiP tool into the classroom. The notebook allows information to be conveyed with interactivity. Example problems can be provided with solutions that students can interrogate and learn better from. Examples from the book can also be re-created in the notebook, allowing students to investigate those examples more fully, too.

Students write code in labs, where they are able to work together, spend more time investigating concepts, and easily get help in order to maximize their understanding.

Key takeaways (summarized by Dawes in his slides): Find real-world frameworks Re-create examples so they can reinforce what they read in their books Don't be afraid to give out fully-worked examples Encourage tinkering

Carol Willing—JupyterHub: Interactive Learning and Classrooms at Scale

What is learning? "Learning results from what the student does and thinks, and only from what the student does and thinks" (Herbert Simon). Willing points out many ways in which Jupyter Notebooks inspire people to "play" with them: change elements, experiment to get different results, and do lots of different things with very few lines of code.

Jupyter has diversified since the iPython Notebook days. The iPython core team develops the Python core of Jupyter, and then other community developers contribute modules to run over 70 other languages. This has allowed the concept of the Notebook to move beyond the boundaries of any one language.

Jupyter Hub allows for web-based Notebooks, sharing, and collaboration without requiring local installs of the software. Goal is to remove the hurdles for getting started with Notebooks.

After discussion of the value of Notebooks for providing valuable features and experiences, Willing discussed some additional pedagogical concepts. Discussing how the learning process works, Willing listed three major steps:

• Motivate - Teachers must inspire students to learn.
• Develop Mastery - Students must actually build skills and understanding
• Apply Knowledge - Students can use their skills and understanding to solve real problems in real situations.

Willing describes how Jupyter Notebooks help accomplish these goals. Notebooks can help motivate, help develop mastery, and then can actually be a tool used by students to apply their knowledge. Willing also goes through several examples of how Jupyter Notebooks and Jupyter Hub have been used in real courses.

Motivation

The nbviewer is an example of something that can motivate students by showing them cool things that people have done and enticing them to jump in and start experimenting with an existing Notebook.

Willing worked with Jessica Keller's Intro to Python workshop to translate the workshop information and goals into Notebooks that students could work through and then take home and continue experimenting and learning (or refer back to). The Notebooks reduced the stress level of students during the workshop.

Willing describes a course taught by Demba Ba at Harvard to teach singal processing using wearables and Jupyter Notebooks. (See SciPy 2016 talk for more details.)

Develop Mastery

Kristen Thyng teaches Python for Geosciences and uses the nbgrader tool to handle grading. Her course also begins with simple information and then seamlessly allows students to increase the complexity of what they are doing.

Tanya Schlusser teaches R with Jupyter Notebooks. She uses a flipped classroom approach to allow students to engage with materials at home, and in class they focus on working in Notebooks on code, examples, problems, etc. Class sessions are designed around short, practical challenges.

Brian Granger teaches Data Science at Cal Poly SLO that includes Jupyter Notebooks, Ansible, etc. Notebooks allow students to get much further than would otherwise be possible.

Apply Knowledge

UC Berkeley Data Science course (Data 8) is a cross-disciplinary course offered campus-wide that teaches data science with Notebooks and includes instruction for deploying Jupyter Hub with Kubernetes.

Jupyter Labs is a new product coming that will allow more complex Notebook creation including more media. Keep an eye out for this to launch and bring new features.

Al Sweigart—Tortuga.py: Expanding Turtle.py Beyond English

Sweigart points out that Python has some great features for teaching new coders using the classic Turtle programming module turtle.py. Considering teaching students whose native language is not English, Sweigart came to feel that the use of English words to program the Turtle ("up", "down", "left", "right", etc.) add to the difficulty of learning to program. Although key words ("for", "in", etc.) cannot be translated into other languages, the rest of the turtle.py module has been translated into Spanish and is available as Tortuga (pip install tortuga).

This definitely makes the code for Turtle scripts more accessible to coders in foreign languages. Sweigart has outlined a technical plan for improving the turtle.py component in the standard Python library in order to make this change fully accessible to people wanting to teach code in a foreign language. Each added language only adds about 4k of code to the package, which should allow for many other languages to be supported.

Sweigart hopes to encourage people to work with him to complete the project and eventually get the translations added to the standard library. He welcomes anyone to come work on the Tortuga.py project with him. Check out his presentation here.

