Don't lose signal by giving it a bag of words, design a system with the desired domain. Otherwise how can you tell the difference between an image of "red green blue" and one of "green blue red"? It's too much work to imagine the underlying words in order to decide whether two potentially radically different avatars are the same. It's easy enough to try both though 🤷🏻‍♂️

What's wild isn't that this is perfect, but that it cost me less than 1¢ and 30 seconds to produce, and it's not a bad start

Claude's first take:

### Facial Structure Elements

1. Forehead slope: 5 variations (strongly receding, mildly receding, vertical, mildly protruding, strongly protruding)

2. Nose shape: 8 variations (straight, convex, concave, wavy, bulbous, upturned, downturned, hooked)

3. Nose size: 5 variations (very small, small, medium, large, very large)

4. Chin projection: 5 variations (strongly receding, mildly receding, neutral, mildly protruding, strongly protruding)

5. Chin shape: 4 variations (rounded, square, pointed, dimpled)

6. Jaw angle: 4 variations (rounded, average, square, sharp)

7. Lip prominence: 5 variations (very thin, thin, medium, full, very full)

8. Lip ratio: 3 variations (upper dominant, balanced, lower dominant)

9. Ear size: 5 variations (very small, small, medium, large, very large)

10. Ear angle: 4 variations (flat, slightly protruding, moderately protruding, strongly protruding)

11. Earlobe type: 3 variations (attached, partial, free)

12. Brow ridge: 4 variations (minimal, moderate, pronounced, heavy)

13. Face length ratio: 5 variations (very short, short, medium, long, very long)

14. Neck thickness: 5 variations (very thin, thin, medium, thick, very thick)

15. Eye depth: 3 variations (deep-set, medium, protruding)

16. Cheekbone prominence: 4 variations (flat, moderate, high, very high)

17. Philtrum length: 3 variations (short, medium, long)

### Detailed Features

18. Tragus size: 3 variations (small, medium, large)

19. Nasolabial fold depth: 4 variations (minimal, light, moderate, deep)

20. Adam's apple prominence: 4 variations (not visible, slight, moderate, pronounced)

21. Temple contour: 3 variations (flat, average, pronounced)

22. Mandibular angle: 4 variations (rounded, average, square, sharp)

23. Zygomatic arch: 3 variations (flat, moderate, pronounced)

24. Nasofrontal angle: 4 variations (sharp, average, obtuse, rounded)

25. Lip border definition: 3 variations (diffuse, moderate, sharp)

26. Mentolabial sulcus: 4 variations (shallow, average, deep, very deep)

### Asymmetries

27. Facial plane alignment: 3 variations (aligned, slight deviation, moderate deviation)

28. Lip symmetry: 3 variations (symmetric, slight asymmetry, moderate asymmetry)

29. Nostril symmetry: 3 variations (symmetric, slight asymmetry, moderate asymmetry)

30. Jaw symmetry: 3 variations (symmetric, slight asymmetry, moderate asymmetry)

31. Ear positioning: 3 variations (symmetric, slight asymmetry, moderate asymmetry)

32. Eyebrow asymmetry: 3 variations (symmetric, slight asymmetry, moderate asymmetry)

### Accessories and Modifiable Elements

33. Eyewear: 6 variations (none, standard glasses, rimless glasses, thick-framed glasses, sports sunglasses, fashion sunglasses)

34. Ear piercings: 5 variations (none, lobe only, multiple lobe, helix, multiple types)

35. Facial piercings: 5 variations (none, nose, lip, eyebrow, multiple)

36. Facial hair: 7 variations (none, stubble, mustache, goatee, full beard, designer beard, sideburns only)

37. Hair length: 6 variations (bald, very short, short, medium, long, very long)

38. Hair volume: 4 variations (thin, average, thick, very thick)

39. Hair style: 8 variations (straight, wavy, curly, coiled, updos, braided, dreadlocks, asymmetrical)

40. Hair accessories: 5 variations (none, hat, headband, clips/pins, religious head covering)

41. Neck jewelry: 4 variations (none, thin chain, thick chain, choker)

42. Visible tattoos: 4 variations (none, small, medium, extensive)

43. Visible scars: 4 variations (none, minor, moderate, significant)

### Soft Biometrics

44. Apparent age indicators: 6 variations (child, adolescent, young adult, middle-aged, senior, elderly)

45. Skin texture: 4 variations (smooth, average, textured, heavily textured)

46. Neck musculature: 4 variations (minimal, average, defined, highly defined)

47. Facial adiposity: 5 variations (very lean, lean, average, full, very full)

### Calculation of Total Combinations

To calculate the total possible combinations, we multiply the number of variations for each characteristic:

