Part I : The Business Culture of Science
Since 2000, there has been a 20-fold increase* in research funding for anti-aging medicine. Wow! That’s a good thing. But let’s keep our eyes on the ball. There is danger that this welcome infusion of capital may be biasing research priorities toward those that are most likely to be profitable, and maybe even diverting the best researchers from the radical thinking that will change our understanding of biology.
Whoever discovers an effective age-reversal treatment is destined to become a multi-billionaire!
At first blush, this statement seems obvious, but that doesn’t mean it’s true. There are many historical examples of people who gave enormous gifts to the world, but struggled in their lifetimes for recognition and even for a livelihood. Schubert, Poe, and van Gogh are artists who died poor, while people after them reaped billions from their work. Inventors who never profited from their inventions include Johannes Gutenberg and Nikola Tesla, Jagadish Chandra Bose, and Antonio Meucci (who?). Reginald Fessenden invented radio a generation before Marconi. Rosalind Franklin got no credit for being the person whose diffraction data and analysis was stolen by Watson and Crick for their Nobel research on the double helix.
More to the point, there have been great discoveries that had no commercial value, or even negative commercial value. Linus Pauling spent the last years of his life documenting the anti-cancer action of intravenous vitamin C. To this day, vitamin C is under-utilized and under-studied precisely because it is so cheap that no one can get rich from it. I believe that aspirin and metformin may be two of the most potent life extention drugs that we currently know about, but we can’t be sure, because they are both long out of patent, and no private company can justify the investment to study them.
Rumors abound about cancer cures and energy technologies that are being suppressed because they would derail two of the most profitable businesses in the history of capitalism. I don’t dismiss such claims out of hand.
If there were a drug that could increase average human lifespan by 15 years (with side-effects that were wholly salutary), there would be a dozen companies tinkering with it, adding a methyl group here or a double bond there, looking for a variant that might boost lifespan by 18 or 23 years. In fact, there is about a 15-year advantage for people who are in a loving relationship, have deep community ties, assume responsibility for leadership, make lots of money, enjoy frequent sex, and remain close to young family members; in comparison, the typical middle-aged American is lonely, alienated, struggling financially, and sub-clinically depressed, with a life expectancy 15 years shorter than it could be. The most effective things you can do to increase your statistical life expectancy are psycho-social, but who is conducting research into optimizing the life-extending benefits of community and relationship?
What I believe
I am convinced that the primary basis of aging is an epigenetic program. Systems that repair and protect our cells and tissues are gradually shut down, and destructive systems including inflammation and apoptosis are ramped up at late ages. Gene expression changes, modified systemically by transcription factors that circulate in the blood. I believe that these blood factors are the holy grail of aging research. Control over aging will come when we learn enough about the basic language of epigenetics to reprogram gene expression with our interventions.
The difficulty is that there are dozens of known epigenetic mechanisms, of which only a few have been studied in detail. A few years ago, it was understood that modifying non-coding regions of DNA could affect the transcription of nearby genes (cis epigenetic signals), but now we know that transcription of genes far away from the modification can also be affected (trans signals).
There is yet more complexity: most hormones and regulatory molecules have secondary roles that affect transcription. Imagine an ecosystem of signal molecules that maintains itself homeostatically, but also changes with age. Sixty years ago, we learned that the genetic code is as simple as it can logically be; every codon three base pairs on a DNA strand is uniquely transcribed to one amino acid, and a protein is built by chaining these together in order. Today we are learning that epigenetics is about as complex as it can be. So in my paradigm, basic research in epigenetics is an essential foundation for anti-aging medicine. If we are lucky, a dozen synergizing interventions will do enough reprogramming to re-set the aging clock. Perhaps there is even a region of the brain that is a common source for the molecules that induce age-related change. If we are unlucky, it may require re-balancing blood levels of hundreds of different substances.
I am optimistic that this can be done, but it will require collaboration on a broad scale. The process is unlikely to end with a single patent-holder who can rake in $ billions. The secrecy and the balkanization of corporate research is slowing progress.
Biases in Corporate Aging Research
For the last five years, Google CALICO has been the 800-pound gorilla in the room. Of course, we welcome their funding, the legitimacy they lend, and their collective brainpower to our field. But they don’t play by academic rules. They are not following the open-source / free-to-the-public model that has been so successful for Google in software. They trend secretive and are not collaborating with university experts outside their walls.
CALICO isn’t announcing its philosophy or paradigm, but we might guess from its lineage that their methodology is rooted in data mining and artificial intelligence. Other companies that have announced publicly that they are taking this approach include Unity Biotech, InSilico Medicine and Spring Discovery. They have in common a data-intensive approach founded in theoretical agnosticism.
Machine learning has been used successfully to create algorithms that translate languages, that drive cars, and that recognize faces. The best thing you can say about this approach to anti-aging medicine is that it is free of the theoretical biases that have plagued aging research through the decades. The worst thing you can say about it is that it misses a fundamental difference between organisms and machines.
