Is intellectual progress in the head or in the paper?
Which of the two generates more value:
A researcher writes up a core idea in their field, but only a small fraction of good people read it in the next 20 years
A researchers gives a presentation at a conference to all the best researchers in his field, but none of them write up the idea later
I think which of the two will generate more value determines a lot of your strategy about how to go about creating intellectual progress. In one model what matters is that the best individuals hear about the most important ideas in a way that then allows them to make progress on other problems. In the other model what matters is that the idea gets written as an artifact that can be processed and evaluated by reviews and the proper methods of the scientific progress, and then built upon when referenced and cited.
I think there is a tradeoff of short-term progress against long-term progress in these two approaches. I think many fields can go through intense periods of progress when focusing on just establishing communication between the best researchers of the field, but would be surprised if that period lasts longer than one or two decades. Here are some reasons for why that might be the case:
A long-lasting field needs a steady supply of new researchers and thinkers, both to bring in new ideas, and also to replace the old researchers who retire. If you do not write up your ideas, the ability for a field to evaluate the competence of a researchers has to rely on the impressions of individual researchers. My sense is that relying on that kind of implicit impression does not survive multiple successions and will get corrupted by people trying to use their influence for some other means within two decades.
You are blocking yourself off from interdisciplinary progress. After a decade a two fields often end up in a rut that needs some new paradigm or at least new idea to allow people to make progress again. If you don’t write up your ideas publicly, you lose a lot of opportunities for interdisciplinary researchers to enter your field and bring in ideas from other places.
You make it hard to improve on research debt because there is no canonical reference that can be updated with better explanations and better definitions. (Current journals don’t do particularly well on this, but this is an opportunity that wiki-like systems can take advantage of, or with some kind of set of published definitions like the DSM-5, and new editions of textbooks also help with this)
If you are a theoretical field, you are making it harder for your ideas to get implemented or transformed into engineering problems. This prevents your field from visibly generating value, which reduces both the total amount of people who want to join your field, and also the interest of other people to invest resources into your field
However, you also gain a large number of benefits, that will probably increase your short-term output significantly:
Through the use of in-person conversations and conferences the cost of communicating a new idea and letting others build on it is often an order of magnitude smaller
Your ability to identify the best talent can now be directly downstream of the taste of the best people in the field, which allows you to identify researchers who are not great at writing, but still great at thinking
The complexity limit of any individual idea in your field is a lot higher, since the ideas get primarily transmitted via high-bandwidth channels
Your feedback cycles of getting feedback on your ideas from other people in the field is a lot faster, since your ideas don’t need to go through a costly writeup and review phase
My current model is that it’s often good for research fields to go through short periods (< 2 years) in which there is a lot of focus on just establishing good communications among the best researchers, either with a parallel investment in trying to write up at least the basics of the discussion, or a subsequent clean-up period in which the primary focus is on writing up the core insights that all the best researchers converged on.
The complexity limit of any individual idea in your field is a lot higher, since the ideas get primarily transmitted via high-bandwidth channels
Depends if you’re sticking specifically to “presentation at a conference”, which I don’t think is necessarily that “high bandwidth”. Very loosely, I think it’s something like (ordered by “bandwidth”): repeated small group of individual interaction (e.g. apprenticeship, collaboration) >> written materials >> presentations. I don’t think I could have learned Kaj’s models of multi-agent minds from a conference presentation (although possibly from a lecture series). I might have learnt even more if I was his apprentice.
This was presuming that that would not happen (for example, because there is a vague norm that things are kind-of confidential and shouldn’t be posted publicly).
Is intellectual progress in the head or in the paper?
Which of the two generates more value:
A researcher writes up a core idea in their field, but only a small fraction of good people read it in the next 20 years
A researchers gives a presentation at a conference to all the best researchers in his field, but none of them write up the idea later
I think which of the two will generate more value determines a lot of your strategy about how to go about creating intellectual progress. In one model what matters is that the best individuals hear about the most important ideas in a way that then allows them to make progress on other problems. In the other model what matters is that the idea gets written as an artifact that can be processed and evaluated by reviews and the proper methods of the scientific progress, and then built upon when referenced and cited.
I think there is a tradeoff of short-term progress against long-term progress in these two approaches. I think many fields can go through intense periods of progress when focusing on just establishing communication between the best researchers of the field, but would be surprised if that period lasts longer than one or two decades. Here are some reasons for why that might be the case:
A long-lasting field needs a steady supply of new researchers and thinkers, both to bring in new ideas, and also to replace the old researchers who retire. If you do not write up your ideas, the ability for a field to evaluate the competence of a researchers has to rely on the impressions of individual researchers. My sense is that relying on that kind of implicit impression does not survive multiple successions and will get corrupted by people trying to use their influence for some other means within two decades.
You are blocking yourself off from interdisciplinary progress. After a decade a two fields often end up in a rut that needs some new paradigm or at least new idea to allow people to make progress again. If you don’t write up your ideas publicly, you lose a lot of opportunities for interdisciplinary researchers to enter your field and bring in ideas from other places.
You make it hard to improve on research debt because there is no canonical reference that can be updated with better explanations and better definitions. (Current journals don’t do particularly well on this, but this is an opportunity that wiki-like systems can take advantage of, or with some kind of set of published definitions like the DSM-5, and new editions of textbooks also help with this)
If you are a theoretical field, you are making it harder for your ideas to get implemented or transformed into engineering problems. This prevents your field from visibly generating value, which reduces both the total amount of people who want to join your field, and also the interest of other people to invest resources into your field
However, you also gain a large number of benefits, that will probably increase your short-term output significantly:
Through the use of in-person conversations and conferences the cost of communicating a new idea and letting others build on it is often an order of magnitude smaller
Your ability to identify the best talent can now be directly downstream of the taste of the best people in the field, which allows you to identify researchers who are not great at writing, but still great at thinking
The complexity limit of any individual idea in your field is a lot higher, since the ideas get primarily transmitted via high-bandwidth channels
Your feedback cycles of getting feedback on your ideas from other people in the field is a lot faster, since your ideas don’t need to go through a costly writeup and review phase
My current model is that it’s often good for research fields to go through short periods (< 2 years) in which there is a lot of focus on just establishing good communications among the best researchers, either with a parallel investment in trying to write up at least the basics of the discussion, or a subsequent clean-up period in which the primary focus is on writing up the core insights that all the best researchers converged on.
Depends if you’re sticking specifically to “presentation at a conference”, which I don’t think is necessarily that “high bandwidth”. Very loosely, I think it’s something like (ordered by “bandwidth”): repeated small group of individual interaction (e.g. apprenticeship, collaboration) >> written materials >> presentations. I don’t think I could have learned Kaj’s models of multi-agent minds from a conference presentation (although possibly from a lecture series). I might have learnt even more if I was his apprentice.
What if someone makes a video? (Or the powerpoint/s used in the conference are released to the public?)
This was presuming that that would not happen (for example, because there is a vague norm that things are kind-of confidential and shouldn’t be posted publicly).