Conversation

There is no “correct” statistical method. You have to *actually do science* if you want to figure stuff out. Science can't be reduced to any fixed method, nor evaluated by any fixed criterion. It *uses* methods and criteria; it is not defined or limited by them.

143

330

Replying to

What parts are left out then if you reduce science to the scientific method? The culture around it? Or is there no scientific method as such?

Replying to

Right: there is no scientific method

'(·)

@allgebrah

Replying to

@Meaningness

As in "there are many methods and there is no coherence to them", "no single method could hope to capture all empirically accessible knowledge", and/or in another way? (Is there a longform piece about it? Especially the latter version feels technical to show)

1:46 AM · Dec 18, 2017Twitter Web Client

Replying to

Pretty much about that:

metarationality.com

Upgrade your cargo cult for the win | Meta-rationality

Richard Feynman derided “cargo cult science” that sticks to fixed systems. Innovation requires an upgrade to fluid, meta-systematic inquiry.

Replying to

sidenote, I'm pretty sure there's ethnographies of scientists doing their thing, a friend of mine did pretty much that when she studied an academic supercomputer's administration and crew (iirc it was that)

Show replies