Robin Hanson on Medical Doubts

Robin Hanson was asked to write an OpEd piece by an editor, who then never responded after the article was submitted. He has now published the paper on his Web log, Overcoming Bias, as “Medical Doubts”.

But surely modern science must have some reliable way to study the aggregate value of medicine? Yes, we do. The key is to keep a study so simple, pre-announced, and well-examined that there isn’t much room for authors to “cheat” by data-dredging, p-hacking, etc. Large trials where we randomly induce some people to consume more medicine overall, and then track how their health differs from a control population—those are the key to reliable estimates. If trials are big and expensive enough, with lots of patients over many years, no one can possibly hide their results in a file drawer.

Thankfully, we do have a few such studies. Yes, they have limits. They may not include all patient ages, or all kinds of medical care, and they can only see marginal health effects, of the medicine that some get that others do not. But for now, they are the best we have.

Which brings us to the biggest medical news of the 2021, at least for those less inclined to give medicine the benefit of the doubt. We now have one new such study: the Karnataka hospital insurance experiment. From May 2015 to August 2018, 52,293 non-poor but otherwise typical residents of the Karnataka region of India were randomly assigned to get free hospital insurance, an option to buy such insurance, or a control condition.

While the study saw large effects on hospital insurance purchases and on hospital visits, when looking at 82 health outcome changes over a five-year period the study authors “cannot reject the hypothesis that the distribution of p-values from these estimates is consistent with no differences. (P=0.31)” That is, they saw no net effects; people who got more medicine were not on average healthier.

The article goes on to discuss other large scale randomised studies of expanding access to medical care on health outcomes and concludes,

Bottom line: we spend 20% of G.D.P. on medicine, most people credit it for their long lives, and millions of medical journal articles seem to confirm its enormous value. Yet our lives are long for other reasons, those articles often show huge biases, and when we look to our few best aggregate studies to assuage our doubts, they do no such thing. And the biggest news of 2021 is: we now have one more such study.

Read the whole thing


Charlie Munger often says show me the incentives and I will show you the result.

Twenty years ago I played online games with my nephew. They had a saying “FRAPS it or it did not happen”. FRAPS was/is video recording software.

I suspect that gold, silver and bronze medals as well as trophies come from a time when you couldn’t FRAPS it. You could not prove an actual result. The medal and trophy were a way to record it. Maybe the NFT will replace these outdated proofs of accomplishment.

When an organization, community or country cannot show an actual tangible results, they come up with a system to give rewards.

Paper publishing has replaced tangible research results in academia. Their reputation, funding and ego (called an expert) are based on papers published rather than tangible results. I see patents used the same way in corporations. People get rewarded based on patents regardless of whether these have or ever will result in a tangible product sold. If an inventor invents a real product, you can see it in the marketplace (with a price tag on it).

Maybe there are accomplishments in research that are real and important that don’t have an immediate tangible result. The corruption of the reward system harms these accomplishments as much as not having any reward system. Maybe more.

When there is an incentive to cheat and it is easy to cheat, a huge percentage of people cheat. When this intersects with biased parties funding the system, it becomes a dangerous situation for society.


Quote is from Robin Hanson, not our host. Key question is how much of that 20% is actually spent on medical treatment as opposed to regulatory & insurance overhead?

Another factor is quality of life (difficult to measure) versus length of life (easy to measure). Most of us have known individuals with terminal diseases whose life spans were extended by modern medicine, but at a dreadful cost to their quality of life in those expensive closing months or years.

As for average life span – is that a useful concept? Wandering through old graveyards, we typically see a lot of gravestones remembering people who lived to ripe old ages without the benefit of modern medicine; and we also see lots of childhood deaths and injury deaths which have been largely eliminated by good sanitation, traditional vaccinations, and more focus on workplace safety. Of course, the average life span has increased, but only part of it due to medicine.