[Links] Serendipity and discovery, and a tool for making progress on hard questions/problems

post by arikr · 2019-08-14T21:43:09.705Z · EA · GW · 2 comments


One survey of patent holders (the PatVal study of European inventors, published in 2005) found that an incredible 50 percent of patents resulted from what could be described as a serendipitous process. Thousands of survey respondents reported that their idea evolved when they were working on an unrelated project — and often when they weren’t even trying to invent anything.


Kenneth Stanley: Why Greatness Cannot Be Planned: The Myth of the Objective

I believe these links could be useful to people in EA who are focused on hard questions where the answers (and the process to get to the answers) aren't yet obvious. I think the NY Times article provides the "why" and the YouTube video provides the "how."

My sense is that this is a highly undervalued tool/method for tackling hard ambiguous questions where the process to get an answer/solution isn't yet obvious, and it seems that EA is full of questions and challenges like that!

So, I hope this is a useful model to add to your problem solving toolkit as you work on doing good!


Comments sorted by top scores.

comment by Aaron Gertler (aarongertler) · 2019-08-15T12:32:27.777Z · EA(p) · GW(p)

What does it actually mean to have "serendipity" as a model in one's toolkit? Would you be open to writing a brief summary of the "how", or do you strongly recommend just watching the video?

comment by arikr · 2019-08-16T16:23:20.966Z · EA(p) · GW(p)

I think the "how" is roughly:

If you do not know the steps to your goal with high confidence, then do the following:

You can imagine that you're looking at a map, and your distant goal is somewhere on the map, but the map is blurry / not yet revealed all the way to your distant goal

So then identify what options you *do* know the steps to (the ones that _are_ visible on the map), and then pick the option from those that is most novel

This is because the more novel it is, the more likely it is to reveal large and unexpected portions of the map, potentially including the part that gives you a visible path to your distant goal

So when uncertain, identify the most novel thing you know how to do/achieve, and repeat that, and that's likely the best (albeit very roundabout!) route for getting to your distant not-yet-visible-path goal.

If the above is intriguing, I'd highly recommend watching the video – I think it's a very well spent 15 minutes if watching on 2x speed.