Originally left as a comment about   PubMed 15343327 on the now-defunct PubMed Commons.

Rose and Wright review energy-efficient methods by which an extraterrestrial civilization might communicate over interstellar distances (1). Overall energy expenditure may not be the proper figure of merit, however. Instead, one might consider methods according to the amount of energy demanded from the civilization.

For example, page 127 of Stephen Webb’s excellent book, “If the Universe is Teeming with Aliens… Where is Everybody?,” notes two communication methods that modulate the energy output of the civilization’s star. These would not be low-energy according to Rose and Wright, because a star expends prodigious energies. Nevertheless, they would be highly advantageous because they require very little energy investment from the civilization.


  • Frank Drake suggested that unusual chemical elements, e.g. praseodymium, could be launched into a star to change its spectrum, causing a signal, and

  • Philip Morrison suggested that matter be placed in orbit around the star so it periodically obscures the star’s light output.

Both methods would be detectable at long distances, and could be modulated to give clear evidence of life (e.g. modulating in a prime number pattern). Certainly, the success of the Kepler probe proves that Morrison modulation can be detected over interstellar distances.

The comparatively low signal-bandwidth of these methods need not be a hinderance. Once a civilization is detected by such a method, an investment in higher-energy, higher-bandwidth communication is easily justified.

Thus, the best approach to interstellar communication would seem to be modulation of stellar electromagnetic emissions. Realizing this, our civilization could launch a program called “Stellar Oscillations To Observe Sentience.”

(1) Rose C, Wright G. Nature. 2004 Sep 2;431(7004):47-49. Inscribed matter as an energy-efficient means of communication with an extraterrestrial civilization.   PubMed 15343327

Rejected by the New England Journal of Medicine in 2017.

To stimulate data-sharing, Bierer et al (*) propose a new type of authorship, “data author,” to credit persons who collect data but do not analyze it as part of a scientific study.

This interesting proposal could be expanded into a more general revision of the authorship concept, without conferring a narrow specialness to data authorship that could equally be claimed for “statistical authorship,” “drafting authorship,” “study-conceiving authorship,” “benchwork authorship,” and so on.

Instead, reviving decades-old proposals for fractional authorship (1) could better achieve the same laudable aims, especially if open-source “blockchain” software technology (2)(3)(4) were used to conveniently, publicly, quantitatively, and securely track fractional authorship credits in perpetuity.

Authorship would thereby have some features of alternate currency (e.g. BitCoin): senior authors could use future authorship credits to “purchase” data from owners according to the data’s value. They could also assign roles from a controlled vocabulary (data author, statistical author, etc.) to some or all authors. Over time, norms for pricing and authorship roles would coalesce in the scientific community.

Overall, a blockchain fractional-authorship system would be more flexible and extensible than a special case made for data authors.

(*) Bierer BE, Crosas M, Pierce HH. Data authorship as an incentive to data sharing. N Engl J Med 2017; 376: 1684-1687.   PubMed 28402238

(1) Shaw BT. The Use of Quality and Quantity of Publication as Criteria for Evaluating Scientists. Washington, DC: Agriculture Research Service, USDA Miscellaneous Publication No. 1041, 1967. Available at: http://bit.ly/2pVTImI

(2) Nakamoto S. Bitcoin: A Peer-to-Peer Electronic Cash System. October 31, 2008. https://bitcoin.org/bitcoin.pdf

(3) Tapscott D, Tapscott A. Blockchain revolution : how the technology behind bitcoin is changing money, business, and the world. New York: Portfolio / Penguin, 2016

(4) Sotos JG, Houlding D. Blockchains for data sharing in clinical research: trust in a trustless world. (Blockchain Application Note #1.) March 2017. https://simplecore.intel.com/itpeernetwork/wp-content/uploads/sites/38/2017/05/Intel_Blockchain_Application_Note1.pdf

Rejected by the New England Journal of Medicine in November 2016.
Written with Dr. Lester Russell.
See this post for a more expansive version.

In assessing the possible impact of machine learning on clinical medicine, Obermeyer and Emanuel(1) describe the narrowing gap between human vs. computer analysis of images, and declare that “machine learning will displace much of the work of radiologists and anatomical pathologists.”

