This transcript was produced by robots and left as-is. Accuracy and elegance are not guaranteed.

Someone you’ve never had a real conversation with has a working model of you. They use it to decide if you get brought in, if you get trusted, if you’re the one things get escalated to. And they built this from secondhand signal over the course of months. You don’t get to opt out of being modeled. You only get to decide how much of the input is deliberate.

The last four episodes were about individual communication acts. Episode 18 was when you answered a question nobody asked, real time, in person. When verbal diarrhea gets the better of you and you give every detail, whether the recipient needed it or not. Episode 19 was about written communication and how you can bury the signal in volume.

Episode 20 was about going quiet, providing no signal at all, while the people around you fill the gap with their own story.

Episode 21 was about having the right information, but presenting it in the wrong currency for the room. Each one of those was a single moment failing. And that’s what engineers focus on. The actual problem isn’t any one moment. It’s the pattern. Communication isn’t an event. It’s a continuous feedback loop.

Other people are running a predictive model of you. Every interaction is a data point. So is every interaction they expected but didn’t get. That model is what gets used in rooms you’re not in and rooms you’d like to get into. Sometimes that model opens the door. Sometimes it slams it shut. And oftentimes, you don’t know which one happened.

You need to feed that model deliberately though. Otherwise, it gets filled from the leakage.

This is system identification. Inputs and outputs. Their model of you is the output. Your behavior is the input. If you don’t design the inputs intentionally, the model gets built from the noise floor. Your worst meetings, your late replies, your wrong currency answers. Your deliberate signal raises the signal to noise ratio.

A few years back, I was having a beer with a competitor. Over the course of the evening, he had more than I did. Got a little loose with his comments. After probably one too many, he stopped, looked me in the eye, and told me that some of my clients and former clients thought I was a real piece of work. My words are kinder than his were. It hit me twice. Once because nobody had said that to my face before. Particularly not that specific phrase.

The second time was because I thought I had built really strong relationships and a really strong network. The reflection that followed was uncomfortable. I’d been counting only the deliberate signals that I’d sent. The times I stood up for my team. The times I held the line because I knew I was right, or at least thought I was. The clean proposals. The strong wins. What I hadn’t accounted for was what some of those moments looked like from the other side of the table.

When I added in the leakage to that equation, those vague responses or late replies, or sometimes wrong currency answers, it started to paint a different picture. I’d never asked for feedback from any of those relationships. I’d assumed no news was good news and that things were going well. That moment made me reflect on how much of the model I’d built had been deliberate and how much had been leakage.

Most engineers think they have a communication problem. In reality, they have a system problem.

One bad meeting, that’s forgettable. Three over six months, that’s a pattern. And patterns become models. The model becomes the decision someone makes about you when you’re not there. Whether you’re a fit for the project, whether you’re a candidate for the raise, whether they want to work with you again or at all.

Back in episode 17, we talked about your decision surface, and I recommended picking three people from it. Take those three. For each one, predict the model they’re running of you. Not the one you wish they had, the one they’d actually predict from the signal you’ve sent over the last 90 days.

If you can’t predict it, your inputs aren’t deliberate. If the prediction comes back too rosy, maybe take a pause. Make sure you’re not just counting the deliberate signal and ignoring the noise. That’s the trap I fell into in the story over the beers with the competitor.

If you can predict it and you don’t like the answer, remember those inputs are your lever to make a change.

The last four episodes were the mechanics. This is the system those mechanics feed. Three systems, three jobs.

Episodes 10 to 13 built the internal operating system. Really, how you operate when nobody’s watching.

Episodes 14 to 17 built the relationship system. How you connect with people directly.

Episodes 18 to 22 built the communication system. What carries you across distance and time to the people you haven’t met yet, in rooms you’re not in.

Three systems built by an engineer for engineers. Use them.