The Need for Long-term Research
Author: Kevin Kelly
Editor’s Note: this essay was submitted to SoS and went through our community-based peer review process. The article received 13 “yes” and 14 “no” votes and received numerous comments from the gardeners which were also mixed in their feedback. Ultimately, we decided against publishing the article in our peer-reviewed format and instead decided to publish it here on our Substack where our criteria are a little less stringent (the main critique was the article’s brevity, something which doesn’t matter as much for our blog format). Despite the rejection, we believe this essay and the comments it generated make for a fascinating and scientifically valuable document and we are excited to share it with our Substack readers—enjoy!
Kevin Kelly is Senior Maverick at Wired, an award-winning magazine he co-founded Wired in 1993. He is co-chair of The Long Now Foundation, a membership organization that champions long-term thinking. He is founder of the popular Cool Tools website, which has been reviewing tools daily for 20 years. He is also the author of multiple best-selling books about the future of technology. His newest is Excellent Advice for Living, a book of 450 modern proverbs for a pretty good life. He is best known for his radical optimism. This essay appeared originally on his blog, The Technium.
The Need for Long-term Research
I am imagining a type of institution that would specialize in long-range research. These institutions (plural) would manage research that had a 25+ year horizon. This might be “pure” basic research with no apparent value within 25 years, as well as long-term longitudinal studies that take 25 years or more to complete. Some programs would be large-scale projects with ambitious decades-long goals. Others would be small individual limited projects with a single PI. Research with a 25-year horizon to either completion or payoff would create a generational research agenda.
At first approximation, this is the type of research that universities are supposed to be doing now. However it is rarely done by universities. Rather the typical research project might last 4 years at the longest because that is the max duration of a PhD and postdoctoral researcher. An average PhD candidate wants their own unique project; they don’t want to continue someone else’s project. So the academic world is a rapid succession of short-lived projects. There are many advantages to this frothy, bubbling, rapid pace, but what we do not get are the results gained from ambitious multi-generation work. There are some hard things that require long-durations to reveal, and many other discoveries that require work beyond the scope of single investigators.
Long-range research is the type of work that is sometimes funded by Darpa in the US, NASA, and some international agencies and research programs. Sending telescopes and probes into deep space, or building high-speed particle accelerators are long-lived research agendas. These are also huge-budget projects, with attendant high-profile politics. We need more long-term research projects that require smaller budgets and resources to endure. That would help make them common rather than rare.
Some of the research labs of big tech companies (AT&T, Microsoft, IBM, Google) are occasionally engaged in this extended time horizon. This long horizon is harder to justify to their stockholders, so the percentage of long-range research they do is a small fraction of all their research, and it is rarely on the 25-year horizon, but in the mix this longer-term commercial research is helpful and important.
We can’t really expect corporate enterprises to develop 25-year programs. Governments should continue to fund as much long-horizon research as they can. I see two frontiers where long-range research might blossom. One is in international cooperation. Pooling resources from dozens if not hundreds of nations reduces political risk and makes it possible to fund things that may never pay off or pay off in another generation. More likely, non-profits can operate here. In many ways, they are the best-shaped organization to do multi-generational work. They don’t have to show results next quarter, they are free of many political mandates, they can make failures without punishment, and they can aim for the future.
We might create incentives for non-profits to focus on long-term projects. Perhaps in addition to the usual tax advantages, we give to non-profit donations, maybe there are additional advantages if they aim to “payout” the next generation. Maybe a donor gets twice the tax write-off if the chosen non-profit funds long-term research, or long-term projects with a plus 25-year horizon. For example, maybe the non-profit is doing solar-panel material research that might take a generation to arrive, or perhaps they want to develop a system of planetary ocean monitors that would take a generation to install and track. We could invent tax and legal incentives to promote this kind of time horizon.
Today, long-term research is rare. It is hard for corporations and national governments to do. However, we could encourage non-profits, universities, and international agencies to do more by developing additional incentives and by promoting their successes when they do so.
Wanpeng Tan (PhD in physics):
The article discusses an interesting topic about long term research support. However, it lacks depth in discussion of the main idea and ambiguity of the key concepts also makes it difficult to be published as it is. Therefore, I'd suggest a major revision before publication. Here are the points I'd suggest the author(s) to consider.
