A Written Interview with José Luis Ricón Fernández de la Puente

ExMedPod: Although you have a background in Industrial Engineering and Aerospace Dynamics, you also have interests in economics, tech, machine learning, data science, and biology. You’re a prolific blogger and an Emergent Ventures winner, whose works have been published in many places including on a16z and Works in Progress. What you write about on your blog is, frankly, not easy to summarize. You’ve written about the trends and data underlying California Wildfires, how we should change science funding allocation, how to create a better Google Scholar, the benefits of one-on-one tutoring for educational instruction, the economics of the USSR, the science behind aging and longevity, and a whole host of other topics.

A lot of your work seems like an attempt to survey the landscape of information about a topic, distill it down, reorganize it and then summarize it simply and clearly such that a non-expert can understand it. How do you think about your wide range of interests and how they connect? How would you explain your work to people who don’t know anything about you?

José: Yes; most of what I write tends to be driven by wanting to know something after having encountered experts disagreeing about that topic. Normally trusting experts is the best choice to get an answer to a question, but in the presence of conflicting data or arguments, who is one to trust? Scientific evidence can be hard to interpret and one may need to muster multiple different strands of thought to get at an answer. Often one can’t take a given publication at face value, one has to check the supplemental data and the models they use to appropriately weigh the contribution of that work to the final answer. I of course first try to find if there’s a good answer to my question, I don’t want to duplicate work, I just write what I wish I had been able to find in the first place. For example at times I see claims that the consumption of relatively large amounts of eggs do not raise the odds of CVD and then I would consider writing on that but fortunately the kind of comprehensive review for that already exists (spoiler: eating lots of eggs is not such a great idea) so I don’t need to write it.

The rationale for why I write is roughly the same in all cases, but there is no common thread that unites all the topics I have written about. I have found different themes interesting at different points in time and sometimes I’ve thought that it would be worth writing about them, but there’s no further explanation beyond that.

I would explain the kind of writing I do as an enlightened version of the “Someone is wrong on the internet” reflex :). I don’t feel like I’m entitled to my opinion in the face of expert disagreement and if I really want to know, I’m left with no choice but to assess everyone’s claims and how good their evidence is, and come to an all things considered view. I don’t make any money from writing nor am I forced to write about anything specific. I don’t sell anything in my blog. I try to make sure that incentives do not get in the way of being as correct as possible. In some cases I even pay readers that find mistakes in my writing!

ExMedPod: Our next question is about science funding and the process of knowledge creation. Science today faces many problems. There is a replication crisis, lack of data transparency, extensive paywalling by journals, an exponential increase in publications including X-is-associated-with-Y research, citation cartels, and inefficient funding mechanisms. Additionally, the Covid crisis of the last few years has brought other bureaucratic problems in science to the fore. In response, many solutions have been proposed. Some have suggested that we build alternatives to the current bureaucracies. Others would like to abolish peer review, treat research more like a live software product, or fund people rather than projects. Alternative organizations have even been created to diversify the funding landscape in science. A lot of these efforts seem like attempts to circumvent the ubiquity of vetocracy within science. How do you think about the major problems science faces today? What metrics should we use to judge the success of scientific progress & science funding mechanisms?

José: Last year I wrote this article “We don’t know how to fix Science“. The title is admittedly facetious, as we do know some things we’d want to change, but the gist of it is that it’s hard to know from current “hard data” how we should prioritize various proposals, and by how much they will improve outcomes. Even these outcomes are hard to measure! If we agree that the value of a given paper is the downstream impact on other publications, plus the social value derived from novel inventions made possible by that work, measuring this chain of value is quite hard. Citations wouldn’t be enough; PCR is frequently used in the life sciences but no one cites Kary Mullis’ original work every time they use it, so this would greatly underestimate the value of this technique. In the field of the economics of innovation or science of science, citations or patents tend to be two commonly used indicators of scientific success of a researcher or an institution because they make some a priori sense, but chiefly because they are easy to measure and in a way relatively objective, though of course citations derive from individual value judgements about the value of a given piece of work, but if a paper has lots of citations, it can be taken to mean that there is overall consensus that it was good work, if cited positively at least. On the other hand we have more subjective measures like looking at high profile awards like the Lasker or the Nobel Prize. With citations, one is relying on some assumptions to make the claim that high citations mean good work, with the Nobel almost definitionally, experts in the field are directly teasing out what the good work is in a more direct way; the downside of course is that the Prizes are at the mercy of a smaller committee that may reward a particular scientific taste. They are also too binary to be taken into consideration alone: Doudna and Charpentier won the Nobel in 2020, but there is a long tail of researchers that worked with them, or whose work directly led to theirs that was not recognized by the committee. There are half Nobels but no microNobels.

