Why we exist

Motivation

Stalled Progress

Technology has advanced at an amazing pace in my lifetime. From smart phones that make the computers I grew up on look like toys, to the internet, and now to advances in Artificial Intelligence. It has been insane the level of change and progress.

The same can not be said for medicine. There have been some advances but in general (mostly in prevention), most of the major diseases are still diagnosed and treated with 50-plus-year-old tech.

Chemo: 1940s
Radiation: 1900s
MRI: 1977

Every week we see articles about medical breakthroughs but they never actualize. You read the fine print and it is in highly controlled settings or in mice. Splashy headline that you then never hear about again.

The supplement industry is a billion-dollar industry not because most supplements work but because people turn to it because standard medicine doesn’t have answers.

Nutrition information is just as bad. Should I eat eggs? Yes/no, how about eggs and coffee? Everyday there are seemingly conflicting studies. It is a mess.

I am not alone in the conclusion that scientific progress is slowing or stalling.

Despite more money than ever, our return on investment is decreasingly returning with fewer groundbreaking results. We are getting less bang for our buck. Spending and the number of scientists are at an all-time high, but based on several surveys and studies progress is failing.

One survey of prominent researchers, asked to compare the quality of Nobel Prize–winning discoveries, and those surveyed think that the quality is going down.^[1]

Other researchers have tried to track the number of disruptive papers. Papers that reshape a field so much that they mark a paradigm shift. They do this by looking at how that paper impacts subsequent citations on papers, which they refer to as a consolidation–disruption (CD) index.

Using this methodology they “found a significant decline in average disruptiveness”. The number of papers has gone up, but according to this, the quality has trended significantly down.^[2] ^[3]

In the world of computer hardware we have Moore’s Law which states that the number of transistors in an integrated circuit (IC) doubles about every two years.

In the pharmaceutical industry they have the opposite of Moore’s Law, Eroom’s Law (“Moore’s Law” spelled backwards) which states the cost of developing a new drug doubles about every nine years, meaning drug discovery is becoming slower and more expensive over time. The exact opposite trend we expect with technology.

Dysfunction

There are probably numerous reasons why this is occurring but here are some theories to highlight.

Goodhart’s Law and the great narrowing

If you want to innovate, you need to go wide and explore new areas. You don’t know what will lead to a new breakthrough. Penicillin was discovered partially accidentally. I say partially because Alexander Fleming was a great scientist and was curious enough to investigate what was going on with some of his petri dishes. This is a good example of how something unexpected happened and having the curiosity but also the time and space to explore why.

Accidents and exploring things are often what lead to innovations.

But that doesn’t really happen anymore. It is hard, and there are no guarantees of success. So everyone stands where the light already is. Drug companies are looking for more drugs. Scientists are looking for things they can publish successfully. You don’t get credit in our current system to publish a failure, even though failures are just as important.

Everything in medical research suffers from Goodhart’s Law, which basically states that many measurements become the goal instead of the original goal.

Your goal as a research scientist isn’t really to do science; it is to publish papers, and not just that, publish papers that are cited by others. You do this so you can get grants to keep paper publishing cycle alive.

While that may sound like the same thing as doing science, it is not. If you have to publish x papers in y time, maybe you don’t get real time to explore the true unknown. If you have to publish only successes, maybe you find easy wins or cheat. There is a term for this “publish or perish”.

I think the same can be said for many organizations as well. I am sure many people in them want to do good, but over time, as the organization grows, so too does the pressure to keep it alive. Maybe the goals are aligned, but it seems more and more the organization is prioritized over making real progress.

Organizations become complacent, and groupthink is a real thing, slowing down or inhibiting progress. I don’t hate pharmaceutical companies, but their goal is to make the company successful, not cure people.

In the book Loonshot by Safi Bahcall calls this franchise mode, where people focus on getting the most out of their current investments instead of searching for new investments. Instead of new movies, we get yet another super hero movie.

Innovation is about exploring. When every step has to be “right” or proven ahead of time, then it is impossible to learn anything new. You need to find new paths, and that means that initially they will not be proven or may in fact seem entirely wrong, but eventually they will be right.