First of all, what kind of long term support the authors are imagining? A fixed "exactly planned ahead" funding scheme or a flexible one that should be renewed with possible modifications? I guess that the latter should be preferred, at least for large and medium-sized long term projects. The article sort of touches on the size of projects but does not elaborate on it.
Actually, the funding agencies have done fairly good jobs on supporting both large and medium-sized long term projects, some of which are mentioned in the article. Other examples include LIGO, LHC, and other large facilities in physics research that has been supported for many decades. So are many medium-sized collaboration teams, at least in the field of physics which I am more familiar with. Many of such collaboration units (typically called some laboratory, center, institute, etc.) have a grant that is renewable every 3-5 years. For each renewal, things can be renegotiated with the funding agency. I guess, it needs more discussion on how long a term for renewal should be. But for large or medium-sized long term projects, some sort of renewal process is necessary to avoid corruption or the like.
I guess the real issue is how to fund small-scale long term projects, where the truly disruptive discovery or invention may emerge. Federal funding agencies in the U.S. do not have a good mechanism to support such projects. They typically use the same or even more frequent renewal mechanism for small projects. As a matter of fact, longer terms for each renewal would probably be more suitable and the focus may be better for funding individuals instead of projects in this case, as small projects or individuals need the most flexibility and freedom to explore.
The article talks about involving non-profits or private foundations for long term research, which is a very good point. I guess that they would be best fit for supporting small-scale ones or better yet individuals. But the critical part is how to make them go after truly high-risk projects. In my personal experience, they are more likely to support mainstream studies even though they claim to support high-risk high-reward projects in their mission statement. They are also more near-sighted and seldom support longer-term projects. Maybe as the article suggests, tax advantages and / or other incentives may help change their funding behaviors. More detailed discussion on this aspect is needed.
The manuscript concerns an important topic - how to create structures for long-term research. The ideas, briefly mentioned, are reasonable. However, I'm not sure it adds enough in terms of substance, detail, and novelty. I'd love to see a version more worked-out and referenced (arguably not having any references is a stylistic choice, but in this case the consequence seems to me that it limits the content) - there's more detailed work on the matter out there already that could be brought in and discussed, and maybe new, more specific ideas or hypotheses generated off of that existing work. A quick search yield a potentially useful Nature article - “Long-term Research: Slow Science”.
I like it, seems like a precis version of something that would be filled out further. But approvable as it stands. Also, is all long term research good research? Should there be a priori controls or guardrails on long term research that could produce a dangerous outcome, whether accidentally or malevolently?
My first thought was that this needs to be fleshed out a bit, the proposal is at a pretty high level of generality and does not provide any evidence of its claims that there is a lack of long term research or provide more detailed examples of its value. But then I thought, well, if you add all of the flesh to it, I would still walk away with the main point that we need more long term research. I do think we generally need to be looking farther out with everything, but we are trapped in a political and economic system that has a short reward cycle. The idea of tax credits is interesting. There is a seed in there.
Mario Pasquato (PhD in physics):
The manuscript discusses the current landscape of long term research, defined as research with a 25+ year horizon. It points out that neither academia nor the private sector are currently prioritizing long term research, and sketches possible solutions to this issue. I find the question of how to promote long term research interesting, but I believe the article would benefit from a thorough rewrite and I cannot recommend publication in its current state. The discussion of the present state of long term research in academia and private labs should be much more detailed and supported by data and references. Similarly, the current discussion of proposed solutions is merely a sketch.
Enrique Muñoz Ulecia:
Despite I agree with the author(s)' idea, it is not very innovative or thought-provoking, and the description of the outcomes is too vague. This could become a yes with a more detailed description, specific research areas where this long-term approach could offer the expected outcomes, or more detailed policy measures to promote it.
There’s not much to disagree with in this short note. It all sounds like common sense (even though there are surely “free market” advocates who would strongly disagree with it) but it doesn’t offer enough detail to help solve the problem. What’s missing is (1) any references to existing literature on the issue; (2) enough interesting example proposals for LT research that is not being done; (3) better analysis of why the problem exists and how to solve it.
A quick look around showed me that there is a large literature on time discounting that could be relevant.
There might also be useful ideas from long-term, but not monolithic, projects that actually exist, such as: fusion power, accumulation of DNA banks or seed banks, weather forecasting, various mapping or cataloging efforts (e.g., of the universe, of biological species on earth, of geological strata, of climate history, of planets in our solar system). And those examples suggest there might be something relevant in where a project is on the spectrum of pure science to pure application. They also open the question of what makes a long-term project possible when it actually consists of linked but independent efforts. What makes a society loosen its time discounting for some projects but not others?