5 × 8 × 5 × 5 × 4 × 4 × 5 × 3 × 5 × 4 × 3 × 4 × 5 × 5 × 3 × 4 × 3 × 3 × 4 × 4 × 3 × 4 × 3 × 4 × 3 × 4 × 3 × 3 × 3 × 3 × 3 × 3 × 6 × 5 × 5 × 7 × 6 × 4 × 8 × 5 × 4 × 4 × 4 × 6 × 4 × 4 × 5

This equals approximately 1.8 × 10^38 possible combinations, which is an extraordinarily large number - far more than the number of humans who have ever lived or will live in the foreseeable future.

Too many variations will become too difficult for people to differentiate. You want a set of memorable characteristics and memorable variations, so when two AIs make the same caricature, you'll recognize them as the same underlying persona

Sorry, vacation brain. What I described is 96^32 ~= 2^210. Anyway, you can use AI to generate a reliable system, even though AI itself is (intentionally) unreliable

LLMs are a source of creativity. Creativity is by definition unpredictable. But, we can use creativity to design complex systems, and those systems can be predictable. I would say, have the AI come up with 32 characteristics an avatar could have. Then have it come up with 96 variations for each characteristic. Sanity check that these are likely to be differentiable. Then use these to produce imagegen prompts and see if it works: whether or not the characteristics are differentiable in the output. Now you can produce 2^128 prompts that won't be perfect outputs, but could be reliably close.

Only the army that takes the land truly owns it. Are you raising one?

HONK HONK

"Nostr is just signed JSON over websockets, so it's actually like, it's not that exciting if it's explained in that way" nostr:npub1xtscya34g58tk0z605fvr788k263gsu6cy9x0mhnm87echrgufzsevkk5s

Au contraire! This is exactly what makes it exciting

MTGAP? I think I'm skipping straight from qwerty to dictation

Trick question. I bet he would though

Just arrived and already speaking the language! You'll be buying real estate before the flight out

Does a tailor buy clothes?

Goose, build a multi band satellite imagery analysis tool with a tiling web interface.

Thanks

Moving gold in a KC-135 isn't difficult. Bitcoin is not a revolutionary reserve currency

"Who controls the past controls the future. Who controls the present controls the past."

- George Orwell, "1984"

0, position: are we there yet?

1, velocity: how quickly are we getting there?

2, acceleration: how quickly are we going faster?

3, jerk: how quickly are we accelerating?

4, snap: how quickly are we jerking?

5, crackle: how quickly are we snapping?

6, pop: how quickly are we crackling?

Very few people have an accurate measurement of our current position. Stay buoyant

If it makes anyone feel any better, ASI is going to zero day all of the things by this fall anyway.

Did you know there's a ghidra MCP?

nostr:note12qdqd00r69anpt94snyclvtxhuxvd375wy7hsn7zqp2g3rlfn5zsx8jcsd

"And the thing that probably kept me out of jail was books, because I could go read what Aristotle wrote or what Plato wrote. And I didn't have to have an intermediary in the way. And a book was a phenomenal thing. It got right from the source to the destination without anything in the middle.

"The problem was you can't ask Aristotle a question. And I think as we look towards the next 50 to 100 years, if we really can come up with these machines that can capture an underlying spirit or an underlying set of principles or an underlying way of looking at the world, then when the next Aristotle comes around, maybe if he carries around one of these machines with him his whole life, his or her whole life, and types in all

this stuff, then maybe someday after the person's dead and gone, we can ask this machine, "Hey, what would Aristotle have said, what about this?"

"And maybe we won't get the right answer, but maybe we will. And that's really exciting to me. And that's one of the reasons I'm doing what I'm doing."

- Steve Jobs, Aspen, 1983

nostr:nevent1qqsytreya3959tcxgnatp39mq4nrl842lawh4de7aq27um4fvlsfklszyptk6g7u8keq2mfq3py5vtlwxkx0nu8nxy9zcc7tdsn85jultpy0jqcyqqqqqqgnk3jm4

nostr:note1zdknppwcd5drfahxef6tt67uysktnsqjz8rxjglvszuqazugj3es2leupr