Every Text I Send Goes Through Five Drafts
The performance looks normal. The resource running it is not.
There is a point where the distance between the version of me they are seeing and the state I am actually in has narrowed to almost nothing. Someone asks me to join in. And there is almost nothing left. What is close to being said has a specific shape. I know exactly what it is.
I do not say it.
There are a couple of competing drafts in a document.
I create another.
This is the one.
Before I copy the 32 words into WhatsApp, I run a final pass. Is it too long? Too fawning? Too abrupt? Should I use a simpler word there? Is it too familiar? Too formal? Too distant? Too warm? Everything is checked. The person who receives it has no idea it went through five versions. It looks like a normal text.
There is no signal telling me what the right version is first time. So I build toward it.
This is a process that costs. I have come to call it the manual layer. The same function that runs unconsciously in other people, running in the foreground.
It is not only text messages that receive this level of attention. The manual layer runs in most of my interactions. Before I speak, I edit hard. I redact my opinions. I drop ideas that run counter to what the group is saying. In social settings, when someone asks me directly for my opinion, I cut what I think down to what I can say without changing the temperature of the room.
What I actually think has conditions. It depends on factors I would need to lay out before the answer means anything. I give the room something smaller. I say the subject is nuanced, that I have no personal experience of it, but am interested to hear their thoughts on the matter.
The people around me are not doing this consciously. They are just talking. Not running checks before they speak. Occasionally, in particular formal situations — a board presentation, a deposition, a panel — they are actively monitoring themselves in a way they do not normally. But most of the time they trust the process that runs in the background.
The process I am running is unknown to them.
Over a day with people who I must run this process with, I notice my capacity to perform wear thin. The manual layer runs on a resource. There is an awareness that the energy sustaining it is running out. Nothing on the surface indicates it.
There is a point in a long, busy social day where the distance between the version of me people are seeing and the state I am actually in has narrowed to almost nothing. Someone asks me to join in. To come over. To engage with one more thing. And there is almost nothing left as I reengage.
What is close to being said has a specific shape.
I know exactly what it is.
I do not say it.
I know a genius.
IQ somewhere above 150.
After school, he tried to get a job using what he had taught himself about coding. He was told multiple times by different companies that without a degree they could not hire him. So he got one. I once asked if he learned anything new studying in university. He said he learned nothing, except most people were bad at coding, and that working in a group held him back.
Today, he manages a team. Everyone who works under him is scared of him.
He does not intend to create fear. He can see the most elegant solution with a glance. He holds his team to his standard. What they bring to him is never at that level. He gives them feedback on their work directly.
He does not mind that they are scared of him. He finds it funny. In his view, if they were better, the feedback would be better.
A few years ago, I started reading books about emotional intelligence. When I told him about it, he said he does not believe emotional intelligence is real. I understood what the statement was doing. Emotional intelligence was not his problem. It was theirs. If they were more capable they would not need what he provides. This is a coherent position. It holds together internally.
I had done a version of it myself.
For most of my life, I told myself that most people were conformist drones.
It was hard to falsify. People did seem to follow convention. Agreeing with whoever spoke last. Rarely thinking for themselves in any way that I could observe.
What I was wrong about was the explanation. I could not see they had a layer I did not. One they never had to build. They were more efficient. More social. Using infrastructure I did not have and did not know existed.
The conformist drone hypothesis was easier than seeing that.
It is not wrong. Most people’s perception is dictated by the consensus more than they realize. What I saw as compliance they know as common sense. The hypothesis held.
It also let me avoid looking inward.
Every conversation is managed output.
The thoughts and observations people have are known to them, but what arrives in their speech has already been filtered. What gets cut is never a decision. It is gone before the sentence forms. People automatically know what to transmit and when to hold back. The filtering happens before language, before conscious thought. As natural and effortless as reading.
In my experience, nothing gets cut without a decision being made.
For most people, the filter is not effort,or as a filter that even exists. They are simply in a conversation, and their words have already been calibrated. It runs as an invisible instinct.
This is the default.
For some, it runs in the foreground. The manual layer partitions attention in the space where other things could otherwise go. The review, the check, the adjustment, the decision, the release. Each step taken deliberately. Each step costing something. They are doing the same thing without effort. From the outside, the outputs often look similar.
The difference is what it costs.
What other people do automatically and almost for free, the manual layer does at a price.
That price is not dramatic. It accumulates across a conversation, across a meeting, across a day. It presents, eventually, as the depletion. The edge. A resource that ran out before the room did. And in complex situations with many people, the manual layer can stop working.
The structure produces the cost. The cost does not reveal the structure.
Other people know when to stop, when to speak, what to say, what to do. Automatically. They just know. There is no equivalent signal in my experience.
For a long time I did not look at people’s faces much when I spoke. Watching and speaking at the same time was too much to run in parallel. What I eventually understood: the face is where the information is. Not all the information. The most legible information. So I learned to look.
It helped. It did not solve the problem.
The problem was the process, not the looking. When the process ran lighter, there was space to focus on faces.
A face is data. It is not a signal. Reading it is still a step in the manual process, an input to observe and interpret and act on. Other people do not read faces. They receive them. The information processed effortlessly.
The manual layer has no instruction to stop. When it finds an error, it is often found later. Sometimes because someone mentions it. Sometimes because a look from yesterday surfaces at two in the morning as information. A review. What was said. What the look meant. What should have been said instead. Whether anything can be done. The process runs to completion before sleep is possible.
The off switch exists. There is just no signal telling it when.