The appeal for tax credits seems weak or even pointless when politics (in the U.S. for sure, and presumably elsewhere in democracies at least) is an endless series of one faction trying to undo the policies of the other faction, happening on an inter-election time scale much shorter than the exemplar 25 years horizon.
I completely agree with the argument, but the proposal is insufficiently developed. First, there are two different types of research being discussed (fundamental and longitudinal) that need to be distinguished, with separate challenges. Second, there needs to be more examples, from the past and for the future, to give specificity to the proposal - and also to engage with the challenging specifics of such research. Finally, any funding still needs prioritisation, there is a lot of uncertainty in the future, so how can this research be evaluated for funding beyond merely being long-term.
The article is clear and well-written. Here are some comments and questions I have.
It would be interesting to have more data to support some of the claims, such as the rarity of long-term research (LTR henceforth) and motivation for the definition of LRT (why 25 years ?). I see a lot of potential to dive deeply into a meta-analysis of research projects and into the history of science.
Does LTR vary from country to country? How long are research projects on average? Is there a comprehensive review of LTR (the author(s) mention Darpa, NASA, big tech companies - are there others?)
As it stands, the prospects feels a bit utopic but it feels good to dream. As a researcher who was, academically, part of LTR (during my PhD I had the opportunity to carry out research started by one of my supervisors a decade ago) I can see the appeal.
This essay would deserve some development. Why is 25 years chosen for long-term projects? Why are some fields, space research as mentioned, more prone to fast research than others? How would such long projects be implemented? The take is interesting but a bit too limited in its current form to be really interesting as a published article.
This is a very short opinion piece that discusses the desirability of long term scientific research on a 25 + year timescale.
Whilst any appeals to increase research funding are welcome, the author largely ignores the huge challenges in implementing long-term research, such as securing consistent funding or managing changes in leadership and personnel.
More fundamentally the author fails to address the central problem of who or what organisations decide what long-term projects should be funded in the first place, after all there are countless examples in the history of science where advances have been made which have been counter to the prevailing orthodoxy.
The basic question for any research funding, is how the research will advance knowledge, a judgement that inevitably raises the paradox of funding – that it is the tastes and preferences of consensus opinion that determines how funding should be allocated. But is existing knowledge, as represented by a consensus, the best predictor of how knowledge will evolve?
To take a particular favourite example of mine – the Theory of Continental Drift, which would, as it turns out, have been a perfect long-term research goal. Yet Continental Drift was dismissed as laughable by most if not all US geologists for over 40 years. The key phrase here is US geologists. In fact, as Naomi Oreskes and others have shown, in the growing field of Philosophy of Geology, many European geologists (notably Wegener) had amassed a great deal of evidence to justify long-term research in Continental Drift and later Plate tectonics, which the American geology establishment adamantly refused to consider, until in the end the evidence became overwhelming, (see: The Rejection of Continental Drift Theory and Method in American Earth Science, Naomi Oreskes 1999)
New directions in science, often opposing the existing consensus, seldom have the credibility to justify any investment, let alone long-term funding. I suggest that a practical and perhaps more modest seed-funding system would be preferable to the huge organisational problems of long-term funding. There is much to be said on this important topic, but I realised I might run the risk of submitting a review that was longer than the original paper.
Given the lack of supporting evidence in the article, it was tempting to reject, but on balance and purely because it touches on a vital subject, I submit a qualified acceptance for publication.
Dr. Payal B. Joshi (PhD in chemical sciences):
The article is written in blog post format and does not provide any insight on an interesting topic. The argument presented in the article on a 25 year long research project is neither nuanced nor engaging. Though the idea of the article is interesting, it is seriously recommended to provide an insight from the author's perspective. I suggest authors to rethink under following points:
"...Government should continue to fund..." is a generic statement and does not provide any support to this statement. Why must the government provide any funding under inflation/hyperinflation, post-pandemic era?
Authors need to describe the potential challenges of long-term research endeavors and also suggest the methods to undertake them. Unless these arguments are not presented, the idea of this article is nothing but a mere generic imagination.
Authors have pointed out the long term research horizon. If one applied critical thinking to this concept, they can provide components of this particular horizon. Apart from the IT industry, material sciences perform long-term projects such as Europe's graphene project, Open Science Project, Rover Challenge and many others.
