Minutia Matters — Hello World Year 2
Why I Write the Blogs I Write
I’m currently buried in a full-semester course that’s being run over just four weeks as part of my M.S. in Health Data Science program, so I might not get back to a time-consuming blog post till February. This is really a shame, because the back propogation / deep learning blog I’ve got forming in my head is going to be pretty cool, I think (if you’re into “wow, maths are amazing” kind of revelations — and all three of you are really going to enjoy it.) That said, at the start of the new year I wanted to pull back just a bit and reflect on why I do this. In particular, why do I spend so much time going over algorithms and approaches that have been around for a long time, with countless books, blog posts, and YouTube videos available that presumably cover the same ground? Does my perspective and approach actually add anything of value to the collective body of work? Or put another way, is the juice worth the squeeze?
My first thought — and if you’ve been a reader for a while, you probably see this coming — is I really don’t care. While I have you, the reader, in mind while I write, I’m not actually doing this for you. My blog project is me “learning out loud” (please see my first blog post for context) so my primary goal in writing these is to force myself to learn this stuff well enough to be able to explain it to someone else. This inevitably deepens my own understanding of the concepts, and I often find myself correcting misconceptions I had formed mid-writing as I put together the examples to prove the points. So whether or not anyone else finds this valuable, there’s enough value in this just for me to continue doing it.
There Are Gaps in “The Literature”
That said, I do think my approach and perspectives are relatively unique, even for well-worn topics.
The Last Mile
Compared to most blogs, I strive to be a “last mile” writer — meaning, I don’t just cover a topic to the same depth that most folks do, but I work decently hard to come up with ways to explain things so that you really understand how something works under the covers without being required to understand fancy math notation and so on. (Or, if you do, I go out of my way to explain **that** before moving into it…) If I’m honest, when I read most other blog posts on topics that I write about, I’m not completely convinced that the writer actually has a firm grasp on the concepts. Most folks keep it high level, seem to start and stop in the same basic places, and throw out a bunch of formulas to “explain” the concept without actually digging into them in any meaningful sense. When it’s a new topic for me, I’m usually left going “but how does that work in the real world?” They could just be parroting surface-level phrases and concepts and you wouldn’t know the difference, in many cases. In contrast, I work pretty hard to simplify things as much as possible and provide as much direct, discrete evidence as to how something works, including math, code, and (poorly drawn) visualizations that build on the concepts.
On the flip side, you have the writers who **obviously** know what they’re talking about to the nth degree, but an awful lot of them have apparently lost the ability to talk about the topics in a manner that a person not already deeply embedded in the field (i.e. normal humans) would understand. In this sense, I strive to be a “bridge” writer — filling in the gaps between the academic complexities of the topics in a way that doesn’t require a super-deep understanding of the entire field in order to parse. You should be able to start at the beginning of my blog, read through a series of related posts, and assuming you have a basic grasp of high school algebra and geometry, in most cases get everything that I’ve been saying. (And if not — you should be asking more questions…)
In short, I don’t think you should have to choose between completeness of answer and understandability. This means I put a lot of effort into comprehending complex subjects, and learning them well enough that I could explain them to an average (motivated) high school graduate with successful understanding. I don’t know that I always nail it, but I think I do better than most folks in this regard (with a nod to Josh Starmer / StatQuest — that dude is a kindred spirit, and hella smart…) and thus I think there’s value in my work for folks who are starting in the same place that I did/am.
You Know Where I Stand
A third benefit is that I’m not trying to put on airs, and you know exactly what I mean when I say “I work in AI/ML” etc. Whether or not I’m an expert on a given topic can probably be answered by the question “Has Jason blogged about it yet?” I’m not trying to pretend to be smarter than I am in any given area. On the contrary, I’m putting out content that may very well be inaccurate, in part for the hope that someone catches my error and corrects me. But I’m also producing evidence of what I’ve learned, as well as how well I have learned it, and that’s important too.
If you’re following my blog, you are on a learning journey with me, and if work/school scheduling allows, in as near real-time as I can get. I’ve worked in and around data science for a number of years, even written introductory course materials on it in past lives, but I’ve spent exactly one year now focused heavily on learning **how** it actually works on a fundamental level. The general broad strokes I’ve been pretty solid on for a while (being married to a data scientist has benefits…), but the hard stuff? I’m getting there — making what I believe is good progress, but fully aware of how far I still have to go. By this time next year, I will likely have finished all of the “eat your vegetables” core algorithms and approaches stuff and moved into more complex topics like ML Ops and more advanced tooling and processes. But it was important for me to understand the core stuff exceptionally well — perhaps better than a number of data scientists in the field do — so that when I start writing about complex workflows and architectures, I’m writing from a solid understanding and not just relying on an 80% understanding to get me from point A to point B. I believe it will make me a better writer *and* better at any job. It means I’m taking 12–18 months to get to “the good stuff” when I could have skipped this step, but in the long run I think I’m going to be better off for it.
For better or worse, I am a WYSIWYG kind of person. I think I have value and add to the conversation in unique ways, but even if that’s just a nice delusion on my part, it doesn’t matter. This is valuable enough to me, for me, to keep going, and if you happen to benefit as well, that’s just a bonus.
I hope you’ve enjoyed the ride in 2021, and am looking forward to another interesting year of thought processing in 2022. Be safe out there, and I’ll see you again closer to February.