It’s fascinating to think about the number of PUs (procedural units) it takes to make a modern tool. Something as simple as a modern hammer must number in the thousands and a mobile phone in the millions or billions.
You might find the essay "I, Pencil" by Leonard Read interesting. It's told from the point of view of a pencil who talks about the complexity of his own creation and all of the components involved in the process
That essay also loosely inspired the opening scene of Lord of War, which showcases the journey of a bullet from an underground mine to an Eastern European factory all the way into the head of an African child soldier.
There's also Thomas Thwaites' "building a toaster... from scratch" project that goes into the details of how even an extremely simple appliance involves materials and processes that are effectively impossible for a single person to replicate (spoiler: he has to give up and "cheat" on some things).
That idea is pretty similar to "assembly theory" no? In the sense of how much information or evolution was necessary to generate some artifact, be it a benzene molecule, a stone tool, or an iphone
The same came to my mind. I think there may be some elements of assembly which have to do with biological process that don’t apply to “accumulation of technological knowledge,” but I need to reread it.
Here, we introduce AT, which addresses these challenges by describing how novelty generation and selection can operate in forward-evolving processes. The framework of AT allows us to predict features of new discoveries during selection, and to quantify how much selection was necessary to produce observed objects, without having to prespecify individuals or units of selection.
It took humans 300k years, all of us, our collective output to reach this point. Yet people insist on comparing a human who is part of society with a LLM alone, who doesn't even have search, and very limited contexts, just closed book mode remembering.
the genius of photolithography (and its descendents) is that each chip is printed all at once, all the transistors at once over several steps for the several layers. This is what makes chips inexpensive.
alun|1 year ago
nexuist|1 year ago
brazzy|1 year ago
alickz|1 year ago
Anotheroneagain|1 year ago
levocardia|1 year ago
adolph|1 year ago
https://www.nature.com/articles/s41586-023-06600-9
Here, we introduce AT, which addresses these challenges by describing how novelty generation and selection can operate in forward-evolving processes. The framework of AT allows us to predict features of new discoveries during selection, and to quantify how much selection was necessary to produce observed objects, without having to prespecify individuals or units of selection.
visarga|1 year ago
w-ll|1 year ago
NL807|1 year ago
fsckboy|1 year ago
etskinner|1 year ago