Authors can elaborate on different proposal formats that can be utilized for long-term projects in a commercial sense.
Until the above are resolved, the article is not deemed fit for publication. Since the idea is unique and interesting, I recommend major revision with a possibility of further reconsideration with adequate arguments, critiques, critical perspectives and examples of long term projects.
It is clear, well-written and concise, but in this case too brief! There are governments funding long-term research outside the USA - the EC has its own long term research funding agncy, UK has Foresight and now ARIA - needs more detail and discussion before publishing.
Reminds me of Robin Hanson's Long Views. He himself seems to think in longer time-lines (e.g., Grabby Aliens) and he seems to predict that longer views are coming: ”Arguably, competition and evolution will continue to select for units capable of taking longer views.”
See “Long Views Are Coming” - I think promoting such long views is a natural outcome and probably something he would support.
What I like are:
- Existing longer-term projects.
- Some arguments of what prevents existing organizations from pursuing such goals.
What I am missing in the article are:
- Arguments why this is a good idea.
- Why specifically 25 years? Are there specific reasons for that or is it just a round number?
- A deeper look at the mechanisms that (de)incentivize such research.
- More specific actions - for example Robin Hanson suggested cascaded prediction markets for research outcomes.
The way it is presented right now I would be dubious of the motivations people would have to apply for such research.
I disagree with the premise. The nature of most research *is* long-term. Indeed, what are fields of study if not long term research projects? The field of statistics has as its overall goal to better understand statistics. The same goes for psychology, physics, and so on. These are long term research programs, carried out by thousands of researchers over decades and centuries.
To say that each individual research project within these fields is not long-term misses the point of (and the underlying epistemology guiding) research. Most research is incremental not because of bad incentives, but because we don’t know what will work. Indeed, if you know what will work a priori, it’s not research.
It’s notable that the author does not give examples of specific long-term research programs other than those we already engage in (which are not research per se — more like engineering projects for which we can see the finish line in advance).
A useful thought experiment is to take yourself twenty years backwards in time and ask: What should have been the long-term projects? After some reflection, it becomes clear that the breakthroughs were not predictable in advance.
Regarding publishing: Even though I disagree with the premise, the article could be useful if fleshed out. I don't think it's strong enough in its current state, however.
This article, while brief and somewhat under-developed, provides an interesting idea worth thinking about further, and in that sense qualifies as a “seed of science”. This is an interesting edge case for SoS - how small is too small to even be considered a “seed”? I’m reminded of a recent philosophy paper, titled “Can a good philosophical contribution be made just by asking a question?” - and that’s it, there is no body to the article. If a question can be considered a good philosophical contribution then I don’t see why this article can’t be considered a useful scientific contribution as well.
I will share a few extended reflections in order to demonstrate the fertility of the idea.
The author defines long-term research as any project with a 25+ year horizon (one well-known example of such a project is the 35 years and running Long-term E. coli evolution experiment). Obviously this definition admits a wide range of time horizons, from projects lasting 25 years to thousands of years or more. The author also does not spell out exactly what is meant by long-term research - is it one experiment that is done continuously over 25+ years (like the E. coli evolution experiment) or a long-running project which may encompass many distinct investigations all aimed at the same general goal? Both are considered in what follows.
The paper focuses on the organizational/financial challenges of conducting long-term research, but another factor, one which becomes more important as we consider truly long-term projects, is the motivation to continuously work on such a project. For a project lasting 100s or 1000s of years, resources are ultimately secondary - if enough will to conduct the research remains, then a way will be found.
How can we organize and support extremely long-term research? The article suggests governments or non-profits but even those are too fickle and too fragile. Only one kind of organization has demonstrated the ability to sustain activities over such a time horizon: religious institutions (e.g. the Catholic Church). Scientific research requires a leap of faith - you must be willing to conduct activities that offer no immediate gratification and perhaps no delayed gratification either (success is never guaranteed). Religions have thousands of years of experience with getting people to perform tasks of this nature (“your reward will be in heaven”) so it would make sense that religious institutions may be best suited for research projects of these timescales. As one example, Zoroastrians in Iran have kept Atash Bahrams (the eternal flame) burning continuously for over 1,500 years; one way to motivate long-term research may be to have a long-enduring religion declare the ritual value of an experiment.