We want to believe that new knowledge is a straight path, but it is not; sometimes you need to step back and move laterally to find new paths.

Studies are not understanding

Ok, I will pull that punch a bit. When taking something new I would really would like there to be some double blind studies done to show it at least won’t kill me or worse.

They can also help point in the right direction.

That being said, I think most studies tell us very little. The problem is that after the study you can’t extrapolate. You haven’t learned a general rule. You can show that smoking in general causes cancer, but you can’t say why.

If I drink coffee, eat eggs and stand on my head for an hour, is that good or bad for me? How do you design studies when everything interacts with everything else? Our bodies are a crazy complex system, and almost everything has an effect, but the combinatorics of trying to create a study for all these things just don’t work.

Which brings in the next part, everyone is different. I think one of the reasons that medicine has tapped out is because many of the “general” cures and fixes have been found. Now we are in the territory of individual make ups. Not just genetics, gut biome, lifestyles, environment and on and on.

That is why you hear these stories, of oh my grandmother drank and smoked until she was 94. It might not just be that she drank and smoked, maybe she just had amazing genetics and/or there was something else she was doing that counteracted the effects. She was probably built differently and had a different lifestyle. You grab 100 people for a study hoping that represents the general population but it doesn’t.

I truly believe the next phase of medicine is almost 100 percent personal.

So if that is true, once again a study of a group of people becomes useless.

Stamp Collecting

I feel like modern medical research is focused solely on fact collection. Same issue with these narrow but possibly “successful” papers. You find some small new detail, write it up in a paper and call it a day, but the best ideas create more space and more questions. Platform level ideas where you jump off from and build on.

It is like finding a new coffee shop vs a new neighborhood. The new neighborhood is a new open question which can then lead to more new. A coffee shop is known and maybe useful but it is not opening new horizons. DNA was one such idea. After DNA was discovered there was so much space to explore.

It seems to me that there is some implied idea that if we just collect enough of these small pieces, something larger will emerge. I don’t think that works.

A good example of this is AlphaFold. One of the biggest challenges has been understanding how proteins fold to create their 3D shapes. How could you predict based on DNA the end structure of a protein?

Let’s say there are roughly 20,000 of them in the human body. Prior to AlphaFold, the way to figure this out was to have a grad student just basically hunker down for a few years and use various experimental techniques to figure out just one.

Not a great plan. One, there are a lot of them, it is going to take some time and a lot of hours. Two, you haven’t really learned anything fundamental. If you find new proteins, you are back to the same place. You have collected facts that end with just that fact.

AlphaFold used various machine learning techniques to figure out the underlying structure and to be able to make accurate predictions just based on the genetic information. This is a huge leap forward and the team won the Nobel prize for this work.

With AlphaFold, you don’t know one fact but can extrapolate new facts.

There are two problem modes for knowledge acquisition, depth limited vs integration limited. We are intergation limited at the moment.

Too few people

Let’s take cancer for instance. It is hard to get exact numbers but let’s use the American Association for Cancer Research’s member count.^[4] They claim 58k members but probably not all are doing active research.

Either way that is less than the employee count then many of the large tech companies. In the US that is 58k per 300 million people. That is not great.

DNA is not the whole answer

The structure of DNA was first described in 1953. Since then the field has been obsessed with DNA. I am not saying DNA is not important but it has only one function, store data. That is it.

Important for sure but it is like studying computers and only focusing on the hard drive. It is needed but there is a lot more to software than storing data.

There is a reason Crispr can’t fix everything. That is because there are only a fraction of diseases that can be easily tracked down to one section of DNA. Many if not most are the result of complex interactions that can’t be traced back to one mutation.

Outside of our own cells, lives an entire biome of bacteria. We are also learning about how important gut-microbiome is to health.

Missing tools

It may also be that we have reached a limit on what one or a team of people can know and hold in their heads. The human body is so insanely complex that there could be a need to have something that can manage and meet that level of complexity that the normal human mind can’t.

Conclusion

I don’t want to dismiss how hard this is; I think solving human biology is one of the hardest problems we face. That being said, I think something is majorly wrong with how current medical science operates and I think that past measures are inadequate. I think a new way forward is needed.