Experience well-being and household income seem to follow a logarithmic curve. This is unexpected to me! https://i.imgur.com/4cMP5vs.jpg

🇧🇷🇺🇸-  Book: Making Things Think: https://holloway.com/mtt. Investor in Wander, Carry, Footprint, Merkle Manufactory (Farcaster), Dynamic, Paragraph.

Common myth - the log increase signifies diminishing bounded returns, some fairly-low good enough point.

Reality - this data indicates UNBOUNDED happiness returns vs increasing wealth, all else equal.

So what does the log part signify then? = Simply that returns  percentage-relative, not absolute-value-relative.

Same idea with the mind blowing scaling law curves seen with large language model AI.

Why mind blowing? 

Those curves strongly imply unbounded intelligence potential.

Not human-level-bounded maximum intelligence bounds, as people instinctively assume.

Incredible. https://i.imgur.com/YlYpIqh.png

^^ Funny, just noticed that original Neural Scaling Laws paper makes this same bad-intuition mistake!

"though performance must flatten out before reaching zero loss"

Well, here, "reaching zero loss" = "reaching infinite skill".

So no, no math reason why curve must ever flatten out, bc infinity can't ever be reached!