Thread by @Ben_Reinhardt

1/ Recently read "The Decline of Unfettered Research"

It has a ton of sharp insights about how science/technology creation has changed that I hadn't seen before.

The uniqueness, coupled with the fact that it was written in 1995 makes it seem worth a PAPER REPORT THREAD. 🧵

2/ First, the term "unfettered research" (and the converse, fettered research) feels important because it acknowledges that the external incentives on researchers are important.

3/ The first sharp insight (and core thesis) is that, despite what we might like to think, research has become much more of a commodity.

4/

5/ The story goes, more demand for research -> more researchers -> more competition -> commoditization.

6/ As a result of more researchers, more things are being explored simultaneously - which seems like a great thing, but leads to decreasing profit margins on research.

Sobering thought: what if (things like) room-temperature superconductors died *because* of this effect?

7/ It's also good to note that the platonic ideal of unfettered research never quite existed in the first place (but there was still more of it than today.)

8/ It's also worth considering whether science shifted from more of an absolute game to more of a relative game thanks to increased numbers -> competition -> metrics-dominated career prospects.

9/ Another point is that in large research projects like CERN, there is only so much room for unfettered exploration because of coordination costs.

I like to think of it as much less room to go "huh, that's funny..."

10/ This feels accurate. SpaceX and recently GPT-3 may be exceptions, but they had less direct effects than the polio vaccine. It might be a double-whammy of fewer hits, and fuzzier connections between those hits and unfettered research.

11/ The explosion in research has created more solutions to the same problem. Again, this seems like a good thing but it might show up in numbers as research becoming less efficient.

12/ Arguably systems and infrastructure have become much more complex over time, so the interoperability requirements keep going up, so the work you need to introduce a new technological paradigm keeps going up as well. Regulations have increased as well.

13/ An uncomfortable idea I've been playing with is that greater understanding of how markets and innovation work has led to more "rational" decisions, which kill outlier outcomes that are enabled by irrational decisions.

14/ The other killer insight is that the paradigm of progress has shifted from "discrete improvements" to "continuous improvements."

Again, a good thing with uncomfortable secondary effects. If you know the system you have now will improve soon, why try something new?

This mindset feels familiar. It’s important to place it in the framework of “what are the consequences of people expecting a continuously improving world?” Start going down this rabbit hole and you quickly end up at “embedded growth obligations” - another story for another time

16/ It’s a sobering thought that a radical idea that creates a 25% improvement isn’t good enough, but squares with folk startup wisdom: “anything less than 10x improvement isn’t worth it.”

17/ Parts of the paper feel weirdly prescient. Remember: this was written in 1995.

18/ Again, knowledge of how markets and innovation work may be leading to rational decisions.

19/ It’s always interesting to come across case studies that are almost forgotten now but were presumably well known in the 90’s. I wonder if Theranos and Juicero will have a similar ring in 25 years.

20/

21/ Not a fad.

22/ A fascinating perspective that the same forces that led to many good things like lean development and MVPs also led to the decline in long-term research.

23/ Note that unfettered research has positive externalities even in the context of national economies. Yet there aren't many international research organizations.

24/ Not even universities are spared, thanks to specialization (again, normally a good thing.)

25/ Perhaps one of the few things we can agree on in economics in addition to supply-demand curves is that competition drives specialization. Normally this is taken to be a good thing. However, specialization of scientific disciplines seems to be more problematic. Why?

26/ A big theme of this paper is how much expectations inform our actions. In this case, knowing that cryptography technology was ready to go when it was needed enabled people to move faster. What missing background knowledge is slowing down progress today?

27/ Let's end with some optimism!

end/

In all, "The Decline of Unfettered Research" raises some excellent and sobering points that are worth thinking more about.

If you made it this far, you might want to just read the whole paper:dtc.umn.edu/~odlyzko/doc/d…

Podcast with Andrew (the author) coming soon!