Dan Pozmanter—Teaching Practical Python

Pozmanter wants to ease the frustration and anxiety students feel when they get out of school and into real-world projects. He discusses the need to create assignments/projects that are ambitious enough that students need to ask questions about "perspective":

• Thinking ahead
• Thinking behind
• Thinking sideways

Simple projects do not encourage this kind of thinking. And projects that are too difficult make it impossible for students to do a good job of thinking through the project. It's a challenge to hit the correct level of difficulty.

Some ways to cause students to leverage their perspective might be:

• Reading how somebody else solved a similar problem and determining how to use that code in their own projects.
• Add "scope creep" to projects: Change requirements in mid-assignment, add requirements, cause students to revise and re-envision their work.
• Have students roleplay QA teams where they actively try to break other students' apps and then figure out why it broke.
• Have students work on projects as business groups and "try on" all the roles of the group in order to better understand how various roles contribute as well as to better think about the software architecture.

Pozmanter mentioned a survey of software developers and things that cause "unhappiness":

• Getting stuck
• Time pressure
• Bad code quality
• Unclear requirements
• etc.

He advocates that these "unhappy issues" can be used as tools to teach and will also have the effect of helping students better cope with these issues when they come up outside the classroom.

Ultimately, the goal is to create a well-rounded developer who is more than just a coder:

• Communication skills
• Strategic Thinking
• Love of Learning
• Outside Interests
• Healthy relationship with failure
• Self confidence

The idea of reproducing unpleasant real-world situations in a course is appealing to me, too. I feel that we often plan for the best case scenario, while we know that our students will actually walk into something much more tragic and chaotic.

Scaffolding projects to provide opportunities for these sorts of curve balls can definitely be a great way to help students make the transition from school to work. It's tough to imagine how to do this in lower level courses, but this is certainly something that would help advanced students, and we should consider these approaches for our advanced groupwork or capstones.

We attempt a form of this team-based education in the Professional Practice course we run as part of the Web Dev Certificate at Seattle University. The course runs as, essentially, a big role-playing game for the first seven weeks, allowing students to inhabit the roles of a full dev team and work together.

Sam Redmond—Breaking Bad Habits: The Usefulness of Style Feedback

Redmond is an undergrad who will graduate in 2018. He has been teaching CS41: Hap.py Code at Stanford. This is an intro to Python course.

Redmond spoke about giving stylistic feedback. This is a fascinating topic for me, probably due to my history with teaching writing and art. I've long believed that accuracy is only half the job: We also aim to make things beautiful. Beauty is useful.

Key takeaway Redmond wants to convey: Style feedback should help students "think in Python."

Redmond wants us to move beyond PEP-8 compliance and aim our feedback to help students not only improve their code, but also improve their understanding. Style feedback should teach students about the language features and help them leverage those features more effectively.

It's tough to teach style in Python because Python draws from different paradigms. This causes confusion. Python also has clear bias and preference from a stylistic perspective. If we are teaching students who come from other languages, then it's important to help them navigate between all of these paradigms and considerations.

Stylistic feedback should help students "stretch their mental habits" to "begin thinking in another way."

Redmond proposes a stylistic feedback framework:

1. Pythonic Practices - "Thinking in Python"
2. Program Deisgn - Language agnostic considerations of software design
3. Python Mechanics - PEP-8 adherance

Redmond asserts that the most important aspect of the feedback is the Pythonic Practices. As much as possible, feedback should encourage use of Pythonic features such as:

• List comprehensions
• Iterables and generators
• Don't reinvent the wheel
• Custom Errors
• Magic methods

In order to address style, feedback should:

1. Identify an area of stylistic weakness in the code
2. Articulate the cause of the weakness clearly to the student
3. Offer actionable advice about how to improve the code
4. Word feedback constructively

Redmond's advice for writing feedback is, "Don't violate the dummy rule." If you can add the word "dummy" to the end of your feedback and it fits, then your feedback is probably inappropriate.

We should also take into account the students' level in giving feedback. Prioritize feedback so that students at a basic level get feedback that helps them understand big Pythonic concepts. But for advanced students, feedback can focus on more subtle improvements that build in prior knowledge.

I appreciate Redmond's points, and they are vital to teaching students. The challenge of providing helpful, accurate feedback is huge, and many teachers struggle with this part of the job.