Machines are designed by human logical minds, and each part is engineered to perform a single function and do it optimally. Organisms are evolved by a process that depends on results only and involves no logical thought. We have found empirically that in biology, parts tend to serve multiple purposes. Causes and effects are entwined in tangled feedback loops. Hormones and other proteins are likely to serve multiple, overlapping functions, some of which are metabolic and some of which are regulatory.
With a homeostatic physical system, you can tweak it to the right and it will bounce back to the left some fraction of the distance, so that the net effect is to move to the right but with less than your original amplitude. With a homeostatic biological system, you can tweak it to the right and it may bounce back and end up further to the left. The canonical example of this is hormesis, which is so counter-intuitive that it took experimental scientists two decades to establish its legitimacy among biological theorists.
The Challenge of Using AI to Modify Aging
Machine learning algorithms work by finding optimal paths toward a well-defined goal. The machine learning paradigm needs a clearly-defined goal as a prerequisite. In the previous triumphs of machine learning listed above, the goal was well-defined before the process was begun.
Application of machine learning to anti-aging will require a quantified measure of biological age. This is what has held up the field in the past. We can measure lifespans in worms in a few weeks, but to measure lifespans in humans takes decades. Aging research needs feedback that is faster than this.
Just in the last year, there are epigenetic clocks based on methylation that predict future mortality and morbidity far better than any other metabolic test. The bottleneck now is the availability of methylation data that is correlated to anti-aging interventions. That is why I have promoted the DataBETA project to collect methylation data from a diverse set of early-adopters of anti-aging interventions.
Using theory-free computer algorithms to search for anti-aging interventions is better than going about it with the wrong theory, but it’s not as effective as starting with the right theory.
This is larger than aging medicine
The culture of business has had a profound impact on science in general, not just aging science. A hundred years ago, people who pursued science were motivated by pure curiosity and intellectual ambition, because there was little reward to be had. Today, science is a career for something approching 10 million people worldwide. Then, science was pursued by dogged individuals. Now, science is managed by bureaucracies.
More patents have been issued since 2000 than all of history before. It’s often said that the number of working scientists is 10 times greater than all the scientists who have ever performed research in the past, but the actual figure is more than 100 times.
The advance in scientific data reflects this increase, and more. To the extent that scientific productivity can be quantified, the productivity per scientist has increased as the number of scientists has advanced exponentially.
What we don’t have is exponentially more understanding. It’s enlightening to compare the first half of the Twentieth Century with the second. The first half** brought us revolutions in understanding:
- Milliken made the electron real as Rutherford pointed to the structure of the atom
- Planck told us the world is quantized
- Einstein taught us to think in terms of a fabric of space-time, molded by matter-energy
- Heisenberg and Schrodinger taught us that the quantum world is fundamentally interconnected and indeterminate
- Godel surprised us with a demonstration that there are limits to mathematical certainty
- Hubble discovered that there are hundreds of billions of galaxies beyond our own, and that they’re flying away from us, the further the faster
- Lewis, Born, and Pauling gave us a science of chemical bonds based in quantum physics
- Alpher and Gamow proposed the hot Big Bang universe
- Franklin, Crick and Watson discovered the biochemical basis of genetics
What do we have in the second half of the century to compare? I’d put three things in the same league as the above list, and they are all observations for which a theoretical framework remains elusive:
- Penzias and Wilson stumbled on the 3 degree microwave background, promoting Big Bang cosmology to the status of a quantitative science (1965)
- Observations of distant galaxies proved that the expansion of the universe is accelerating; dark matter and dark energy were introduced as the least radical modification to established cosmology (1997)
- Epigenetics came into its own in the 21st century, as it was discovered that big variations in gene expression are more important for the direction of life than small variations in gene sequence.
With so many more scientists, why aen’t we seeing new and powerfully synthetic theories? It’s just not plausible that no one as smart as Newton or Euler or Darwin or Planck is alive today. Then, are the “easy” problems all solved, and the remaining problems in science so much harder? Certainly that’s true to some extent. But there is a larger part of the story, and it is the canalization of scientific thought. Scientists today are paid to be efficient. There is a model of productivity borrowed from industry that is completely inappropriate to science.
We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct. — Niels Bohr (to Wolfgang Pauli)
Through the culture of business, science has become conservative, which is to say dogmatic. It is more difficult than it used to be to throw out a theory that doesn’t work. Almost everyone is working to push outward in the directions that science has already advanced, but almost no one is digging at the roots, or exploring fundamentally new directions. Almost everyone is engaged in the safe science of incremental advance and almost no one is taking the big risks. Tenure is granted to fewer science faculty members, and they are getting tenure at later ages. Career uncertainty makes scientists risk-averse.
With so much at stake, science is being managed by committees and bureaucracies. They judge on the basis of conventional wisdom and measurable results. Business by nature is risk-averse. But in the long run, science can only advance when we scrap the idea of predictable returns on investment and accept a very high failure rate.
Part II next week: survey of biotech companies doing research in anti-aging medicine.
* 20-fold increase is my estimate, a soft number. I’ve been unable to identify hard statistics, and of course the very definition of “anti-aging” is changing as the idea that all diseases of old age can be delayed has come into general acceptance.
** I’ve taken the license to include two discoveries from 1952 in the first half of the century.