We hesitate to agree, owing to the Jevons paradox and elastic demand for medical imaging.

In 1865, the economist William Jevons predicted that more efficient coal-burning in manufacturing plants would not lower the nationwide consumption of coal. Instead, the lower cost per unit of energy would increase demand for coal energy and thereby increase consumption(2).

Thus, assuming machine interpretation lowers the cost per imaging study, future human case loads will depend on the quantitative balance between a Jevonsonian increase in imaging (if any(3)) and the fraction of cases where computers completely exclude humans (e.g. only 25% for contemporary computerized Pap smear interpretation(4)).

Clearly, major changes are coming, but, given healthcare’s tangled economics, it is premature to affirm that computerized image interpretation will decimate physician workloads.

(1) Obermeyer Z, Emanuel EJ. Predicting the future -- big data, machine learning, and clinical medicine. N Engl J Med. 2016; 375: 1216-1219.

(2) Jevons WS. The Coal Question. London: Macmillan and Co., 1865. Pages 102-104.

(3) Polimeni JM, Mayumi K, Giampetro M, Alcott B. The Jevons Paradox and the Myth of Resource Efficiency Improvements. New York: Earthscan Routledge, 2008.

(4) Bengtsson E, Malm P. Screening for cervical cancer using automated analysis of Pap-smears. Computational and Mathematical Methods in Medicine. 2014; Article ID 842037.   http://dx.doi.org/10.1155/2014/842037

See this post for a shorter version.

Eminent academicians are editorializing in the New England Journal of Medicine (1) (and elsewhere) that artificial intelligence computer systems are about to disrupt both the practice and the profession of medicine, to the point where some specialties, such as radiology, may vanish entirely.

Such prognostications are symptoms of an unfortunate condition I call “techno-hypnosis:” entrancement with an emerging technology while ignoring lessons of history from the wider world.

Let’s look at the radiology claim. It stems from recent breakthroughs in machine learning techniques – a type of artificial intelligence – that now allow computer scientists to develop, with relative ease, software that recognizes disease in medical images, for example, signs of a fracture in a leg bone’s x-ray, or signs of a cancer in a mammogram.

Moreover, formal studies have demonstrated that such software can detect certain features with the same accuracy as humans, and sometimes better.

But this does not mean the sky is falling for radiologists, for several reasons.

First, we have been here before. Computers started reading the squiggly lines on electrocardiograms in the 1970s, and quickly became better at it than 99% of physicians. Yet, cardiologists today still read EKGs – in vast numbers – and they read them faster because the computer does the legwork of making basic measurements and calling out specific features onto which the cardiologist can apply his or her deeper skills.

Second, success in a few trials does not guarantee success in the wider world, a problem widely known in the artificial intelligence field for decades. All software systems have limitations, and – especially with complex systems – these limitations are often not apparent until they leave the test bed and encounter the unrestrained chaos of the real world. Maybe the new AI techniques have licked this problem, but I doubt it. Witness last month’s report that gene sequencing laboratories have for 10 years been incorrectly classifying certain gene variants as harmful (2). The labs were violating statistical assumptions that weren’t even recognized as being assumptions, but are now obvious in retrospect. AI systems in the real world would be subject to exactly the same type of error.

Third, the new AI technologies have a huge problem: they cannot explain themselves to a human, and a sub-genius human cannot understand how they work. Humans discovered the peril of this situation 2500 years ago, when the ancient Greeks wrote stories of wars started by cryptic pronouncements from the oracle at Delphi, pronouncements the oracle refused to explain. Fortunately, DARPA (the Defense Advanced Research Projects Agency) knows its classical literature and is launching an initiative to add explanatory features to the new AI. It won’t be easy.

Finally, even if AI systems are spectacularly successful, human radiologists are not going to go away. For the sake of argument, let’s define “spectacular” as meaning that the cost of an MRI scan, including interpretation, drops to $1 because AI software has superhuman accuracy. If that happened, then every physician visit by every patient would include an MRI scan. (Why guess if your headache is a sinus headache, when it only costs $1 to see?) The resulting explosion of MRI scans would generate large workloads for human radiologists, even if they had to review only 1 in 100 of the AI system’s interpretations.

