Lesson 1: The Algorithm Is Just a System
a lot of people talk about "the algorithm" like it's a wizard behind a curtain, making arbitrary decisions about who gets seen and who doesn't. it's not. it's a system with inputs, weights, and outputs — exactly like a recommendation engine you'd build yourself if someone asked you to keep users on a website for as long as possible.
if you can read a control flow diagram, you can understand how platforms decide what to show people. let me show you what's actually running.
The cold truth: platforms optimize for one god metric — dwell time plus engagement velocity. not views. not likes. not quality. a post that gets 30 meaningful interactions in 5 minutes beats a post that gets 100 in 6 hours. the algorithm doesn't care about good. it cares about engaging right now.
The Real Distribution Funnel
every platform follows the same lossy pipeline. the nouns change but the control flow doesn't:
impression → click/stop-scroll → consume → react/comment/share → follow → repeat exposure → trust → intent → conversion
Each arrow is a lossy step. Your job isn't to "go viral." Your job is to lose fewer people at each arrow. That's it. That's the whole game.
breaking down each step
impression → click/stop-scroll. someone sees your post in their feed. they either scroll past or stop. the first line — or first 3 seconds of a video — determines whether the arrow fires. if they scroll past, the rest of the post might as well not exist.
click/stop-scroll → consume. they stopped scrolling. now they need to actually read, watch, or listen. this is where structure and pacing matter. a wall of text loses people. a rambling video loses people. format for scanability.
consume → react/comment/share. they finished. now what? if your post was interesting but not actionable, they close the app and move on. no signal gets sent back to the algorithm. you need to give them something to do — a question to answer, a take to argue with, a like to give.
react/comment/share → follow. they engaged once. will they do it again? your profile, pinned post, and recent history need to communicate "there's more of this coming." if they land on your profile and see inconsistency or nothing relevant, you lose them here.
follow → repeat exposure. now comes the compounding. they see your stuff again. and again. if the quality holds, trust builds. if it doesn't, they unfollow — or worse, they stay followed but stop engaging, which signals the algorithm to show your content less.
trust → intent → conversion. this is the longest arrow and the one nobody talks about in "growth hacking" threads. trust takes time. nobody buys a course, subscribes to a newsletter, or hires you off one good tweet. they buy off a pattern of being helpful or interesting over weeks and months.
What Gets Surfaced, What Gets Buried
the three signals that control your reach
| signal | what it actually means | how to optimize |
|---|---|---|
| engagement rate | reactions divided by impressions, measured fast | say something polarizing or surprising in the first line |
| dwell time | how long someone's eyes stay on your post | format for scanability — short paragraphs, bold pull-quotes, visual breaks |
| conversation depth | replies and nested threads, not single-word comments | end posts with a question or a hot take, not a summary |
Engagement velocity is the metric that matters most in the first 30-60 minutes. Platforms use a "velocity score" — if your post gets engagement fast relative to its impressions, the algorithm expands its reach. If engagement is slow, reach gets capped. Think of it like a rate limiter with a burst allowance.
what gets algorithmically suppressed
- external links — platforms want users to stay, not leave. link-in-bio exists for a reason. posts with links in the main body get lower reach on twitter, linkedin, and facebook. on instagram, links are basically invisible unless you're in stories.
- screenshots of text — OCR detection is real. platforms know when you're posting text-as-image. they'd rather you actually type the thing.
- repetitive formatting across posts — pattern matching flags bot behavior. if every post uses the exact same structure, the platform assumes automation and throttles you.
- low-signal engagement — getting likes without replies, or replies that are all one-word ("nice", "good", "🔥"). the algorithm can distinguish between real conversation and empty engagement farming.
- high ghost ratio — impressions without engagement. if your content gets shown a lot but nobody interacts, your ghost ratio rises. high ghost ratio = reach gets capped.
The ghost ratio is the quiet killer of small accounts. Most people think "at least people are seeing it" — but impressions without engagement signal to the algorithm that your content is boring. The platform stops showing it. You're better off with fewer impressions and higher engagement rate than more impressions and zero interaction.
Why Early Engagement Velocity Matters So Much
think of the algorithm like a feature flag rollout system. when you publish a post, it goes through a canary deployment:
- t+0 to t+15min: shown to a small test audience (your most engaged followers, or a random sample)
- t+15min to t+1hr: if engagement velocity on the canary is high, expand to a larger audience
- t+1hr to t+24hr: if momentum holds, push to wider distribution (explore pages, trending, "for you" feeds)
- t+24hr+: if the post generated enough total engagement, it enters long-tail discovery (search, recommendations, "top posts" sections)
The curve is ruthless. Most posts die at step 1. They get shown to a few hundred people, get single-digit engagement, and the rollout stops. The shelf life of a post that doesn't hit is measured in hours. The shelf life of a post that does hit can be measured in weeks or months.
this is why timing matters. posting when your audience is active (not necessarily when you're active) gives you a better chance at that initial velocity window. it's also why reply bait — ending posts with a question, a hot take, or a "what do you think" — works. it accelerates step 1.
Why This Matters for Engineers Specifically
you already understand everything in this lesson. you just don't know you do.
the distribution funnel? it's a multi-stage pipeline with error handling. each arrow is a function that can return null. engagement velocity? it's a rate-limited throughput metric. ghost ratio? it's a signal-to-noise calculation.
Reframe: stop thinking of yourself as "posting on social media." start thinking of yourself as deploying content to a recommendation system and monitoring engagement as your observability layer. every post is an experiment. every engagement metric is a log line. every pattern you notice is a system behavior you can exploit.
the difference between people who succeed at organic distribution and people who don't isn't talent or luck. it's whether they treat the platform as a black box they hope performs well, or as a system they're learning to operate.
Check Your Understanding
Practice: Run Your Own Audit
Pick the platform you want to build presence on first. Do this audit right now.
- Find 5 posts in your niche that performed well (high engagement relative to the account's normal output). For each, note:
- What was the first line?
- What format was it? (list, story, hot take, how-to)
- How did it end? (question, statement, CTA?)
- What was the comment section doing?
- Find 3 posts that flopped (low engagement relative to account size). Same analysis.
- Pattern-match. What do the successful posts have in common that the flops don't? Write down 3 patterns you observ.
This is not busywork. This is you reverse-engineering the algorithm in your specific niche. The patterns you find here will directly feed your content strategy in later lessons.
Why this exercise matters: most people skip the audit and start posting blindly. that's like deploying code without reading the existing codebase. you'll write things the platform doesn't reward and wonder why you're invisible. 15 minutes of research now saves months of screaming into the void later.
1. 1. What's the primary metric platforms optimize for?
2. 2. Why does a post with external links typically get lower reach?
3. 3. What is the 'ghost ratio'?
4. 4. A post that gets 30 retweets in 5 minutes beats a post that gets 100 retweets in 6 hours because:
5. 5. In the distribution funnel, what happens at the 'follow → repeat exposure' step?