So for now we can’t run away from the need to use multiple measurements and individual judgment to assess talent. In parallel to that, work should continue on developing better metrics. For example, one could try to use not the direct citations of a paper, but the entire graph of citations to determine importance, in turn this can be used to probe early patterns in citations and use that to see ahead of time if some work will be considered important in the future. This could be in turn used to allocate funding. Obviously the downside is that once the pattern is known there are incentives for scientists to collude and cite each other in ways that will boost a researcher’s algorithmically-derived importance. There’s truth to Goodhart’s law and the way to get around it may involve fuzzier or composite metrics that are weighted, with different institutions valuing different metrics.

Many of these issues happen because of competitive pressures in science, and because we try to use the same institution, the university lab, to do multiple tasks, from training new scientists to doing fundamental research, translational research, or project management, and doing all that on a shoestring of a budget. Starting new institutions for specific purposes is something I very much want to see more of: For concrete large scale projects that are not quite commercial-ready yet, now we have Focused Research Organizations, and likewise last year we saw a number of new institutions popping up like the Arc Institute, New Science, or Arcadia. We will learn from the experience of these institutions and the ones that will arise in the future and iterate our way to a better science funding ecosystem.

ExMedPod: Now let’s talk about your interest in longevity research. Before we dive into the topic, how did you get interested in aging and anti-aging therapeutics? We were taught in medical school about two schools of thought with regards to aging: accumulated DNA damage versus programmed cellular senescence. Obviously, the story is a lot more complicated. What is your current model of aging and what do you see as the levers upon which science can yank to affect it? How do you feel like your aerospace and software engineering background influences your perspective on the longevity research landscape?

José: My interest in aging comes from a number of places. First, I didn’t have a particular interest in biology until relatively recently. The way it was taught to me in high school was as this relatively figured out system, leaving out the wonder and importantly the speed of progress we have witnessed in the past few decades. Then a few years ago, various things happened: I read David Sinclair’s “Lifespan”, which discusses a number of intriguing findings. It was my first time hearing about “partial reprogramming”, the approach Altos is now pursuing. I had also been thinking about progress in science and technology broadly, noting that compared to other areas, the life sciences are the endless frontiers of science. Thinking that biology was going to play a big role in the decades to come, like physics had in decades past was a reasonable inference. Then from there, it seemed natural to want to understand more biology in general. And to that end I chose as a project to guide my study writing an introductory FAQ about longevity.

There are and have been through the years many theories of aging, some monocausal, others implicating a number of factors. A few years ago, telomere shortening was thought to be one such major driver of aging, and while it does play some role, especially in mice, the field has moved beyond that. These days at a high level one tends to find a large proportion of researchers that believe that aging is substantially the accumulation of various forms of damage at the molecular level. There’s also the idea that some facets of the aging process could be quasi-programmed, an example being the involution of the thymus, occurring in every mammal, and which has been implicated in immune senescence. Natural selection did not select for aging as a trait, but weak selection post-reproductive age led to traits that could be beneficial early on and not ideal later in life.

When thinking about how to address aging, there is no single target to drug, and likewise comprehensively addressing aging will require a combination of multiple interventions (beyond the essential diet and lifestyle). In one extreme, there is addressing specific diseases one by one, in the other there’s targeting mechanisms shared by all cell types. In the latter category one could try upregulating DNA repair or autophagy, but partial reprogramming and its relatively comprehensive effects hint at the idea that maybe if only we reset the epigenome to a youthful state, the cell can handle the rest. It remains to be seen how truly comprehensive this might be, as we already know this won’t undo existing DNA mutations, and it probably won’t help with crosslinks in the extracellular matrix, or other extracellular aggregates like atherosclerotic plaques, those will require a targeted approach.

As for the influence of my background, it’s hard to say! I have a lower tolerance for uncertainty than the average person, is that a product of studying subjects where you can prove and calculate results with very high accuracy? Or was I like that to begin with? Be that as it may, this shows somewhat in the kind of work I’ve done, first to clarify what the field says (in my FAQ) and then to start a project to produce “gold standard” data with comprehensive measurements and substantial sample sizes.

ExMedPod: Your previous experience learning about biology as a relatively figured out system reminds us a lot of medical school training. Much medical school instruction involves memorizing pathways, proteins, and associations without much appreciation for how this knowledge was generated or what frontiers remain for investigation. If you have any ideas for how this might be improved, let us know!