This principle, well-known in economics, is called the Jevons paradox, after William Stanley Jevons. In 1865 Jevons noted that the greater fuel efficiency of the new Watt steam engine, compared to the then-standard Newcomen engine, would not lower coal consumption, but would increase it because the lower price per unit of energy would allow more people and companies to consume coal (3).

Clearly, major changes are coming to cognitive medical specialties, but, given healthcare’s tangled economics, it is premature to affirm that computerized image interpretation will decimate radiology workloads.

(1) Obermeyer Z, Emanuel EJ. Predicting the future -- big data, machine learning, and clinical medicine. N Engl J Med. 2016; 375: 1216-1219.

(2) Manrai AK, et al. Genetic misdiagnoses and the potential for health disparities. N Engl J Med. 2016; 375: 655-665.   PubMed 27532831

(3) Jevons WS. The Coal Question. London: Macmillan and Co., 1865. Pages 102-104.   https://archive.org/details/coalquestionani00jevogoog/page/n8 https://en.wikipedia.org/wiki/The_Coal_Question

First published on WSJ.com on Sept. 25, 2016

Having recently proposed [changes at the US Food and Drug Administration](this post) that could benefit all persons who take or will take medications, here I’ll present a simple modification of healthcare payment policies that could yield large dividends, too – in health, in finances, and in patient satisfaction.

The principle is straightforward: pay physicians a bonus for providing continuity of care. In other words, insurers (including Medicare) would pay a bit extra to a physician who sees a particular patient multiple times over a long period, versus a physician seeing the same patient for the first time.

A formula, based on the number of visits and time, would calculate the bonus. For example, a family physician could be paid 10% more when seeing a patient for the 10th time in 10 years. A hospitalist might be paid 8% more for seeing an inpatient 4 times in 5 days. The bonus’s size should entice health plans to modify their scheduling practices. With computers now at the heart of all medical billing, multiple formulae of arbitrary complexity could be defined, even incorporating the diagnosis.

It is surprising that health systems have not to date been incentivized to provide continuity of care, considering all of its benefits.

Pre-eminently, everyone likes knowing their doctor. And doctors like knowing their patients. But, more practically, deeper physician-patient relationships – as the incentive aims to produce – helps physicians practice better medicine, for multiple reasons.

First, it allows physicians to put the patient’s complaint in context. For example, if, by long association, Dr. Smith knows that Ms. Jones is not a complainer, Dr. Smith will take notice if Ms. Jones actually complains of something, even if seemingly minor.

Trust is the second positive from longer physician-patient relationships. A discouraging number of patients do not follow physician-prescribed treatments. A patient who walks into an exam room already having trust in the physician is more likely to adhere to the physician’s treatment – this is simple human nature. Moreover, knowing that a patient will be returning multiple times enables a physician to eschew winning each battle with the patient, and instead follow a strategy to win the war.

Third, and just as rooted in human nature, a patient hospitalized with a serious illness is going to feel better immediately if a friendly, known face walks into their hospital room. This is important because, ultimately, the medical profession exists to relieve suffering, and everyone in a hospital bed is suffering mentally, if not physically.

Fourth, a long term relationship between physician and patient – and the patient’s family – enables the physician and patient to say difficult things at the end of life. Much [expensive] medical care near the end of life is futile, and although physicians know that going in, they will treat aggressively by default. Aggressiveness is fine when all parties agree, but genuine agreement can occur only when openness is full – again, this is human nature.

None of this is news. We all know that trust is earned, and that the root of good medical care is knowing what’s going on with your patient’s body and mind. Indisputably, care today is fragmented.

Would overall healthcare costs decline? It is an experiment worth performing. Improving adherence to medications and enabling more well-informed end-of-life decisions are just two areas from which overall cost savings could come. Large health insurers probably have enough data on which to develop the first formulae.

Ample evidence shows that changes in healthcare payment policies can have rapid, massive influence on medical practice – witness Medicare’s earth-shaking adoption of “value-based” reimbursement. With suitable protections (barring insurers from surcharging long-term patients), incentivizing continuity of care would spread a practice that every one of us would appreciate.