You need to sign in or sign up before continuing.
Reflection — An Honest Take 8 min

Honest Take — Before You Begin

Honest Take — Module 8: Teaching Through Video #

I want to say something most curricula on this topic won't, because they are written by people who teach video for a living and have a vested interest in pulling you onto the platform: most engineers should not have a YouTube channel. Not because video is bad — it is excellent for some teaching. Not because the engineers would teach badly — many would teach beautifully. But because the production overhead, the algorithmic economics, and the time commitment of a sustained channel are higher than almost anyone budgets, and the back-of-envelope math for most engineer-creators is brutal: tens of hours of production per video, single-digit-thousands of views, and a growth curve that does not compound the way a blog's does.

The ceiling is enormous — a handful of educational channels are real businesses — but the failure floor is also enormous, and many engineers with genuinely high-quality content sit at three-digit subscriber counts for years. The honest budget is: assume your channel will not get traction. If you would still want to make the videos, make them. If you wouldn't, don't.

That said, video is the right medium for some teaching, and the module exists so you can make the call on evidence instead of vibes. Visualization-heavy explanations, live-coding demonstrations of a complex flow, anything where the understanding lives in watching a thing move over time — when the topic genuinely wants to move, video is right and a blog post is wrong. When it doesn't, the blog post is faster, cheaper, and more durable. The four canonical models — compressed entertainment, animation-first explainer, long-form lecture, live coding — are genuinely different practices demanding different temperaments, and the selection rule is the one the module states: pick the model that matches what you would sustain, not the one with the highest ceiling. Polished-educator video requires actually enjoying the production half of the work; if you don't, you burn out by video twelve. Long-form lecture doesn't grow algorithmically; if you need growth to stay motivated, you quit. The honest temperament match is the only one that survives Year 2.

The thing nobody tells you about live coding: you will be visibly slow. You will typo, look up syntax you "should" know, miss obvious errors while the chat sees them instantly. The instinct is to script aggressively, pre-record the hard parts, and present a polished surface. That instinct kills the medium's entire value, which is modeling the actual cognitive work of programming. The audience knows you have done this before; what they need is to see how you think when something doesn't work, because that is where their gap lives. Resist the polish.

Similarly, resist the gear: a USB microphone and OBS get you 80% of the quality you need, and the remaining 20% comes from production skill, not equipment. Most channels that quit, quit at video three with $1,500 of equipment. Ship twelve videos on the cheap setup; upgrade only when audio is demonstrably the limiting factor.

One regional note, because the landscape shapes the decision if it is your landscape. If you are teaching into the Indian audience, know that the educational-tech-video space there is vast and has a dominant pattern — placement-prep plus algorithm-interview content, often in Hindi or Hinglish — that has produced large audiences, real value for early-career engineers, and some predictable failure modes: solution-walkthrough instead of reasoning-development, placement metrics over craft, creator-as-personality-cult dynamics. If you enter that market, study it first and decide whether you are entering on its terms or differentiating against them. Both are legitimate; default-mimicry is the trap. The regional applied vertical in this collection goes deeper, and the same study-the-landscape-first discipline applies to whatever regional or niche market you are considering.

Conclusion #

Most engineers should not build a channel as a primary teaching practice; the variance is too high and the production cost too real. Many engineers should make occasional videos for topics that genuinely move. Pick the model by temperament, spend under $200 on gear, resist the polish in live coding, and after four published videos make the call — continue, pivot, or sunset — on evidence. Sunsetting is a legitimate outcome of this module; it just has to be a decision, not a drift.

Predictions #

  • Your first video will take 2-3x longer than you predicted. Subsequent videos will improve toward the prediction and rarely reach it.
  • Your view counts will be discouraging for at least the first ten videos. This is the genre's normal; do not update on it before the data window the module specifies.
  • One video — usually not the one you expected — will outperform the rest by 5-20x. Learn from what worked without converting yourself into an algorithm-chaser.
  • You will over-edit your first five videos, cutting the natural pauses that were giving the viewer time to absorb. Re-watch your own footage with the pauses left in and feel the difference.
  • You will reach for better production gear exactly once before the twelve-video mark. Don't.
  • Somewhere between video four and video ten, the continue/pivot/sunset answer will become clear, and the data will be clearer than your introspection. If both view counts and your own enjoyment are flat, sunset without shame. The skill of stopping is a skill.