In one of your recent articles, you write about partial reprogramming as a particularly exciting anti-aging intervention because of the ability to manipulate an organism in a “top-down” rather than a “bottom-up” manner. Rather than CRISPR or editing the epigenome “one methylation mark at a time”, partial reprogramming holds the promise of coordinating a whole cell or collection of cells to become more youthful. How concerned are you about the possibility of an inherent antagonistic pleiotropic trade-off between oncogenesis and longevity in humans?

Of course, epigenetics is only one way information is stored in an organism outside of the genome. Another exciting form of informational storage is bioelectricity, which some researchers such as Michael Levin are bullish on. Do you have any thoughts on bioelectricity or other non-partial reprogramming top-down approaches to longevity interventions?

José: Even as early as in high school, teachers should spend some time discussing novel developments in biology. They don’t have to be following the state of the art, but research universities could have biyearly research digests intended for diverse educational levels which then teachers can use to instill that sense of wonder and progress. The goal of that, to me, is to help students see various areas more fairly, not as polished edifices, but works in progress that may need their help.

A priori there does seem to be a tradeoff. Indeed if one takes one of the hallmarks of aging, telomere shortening, evolved as an anti-cancer mechanism. Reprogramming itself, when performed in vivo for too long, leads to teratomas. But then there are some other intriguing data points: We’d think that animals with high regenerative potential like some amphibians and reptiles, with stem cells that are poised to proliferate enough to readily regenerate limbs, should be cancer prone. But they are actually very cancer resistant. Granted we are not frogs! We also know that interventions that tend to both increase lifespan and improve health (like rapamycin) also tend to decrease cancer proliferation.

I really like Michael Levin’s work! I will caveat that I am not deeply familiar with it though. He has focused mostly on cancer and regeneration but evidently his work has implications for aging. Reprogramming starts from the observation that we can rejuvenate single cells (in vitro), and from there we may try to rejuvenate cells (in vivo) that form part of a tissue. An approach inspired by Levin’s work would involve potentially replacing entire sections of tissue, as Jean Hebert has proposed, but then relying on regeneration to restore the organ to its original size. Right now it’s unclear how further one could extend this, but as a proof of concept we do know axolotl can regenerate parts of its heart and even brain.

ExMedPod: There is a growing connection between the crypto world and interest in longevity research. Beyond rejecting some traditional assumptions (e.g., many crypto enthusiasts are opposed to centralized banking; many longevity enthusiasts question the underlying premise that death is unavoidable), why do you think there is such an overlap of interest between these two communities? What do you think are the potential benefits and possible downsides to the interaction between the two?

José: There has been indeed such a growing connection: In 2021 alone we saw the launch of the Rejuvenome Project  ($70M, Astera Institute, Jed McCaleb) or Longevity Impetus Grants ($26M, Jed McCaleb, Vitalik Buterin, Juan Benet, among others). Both the philanthropists and the community at large have had such an interest for a long time. Some say somewhat in jest that you can buy all the NFTs you want but you can’t buy time, and the only way to create time is investing in longevity. But this is not quite it: Even those not in a position to fund any major project have been very interested in the topic for a long time. Part of the explanation may be a higher than average contrarianism: If you are the kind of person that’s willing to bet (and so far, win) against a majority that tells you that you are holding worthless fake money, you may also be inclined to go against consensus beliefs elsewhere, and while aging research is very much an established field, at a social level ideas like “death is what gives meaning to life” are not uncommon.

ExMedPod: For our last question: in your Longevity FAQ: Making of post, you comment on the siloed communities within aging research, whereby scientists interested in aging from different perspectives publish in different journals and effectively function like independent communities. Separately siloed communities seem like a prime driver of human conflict, whether it be geopolitically, culturally, or in academia. It’s difficult to move a field forward when there is disagreement on the basics. Do you have any ideas for how to address this siloing problem in longevity research or science in general?

José: Yes; so to put this in context, at the edges of research, beyond what textbooks teach as incontrovertibly well grounded facts, there’s this frontier of half-accepted data points that is waiting to be established or discarded. There one finds different communities weighing different pieces of evidence differently. Here I point to the early debates around the role of the immune system in fighting cancer. For some time there was a group that believed that “obviously” the immune system could have nothing to do with cancer and another that pointed to various experiments that supported the notion that there were immune cells that could fight cancer. Why did that happen? The early experiments were not crystal-clear and it was only repeated data collection in increasingly controlled conditions that led to results that then drove consensus. To be sure, fields will always have their open questions where reasonable disagreement can exist. But I think it would be useful for there to be more conferences or workshops that bring together opposing schools of thought on an open question, and then getting them to co-design studies that would help bring them into agreement, then fund those studies.