google-site-verification: google61e178fb7836a7e6.html
 

Improving Science Discussion Group

Introduction

Scientific and technological progress has been one of the dominant drivers of progress over the last few centuries. Future advances—such as in AI, medicine, and biotechnology—are compelling reasons for optimism about the future. On the other hand, some types of technological progress may also be a major source of risks to humanity. Scientific institutions, therefore, play a central role in the shaping of our future.

 

In this reading group, we will cover some key issues with the modern scientific enterprise, and how scientists are increasingly forced to prioritize self-preservation over pursuing the best questions and making scientific progress. We will cover key questions about how we should think about and prioritise reforms to scientific institutions, assess and identify key gaps in our understanding of the enterprise, and cover recent proposals for reforming scientific institutions. Other topics that we’ll cover include the recent field of the Science of Science, differential technological progress, how AI might speed up scientific progress, and how gambling might help achieve an honest consensus in science.

 

The reading group is an opportunity to cover a wide-range of original work, discuss these with a group of bright and motivated peers, and occasionally have Q&A sessions with researchers.

 

Below is the provisional curriculum with core reading (pieces that are marked with an asterisk are strongly recommended), and further optional reading.

Week 1: Issues with modern science

Core reading

 
Key Questions
  • What are some of the key issues with modern science, and what are their causes?

  • How much do key issues (such as ‘closed science’, poor research design, peer review delays, failure to publish research) slow down progress? Can we estimate this?

  • How confident can we be about estimates of scientific waste?

  • Which of the issues with modern science are most egregious/largest in scale?

  • How much would better scientific priorities regarding which problems to solve improve the value of scientific output?

  • How do the prevalence of these issues differ by scientific disciplines?

 

Optional reading

Week 2: Understanding reforms

Core reading (Reforms)

 
Key Questions
  • What type of experimentation with new scientific institutions should be prioritised?

  • How much should we reform vs. study? Do we have enough information to make sensible changes? 

  • How much should we focus on incremental approaches (improving standards for open science, peer review framework, etc.) vs. larger institutional changes (e.g. reforming how science is funded, how scientists are judged/rewarded)?

 

Optional reading

 

Core reading (Focused Research Organizations)

 

 
Key questions
  • Does academic science lack coordinated engineering or system-building capacities? If so, does this constrain them from generating catalytic ‘public goods’ for science?

  • Do commercial labs fail to generate ‘public goods’ for science? If so, why not just incentivise them to produce more of that type of work?

Week 3: Science of Science + AI & Science

Core reading

Topic 1: SciSci

Topic 2: AI & Science

 

Key Questions (SciSci):

  • What are some substantial results out of SciSci?

  • How much of the scientific budget should be spent on researching science itself?

  • Wil SciSci lead to proposals for better ways of organising science? 

  • What is the ‘theory of change’ for SciSci?

Key Questions (AI & Science):

  • How big of a deal will AI adoption in science be, over the next 5-10 years? How about over longer time-frames?

  • What AI applications in science are likely to be impactful?

  • What are bottlenecks to AI adoption in science?

  • What are the likely effects of AI adoption in science?

 

Optional reading

Science of Science

AI & Science

Week 4: Differential Technological Progress + Idea Futures

Core reading (Differential Progress)

Optional reading

 
Key Questions
  • How likely is faster scientific and technological progress to result in an increased amount of suffering?

  • How tractable is concerted differential technological progress?

  • What types of strategies are available to induce differential technological progress?

Core reading (Idea Futures)

Optional reading
Key Questions
  • How could Idea Futures improve incentives for being right?

  • Could one fund research effectively using Idea Futures?

  • Don't science questions resolve too slowly relative to how long a bettor wants to tie up their money?

  • How should we decide which Idea Futures markets to make?

  • Why don’t Idea Futures markets exist?

Further reading

 
google-site-verification: google61e178fb7836a7e6.html