Showing 468 to 471

Nullius in verba

The Royal Society in 1660 had the slogan nullius in verba – "Take nobody's word for it". We can see it as representing a fundamental shift in mindset that we call the Enlightenment. It used to be near-universal among human cultures to believe in some sort of Fall From Grace: everything was better before, and the most solid knowledge comes from authorities like the church or someone who lived earlier who wrote something, the older the better.

Mapmakers everywhere used to fill-in the regions they didn't know well or didn't know anything about (perhaps they just tried to hide their knowledge-holes in order to sell, but I read in Sapiens: A History of Humankind that it also reflected a different mindset – they acted as if knowledge couldn't progress so what we had was as good as we were ever gonna have), but starting around this time, we see maps with blank areas clearly marked as unexplored, which invited curiosity.

Admitting what we didn't know led to the desire to find out.

But why was truth from established authority no longer satisfactory?

  • Sapiens: A Brief History of Humankind has an explanation for the post-1600 Europeans' odd (enterprising & exploratory) outlook.
    • Counterexample of the Chinese general who once took fleets as far as Madagascar but whose expeditions ended due to lack of interest from the throne. This is the norm for most societies. China didn't believe there was anything of interest far away. Thus they never discovered Polynesia or Australia.
    • Science was supported by empires with ample funds. Every Royal Navy ship brought a scientist or two just because, to document what they found. From the empire's perspective, it was also a way to buy legitimacy for colonialism, "white man's burden".
    • Dutch East India Company. Early stock market.
    • The tiny Netherlands defeated Spain because investors trusted NL finances. Spanish king unable to get loans, while NL could get all the loans they wanted. Early example of the fact that credit-ratin wins wars.

What links here

Created (2 years ago)

Say not "truth"

  1. "Different societies have different truths" – no, they have different beliefs
  2. A belief is not "true" when it matches reality – it is "pretty accurate". See also Examples of type error: a belief is a cons cell binding a proposition to a number between zero and one, never exactly zero and never exactly one, and certainly not a boolean. If your belief put 70% on something later shown real, the belief's accuracy was log2(70%) = -0.51 bits. If your belief put 30%, you did worse: log2(30%) = -1.73 bits. Accuracy is measured in negative numbers up to zero, never hitting exactly zero.
Created (2 years ago)

Terminal values ≠ instrumental values

  1. "I want my sister to live" – terminal value
  2. "I want to administer CPR on her" – instrumental value

If #1 disappears, you will not continue wanting to do #2 for itself.

Yet, English uses the word "want" for both – leads to confusion in discussions about AI.

What links here

Created (2 years ago)
Showing 468 to 471