My father worked Mexico City approach for 41 years. I’ve written before about why I started with that airspace. He is not a consultant I hired or a paper I read with an LLM. He is the person I am trying to build this for, and he is also the person who catches me when I get it wrong.
Nothing ships without passing him. He tests on an iPad. He is not a programmer, so he never tells me how to fix anything. He tells me how it feels, which is harder to argue with and impossible to fake. When I change something small in the airplane performance characteristics, he notices.
I rebuilt the speed model more than once. I kept trying to simplify the model enough to ship. He kept bringing me back to what a controller actually sees on the scope. Two days in a row, he sent me back to adjust the same calibration because it still didn’t feel realistic. Just this week, the 737s and A320s were holding Mach .88 up high, when real pilots keep those between .76 and .80 and only heavies like a 787 or A350 get near .90. A medium jet doing .88 looks wrong to him before he checks the number on the tag.
He does not hand out praise. After weeks of getting the speeds right, he played for a while and wrote back: “ya está listo, quedó perfecto.” It’s ready, it came out perfect. Another time, after a long stretch of voice work, “fue lo máximo,” roughly, it was the best. You earn that by being right about the thing he spent his career doing.
A simulator can look convincing in a fifteen second video. The bar I actually care about is different: a man who did the job for four decades looks at it and says it feels like the real deal. You can’t add that as a feature later. It comes from building something good enough that an expert recognizes it.