On Wednesday, I blasted through some more Protagonist work, and started sprouting the seeds of a few more projects. Most importantly, I made further progress on thinking about how to make further progress.

More Goals Work

At Stacey's Goals Workshop, I submitted the following fill-in-the-blanks exercise:

"This batch I am working on balancing learning a new language (Factor) and getting some concrete projects completed,because I want to feel the creativity of doing something new, but I also want develop my strengths in making real things.

So far I've tried combining those by doing a project idea in Factorand the result was frustration in both areas.

Therefore, my next steps are to develop the project in a language I know (Python), and find other ways to learn Factor."

We went through a process of understanding the goals behind the goals, and other clarifications. I gained two important insights into my current problem.

  • The first one happened when I was trying to explain why I want to learn Factor. I was able to notice that although Factor has some particularly nice typing and polymorphism properties and constructs that I ultimately want to use, the biggest reason I want to learn it, and the hardest part about it, is the fact that it is a stack-based, concatenative language, which is a new and interesting paradigm for me. Answering this questioned reopened the decision point of which such language to learn, which had become the base assumption.

    The central problem that I am having with learning Factor in particular is that there are no teaching resources available, at least that I have found, that can take the learner in a guided way through progressively powerful idioms. If it is the case that there are such resources for Forth, which seems likely, then it might be a good strategy to learn Forth first, and then I will be able to transfer that skill to using Factor for its higher abstractions. A few people suggested this to me before it sank in as a Good Idea. One Hacker Schooler has even offered to pair with me on writing a Forth interpreter.

This brings me to my other insight.

  • The higher problem that I have been struggling with is how to balance my conflicting goals. I have tried tying them together, and it seemed suboptimal.

    (I had a similar problem when I tried to learn Haskell through Project Euler problems — I did much better Haskell learning going through part of a course designed to introduce the language in graduated steps than to try to figure out what I need as needed. The strategy of doing projects in a language to learn it seems to be appropriate only after a certain base level of competence.)

    If I do two things at once that are not tied, then I have to choose how to allocate time, and when to switch. A colleague who was listening to my problem suggested that I make switching be goal based, rather than time based. That is, I could work on Protagonist until I reach some minimal satisfying feature set, and then do something goal-based for learning a concatenative language, and then if I want to do another round of features on Protagonist I can. Or I could do something else. This seems quite sane.

Hacker School Mothers

I got to chat with two other women yesterday who came to Hacker School leaving children behind, and who are facing or have faced the challenges of building a post-motherhood career. This helped me process some feelings of guilt and fear that have been lurking.



  • Protagonist: write out some specs, and continue implementing.
  • Work with another Schooler on machine learning algorithms.
  • Attend Stacey Sern's Goal Setting workshop.


  • Protagonist:
    • Updated the README to reflect what I had in mind so far.
    • Implemented several functions and tests.
    • Figured out the algorithm I want to use for parsing boolean queries.
    • Worried a bit about the problem of the file_id not being human readable.
  • Machine Learning:
    • Talked about how to get started. We wanted to get some real data, and then apply some machine learning to cluster or organise it.
      • So we started learning how to use the Python requests and b24 (BeautifulSoup) modules.
      • I pitched one of my favourite algorithms that I've been wanting to implement for a long time, which is LDA (Latent Dirichlet Allocation), to model the topics in a collection of websites. We could then cluster the documents by topic distribution.
  • Goal Setting
    • As detailed above, attending Stacey's workshop helped me to tease out a few more things about my current directions, which was very helpful.