Create human.md

falseywinchnet · web-flow · commit ea345585ddf6 · 2026-07-06T14:06:29.000-05:00
diff --git a/human.md b/human.md
@@ -0,0 +1,3 @@
+Technically, i dont know much math and im not even a great coder. What I do know is the shape of what good things look like and how to ask for that and discipline it into being. I cannot do integration in my head. I can use the operators, and i can use sum, algebra, etc. i know how to take a pointwise difference. i know geometry, angles, and therefore sin and cos. i understand logistic distributions. But AI did most of the work here.
+
+What I contributed, for a lack of a better explanation, was taste. I saw the BFFT as a novel and rich, accessible representation of the FFT. first it was implemented. Then I had to guide the discovery of the normalization basis and the right SIMD forms. then I guided the development of the DIT. I observed different ways that entropy is either gathered or distributed in the different kinds of FFT and how that interplays with the different kinds of butterflies- splits, mirrors, etc. I contemplated and decided a third walk was possible- along the diagonal. And so it is. its a phase walk. Its novel. I dubbed it the DIP. it offered some properties that immediately smelled to me like they had analytical uses. and so they did. from there, I guided along further revisions a Super resolution method. I resolved to force it to live optimizing inside the FFT itself to take advantage of the superposition properties. The models didnt intuit this. I did. many sessions later...

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+Technically, i dont know much math and im not even a great coder. What I do know is the shape of what good things look like and how to ask for that and discipline it into being. I cannot do integration in my head. I can use the operators, and i can use sum, algebra, etc. i know how to take a pointwise difference. i know geometry, angles, and therefore sin and cos. i understand logistic distributions. But AI did most of the work here.`
	`2`	`+`
	`3`	+What I contributed, for a lack of a better explanation, was taste. I saw the BFFT as a novel and rich, accessible representation of the FFT. first it was implemented. Then I had to guide the discovery of the normalization basis and the right SIMD forms. then I guided the development of the DIT. I observed different ways that entropy is either gathered or distributed in the different kinds of FFT and how that interplays with the different kinds of butterflies- splits, mirrors, etc. I contemplated and decided a third walk was possible- along the diagonal. And so it is. its a phase walk. Its novel. I dubbed it the DIP. it offered some properties that immediately smelled to me like they had analytical uses. and so they did. from there, I guided along further revisions a Super resolution method. I resolved to force it to live optimizing inside the FFT itself to take advantage of the superposition properties. The models didnt intuit this. I did. many sessions later...