Any time I find myself repeating a story to people - first my wife, then a friend or someone I'm talking to in public, and later to my dogs on one of our walks - I know I'm onto something worth writing down and working out. The dogs don't have to react, but the fact that I'm replaying reactions in my head on a walk - where the only other thing I'm doing is occasionally making neighbors look at me funny - that awkwardness is signal. When I have that happening, always subconsciously at first, I know I'm onto something.

I try to test ideas this way. Not from the get go, and not in some obnoxious way like everything is me working out a bit, but there's an attention and awareness to getting a reaction, and if you're paying attention, it's just part of life.

Dave Nadig told me he does a version of this too. I think about it all the time, I realized, and yet I never really write or talk about THIS aspect of the creative process. Which is weird. And so here we are. You can listen to our Notes on Notes discussion about the process behind his essay, Free Treasure, on Panoptica.

Another idea where Dave described an internal feeling I have way too often and don't vocalize: his LLMs have a too-nice problem. Any first read on anything creative, the LLM responds with some praise and a "good job, kiddo!" vibe. I kind of hate that, and I'm very scrupulous about what I'll let AI do and not do any time I show it a first pass.

Part of the way I fight against it loops back to the original idea. The LLM isn't stress testing my idea the way sharing it with a stranger at the bar is. The LLM ask has literally no stakes. It's so friggin easy that - that's the whole idea. And it goes a level deeper too. Beyond strangers stress testing an idea, the next level is the trusted confidant. Not a stranger, not a robot overlord.

Dave calls the trusted confidant who can give you real feedback that won't demoralize you out of your art your "first read" people. We talked about who these people are for each of us. I have my wife and a college buddy. Dave has his friend Barry and the wonderful Lara Crigger.

We talked about testing the LLMs to see if they've gotten better at a first read. You have to keep testing these tools - we both freely admit that neither of us can help ourselves. The result is still pretty unanimous. I feed it something, click the button, and the response usually sucks.

I'll clarify for those who haven't had the experience. If I start with something I’ve written that I'm still not quite sure entirely about, yet I still know I’m onto something, and the AI comes back with "Great job, I especially love the way you..." and then trails off into some garbled version of my own not-quite-clear-yet logic - all I know is I still have work to do. The mistake is running with the confused compliment. AI is just really good at BSing you. Another pass is needed. You can’t fall for it’s friendly repackaging of your mess into a tidier, faux-coherent structure.

But more than anything else Dave said in our talk, the idea that more and more of us are turning to a box that promises we won't need other people for this process - that's the killer idea here. For all the jobs AI is going to replace and all the professionals it's promising to upset, that's a harrowing sentence. The promise that we will need fewer people in our lives.

I don't want fewer people. I want more interesting ones. I want to be in their audience, and when it makes sense, I want them in mine - but only because they looked at something I did and thought it was worthy of their time. That's the only kind of networking I care about anymore. Viral might be cool but I want stickiness. And I definitely don't want a room full of bozos. Social media comment sections are good enough reminders of that.

The concept that makes me feel this hardest in Dave's essay is in how he talks about physical places. I opened our conversation there but we got away from it - and that's okay, I didn't fully know what I was trying to describe when we pressed record. Playing it back now, I see it: the places are the enablers for the people in each vignette. Barry and Camp Kotok, the fishing boat with his grandfather, the church lawn with Pete - in every case it's about the people in the place in the moment. There's nothing viral in any of it. It's pure. And it's exactly where the interesting people show up.

These are the kind of people you can test out a new story on. The kind that test out a new story on you. Not in any transactional way, just because you're there and what else are you gonna do except talk for a minute?

They do it despite what the box might say. Maybe even in spite of asking the box anything. This person in front of you is far less predictable - and maybe we are the kinds of people who just like a little chaos. We choose these spaces and these interactions because they are definitionally not what the magic wand of a little orange up arrow promises when we're out of those moments.

If you want real readers, write something real, worthy of being read. Earn that s***. Like Dave.