Conversation

Why MOST research should NOT be objective-driven. To get to any one objective, you have to build on a library of stepping-stones that were discovered based on subjective interestingness unrelated to that or maybe any objective. youtube.com/watch?v=dKazBM

Quote Tweet

Tapa Ghosh

@semiDL

· Dec 9, 2020

If you read no other book in your life, this is the one

196

Ben Reinhardt

@Ben_Reinhardt

Dec 10, 2020

Poking a bit for the sake of public discussion: I suspect by 'objective' driven you mean something isomorphic to 'deliverables.' I think most a lot of research should be objective-driven in the sense that you are trying to build or learn something.

José Luis Ricón Fernández de la Puente

I think so. The novelty seeking process at the level of the scientific system isn’t based on pure novelty seeking at the level of individuals. Also: nature has pre-existing interestingness to harness, like CRISPR, that pic-breeder doesn’t... & logic -> roadmapping + engineering.

Adam Marblestone

@AdamMarblestone

Dec 10, 2020

But there can be a lot of idiosyncratic taste to what people want to build or learn that is hard to formalize as deliverables.

Adam Marblestone

@AdamMarblestone

Dec 10, 2020

I’m interested in enabling more roadmapping, goal-driven research, & engineering... But I often find that good paths rest on stepping stones discovered in vaguely the way Stanley describes, either by nature or humans: found in a way that seems unrelated to the goal they enable.

Replying to

and

I’m curious how you see this observation interacting with the formulation of FROs?

6:15 AM · Dec 10, 2020Twitter Web App

Replying to

and

It just means that FROs are not the “best” way to do “research”. They fill one very specific gap. One that is hard to fill without a push, because of the scale it requires. There are other gaps, but those requiring less scale to fix have more work-arounds. Thus we limp forward.

Adam Marblestone

@AdamMarblestone

Dec 10, 2020

The search for FROs is part of a more general pattern, though: Asking people what they really need, to support the work they really want to do. Even if that is a new structure. Then trying to go get that. I find several problems desperately need FROs. Others need something else.

Show replies