Can we imagine an institution like the Catholic Church conducting a 500 year-long research project? a 5,000 year-long project? What could even be the nature and focus of such a research project? A first thought is that it would have to be something that speaks to the metaphysical and/or moral commitments of the religion or somehow validates its beliefs. For example, some religious traditions (Advaita Vedanta, Shinto, some forms of Buddhism) hold an idealistic (the universe is fundamentally mental) or panpsychistic view of reality (matter is alive or conscious in some way). We can imagine some future group descended from one of these traditions which believes that research into the nature of consciousness and/or matter will yield evidence for their metaphysical view and conducts an extended project aimed at finding this evidence. Perhaps the Largest Hadron Collider (or its neuroscientific equivalent) will be both a scientific apparatus and a temple and will take hundreds of years to construct. As a further example, one of the central tenets of Jainism is non-violence against all beings. This belief is what motivates their lacto-vegetarian lifestyle, but we can imagine future Jains going above and beyond by launching various long-running projects aimed at producing death-free food (e.g. synthetic “plants” - some kind of artificial system that harnesses light to produce edible plant tissue) or ending wild-animal suffering.
Another reason that religions and their institutions could play a valuable role in conducting long-term scientific projects is that they can, at least in theory, have “purer" motivations (perhaps conducting research simply to learn more about the beauty of god’s creation) than governments who will always have some ambivalence towards expensive research that has no obvious applications or practical significance. It may be hard to imagine present-day religious groups developing such an interest in basic scientific research but we shouldn’t be so myopic - the relationship between science and religion has evolved considerably throughout history and will only continue to do so in the future. New religions may spring up with explicitly scientific practices and goals and existing religions may develop new beliefs and practices. For example, a new strain of Christianity could arise which exalts evolution and sees its preservation and enhancement as a sacred duty. The aforementioned long-term E. Coli evolution experiment (which has taught us a tremendous amount about evolution) requires a bacterial culture to be transferred to a new beaker every day; perhaps this transfer could come to be seen as a kind of sacrament carried out as a part of daily worship. Extended ecological data collection could also come to be seen in a spiritual light; for example, a religious group could set up a long-term experimental forest which is studied in order to gain deeper insight into ecological dynamics. Future scientific advances could enable even more exotic possibilities (one example: high-throughput evolution of synthetic lifeforms as a spiritual project).
It is beyond the scope of this comment to enumerate all of the challenges and opportunities that could arise from further entanglement of religion and science, but I offer one further thought on the matter. Science, like god, works in mysterious ways — patent clerks transform our understanding of space and time and bacterial immune systems provide us with stupendous advances in gene editing technology. I for one won't be surprised when the next scientific revolution of Copernican proportions comes from a long-running spiritual research project that the mainstream sees as misguided or inconsequential.
The author notes how the structure of academic science does not incentivize long-term research projects.
“the typical research project might last 4 years at the longest because that is the max duration of a PhD and postdoctoral researcher. An average PhD candidate wants their own unique project; they don’t want to continue someone else’s project”
In a previous paper (“Amateur hour: Improving knowledge diversity in psychological and behavioral science by harnessing contributions from amateurs”), my co-author and I argued that amateur independent researchers may be able to make valuable scientific contributions by focusing their efforts on “blind spots” in the academic system. In the following passage, we discuss the nature of these blind spots using long-term research as an example.
“We propose that amateur psychologists can most effectively improve knowledge diversity in PBS if they focus on “blind spots”—topics or endeavors that are generally neglected in academia (e.g., because they are not incentivized, or due to some other constraints) but have a large potential to lead to new insights and discoveries. For example, the “slow scholarship” movement highlights how scholars face a general intensification in the pace of work and an increasing pressure to publish (Harland, 2016; Hartman & Darab, 2012). Research indicates that the average number of publications at time of hiring for science faculty positions has been steadily rising in recent years (Pennycook & Thompson, 2018; Reinero, 2019; Van Dijk, Manor, & Carey, 2014); trends like this may influence researchers, especially early career researchers, away from projects that require dedication over a long period of time. This suggests that long-term research projects are generally a neglected area in academia (i.e., a blind spot), and amateurs could do valuable work by focusing their efforts on research that may take a significant amount of time to yield results (Medin et al., 2017). This may involve spending decades to build rich and multilayered psychological theories, investigating psychological phenomena in greater detail, or conducting long-term observation.”
The next question is how to motivate amateur independent researchers to conduct long-term research and support them in doing so. Perhaps religions and their institutions could play a role here as well.