The Good, the Bad, and the Complex
This essay was originally written in October 2018 and published on the blog of my Science Distributed site. That site is undergoing an overhaul, and I’m reposting some items here as examples of early reflections on the use of technology to improve science.
October 17, 2018
Paul Allen’s death this week at just 65 years young was a great loss to humanity. His contributions to the computer age we live in were world changing. His Allen Institute for Brain Science may be an equally critical contribution to the future of science. The Allen Institute's focus is not just on advancing specific areas of bioscience, but also accelerating science worldwide. Their open science approach to expanding quality science across disciplines will improve health outcomes and save lives. This takes time, but the team he built at the Allen Institute understands the long, slow process of quality science. I spoke with some of them last year at the Society for Neuroscience meeting in DC. I was impressed with their forward-looking vision (including blockchain interest) and determination to make a difference. While those impacts won't come fast enough for Mr. Allen and many others, the organization he created and the staff there are a major pillar in the effort towards faster medical miracles.
As a counterpoint, the challenges of conducting science to advance treatments and save lives was highlighted this week by Harvard and Brigham's call for retractions of 30 peer-reviewed publications of cardiac research by one of their star researchers and his subordinates. The whole case involves 50+ papers over 15+ years, many in top-tier journals, with more than $50 million in U.S. taxpayer funding. Bad science also has many indirect costs: misled colleagues funded with this bad lab as a partner, other research built on and referencing this high impact false work, programmatic decision-making made by funding agencies based on the fabricated findings and the opportunity cost of what they didn't fund, clinical red herrings followed, and new directions in treatment not explored.
There is also a deeper layer of cost with the suppression and reprisal that appears to have gone along with this fraud. I've seen first-hand what happens when scientific integrity issues are raised in my time with the federal government. Suffice to say that an honest look into the problems is not usually the first (or second or third) reaction of a large organization. Promising early stage researchers can have their science education warped and their careers stalled or ended, as appears to have happened in this case.
My back of the envelope calculations put the overall cost of this scientific train wreck at half a billion dollars. That's $500,000,000. That’s just a splash in the bucket of wasted science funding.
Most bad science isn't this egregious, but that in itself is a problem because most never gets noticed. This particular case stretches back more than a decade and the cleanup has been ongoing for several years, but it is at least getting addressed in the body of knowledge of the peer-reviewed literature. Only 0.05% of published scientific articles reach this level of necessary retraction response due to clearly identified flaws. A much higher amount, almost 20% of a sampling of the $150 Billion in US based biomedical research each year, is simply not reproducible. This problem receives less attention but steers large amounts of the scientific endeavor off-course.
A start-up that doesn't make money is correctly noted to be an expensive hobby (or a non-profit mission under the wrong tax label). Bad research that gives irreproducible science is not just an expensive hobby, it is an infection to future science and humanity's knowledge.
If it were the case that food was infecting a large enough number of people, there would be rapid efforts to contain it, track it back to the source, and eliminate the problem. If it was medical devices causing adverse events, a similar recall effort would be initiated. We have only the most limited ability to recall bad science in some of the most extreme cases and after many years. The irreproducible science is hard to identify and almost impossible to eliminate, yet it infects everything that comes after. This slows the process of advancing medicine to Improve and Save Lives and drifts us away from fully realizing the impact of the Hundreds of Billions of Dollars$$$ spent on research worldwide each year.
Walmart has piloted a blockchain solution to produce tracking that worked so well (time to identify source of produce in stores went from 5 days to 5 seconds), they are moving their entire produce supply chain to this model. They have gotten top 10 food producers like Dole, Unilever, and Nestle to join a consortium to this end (and now will likely own the standards definition for the entire industry).
Spiritus Partners is working with the University of Edinburgh and the NHS in Scotland to transform medical device tracking, training, adverse events reporting, and recall in similarly dramatic fashion.
What if there was a way to create an open-sourced, or appropriately permissioned distributed ledger of scientific data movements through the bench to bedside chain, allowing for auditing of provenance, contribution, methodological and statistical details, publications, and referencing by future studies? Tech-wise this has probably been possible for years, but there hasn't been a value proposition high enough to seriously try or an incentive system in science fully aligned with the goal of getting higher quality science, fast track-back for errors, more bang for the research buck, and faster improvement in treatment and outcomes.
With blockchain as a new tool, and Open Science as the framework, this might be more achievable and worth the cost of implementation. It’s no magic wand, more like a cordless power drill where only hammers and screwdrivers existed previously. It can speed up some of the things we are already doing (HHS acquisition lead Jose Arrieta has already successfully piloted with purchasing and proposal review), free researchers up from admin time, and give us entirely new capabilities. It can allow us to build a house at the speed and cost we used to build a shed.
It is a complex new area and will take time to map the current system, outline the vision of the future, and take the achievable steps to get to Faster Miracles. Paul Allen's research institute and other forward-looking institutions and individuals will play as critical a role in this effort as Microsoft did in creating the world in which I am writing, and you are reading these words. Get ready.