Half of the Facts You Know Are Probably Wrong

A review of The Half-Life of Facts: Why Everything We Know Has an Expiration Date by Samuel Arbesman.

Dinosaurs were cold-blooded. Vast increases in the money supply produce inflation. Increased K-12 spending and lower pupil/teacher ratios boosts public school student outcomes. Most of the DNA in the human genome is junk. Saccharin causes cancer and a high fiber diet prevents it. Stars cannot be bigger than 150 solar masses. And by the way, what are the ten most populous cities in the United States?

In the past half century, all of the foregoing facts have turned out to be wrong (except perhaps the one about inflation rates). We'll revisit the ten biggest cities question below. In the modern world facts change all of the time, according to Samuel Arbesman, author of The Half-Life of Facts: Why Everything We Know Has an Expiration Date.  

Arbesman, a senior scholar at the Kaufmann Foundation and an expert in scientometrics, looks at how facts are made and remade in the modern world. And since fact-making is speeding up, he worries that most of us don't keep up to date and base our decisions on facts we dimly remember from school and university classes that turn out to be wrong.

The field of scientometrics – the science of measuring and analyzing science – took off in 1947 when mathematician Derek J. de Solla Price was asked to store a complete set of the Philosophical Transactions of the Royal Society temporarily in his house. He stacked them in order and he noticed that the height of the stacks fit an exponential curve. Price started to analyze all sorts of other kinds of scientific data and concluded in 1960 that scientific knowledge had been growing steadily at a rate of 4.7 percent annually since the 17th century. The upshot was that scientific data was doubling every 15 years.

In 1965, Price exuberantly observed, "All crude measures, however arrived at, show to a first approximation that science increases exponentially, at a compound interest of about 7 percent  per annum, thus doubling in size every 10–15 years, growing by a factor of 10 every half century, and by something like a factor of a million in the 300 years which separate us from the seventeenth-century invention of the scientific paper when the process began." A 2010 study in the journal Scientometrics looked at data between 1907 and 2007 and concluded that so far the "overall growth rate for science still has been at least 4.7 percent per year."

Since scientific knowledge is still growing by a factor of ten every 50 years, it should not be surprising that lots of facts people learned in school and universities have been overturned and are now out of date.  But at what rate do former facts disappear? Arbesman applies the concept of half-life, the time required for half the atoms of a given amount of a radioactive substance to disintegrate, to the dissolution of facts. For example, the half-life of the radioactive isotope strontium-90 is just over 29 years. Applying the concept of half-life to facts, Arbesman cites research that looked into the decay in the truth of clinical knowledge about cirrhosis and hepatitis. "The half-life of truth was 45 years," reported the researchers.

In other words, half of what physicians thought they knew about liver diseases was wrong or obsolete 45 years later. As interesting and persuasive as this example is, Arbesman's book would have been strengthened by more instances drawn from the scientific literature.

Facts are being manufactured all of the time, and, as Arbesman shows, many of them turn out to be wrong. Checking each by each is how the scientific process is supposed work, i.e., experimental results need to be replicated by other researchers. How many of the findings in 845,175 articles published in 2009 and recorded in PubMed, the free online medical database, were actually replicated? Not all that many. In 2011, a disheartening study in Nature reported that a team of researchers over ten years was able to reproduce the results of only six out of 53 landmark papers in preclinical cancer research.

In 2005, "Why Most Published Research Findings Are False" by physician and statistician John Ioannides appeared in the journal PLoS Medicine. Ioannides cataloged the flaws of much biomedical research, pointing out that reported studies are less likely to be true when they are small, the postulated effect is weak, research designs and endpoints are flexible, financial and nonfinancial conflicts of interest are present, and competition in the field is fierce. Ioannides concluded that "for many current scientific fields, claimed research findings may often be simply accurate measures of the prevailing bias."  Still knowledge advances, spawning new facts and changing old ones.

Arbesman also delves into what he calls "hidden public knowledge." Consider again that nearly 850,000 new articles dealing with biomedical research were published in 2009; lots of true information will be overlooked in that deluge. Arbesman illustrates the point with the case in which Harvard researchers decided to go back and actually look at all of the prior randomized control trials relating to heart attacks and the drug streptokinase between 1959 and 1988. In 1988, a big trial showed that intravenous streptokinase was effective in treating heart attacks.

Combing through the "hidden public knowledge" in the prior studies, the researchers concluded that the scientists could have found a statistically significant result in 1973 rather than in 1988. No doubt countless lives were lost because the knowledge of streptokinase's efficacy remained hidden in the scientific literature for 15 years.  The good news is that better data dredging techniques, CoPub Discovery, and collaborative networking systems like Mendeley have been developed so that fewer facts will languish unused in the future.

People also cling to selected "facts" as a way to justify their beliefs about how the world works. Arbesman notes, "We persist in only adding facts to our personal store of knowledge that jibe with what we already know, rather than assimilate new facts irrespective of how they fit into our worldview." All too true; confirmation bias is everywhere.

So is there anything we can do to keep up to date with changing facts (other than constantly reading Reason)? Arbesman suggests that simply knowing that our factual knowledge bases have a half-life should keep us humble and always seeking new information. Well, hope springs eternal. More daringly, Arbesman suggests, "Stop memorizing things and just give up. Our individual memories can be outsourced to the cloud." Through the Internet, we can "search for any fact we need any time." Really? The Web is great for finding a list of the ten biggest cities in the United States, but if the scientific literature is merely littered with wrong facts, then cyberspace is an enticing quagmire of falsehoods, propaganda, and just plain bunkum. There simply is no substitute for skepticism.

Toward the end, Arbesman suggests that eventually "exponential knowledge growth cannot continue forever." Among the reasons he gives for the slow-down is that current growth rates imply that everyone on the planet would one day be a scientist. The 2010 study in Scientometrics mentioned above also mused about the growth rate in the number of scientists and offered a conjecture "that the borderline between science and other endeavors in the modern, global society will become more and more blurred." Arbesman notes that "the number of neurons that can be recorded simultaneously has been growing exponentially, with a doubling time of about seven and a half years." This suggests that brain/computer linkages will one day be possible. I, for one, am looking forward to updating my factual knowledge daily through a direct telecommunications link from my brain to digitized contents of the Library of Congress.