Science and Technology links (November 24th, 2017)

Women earned majority of doctoral degrees in 2016 for 8th straight year and outnumber men in grad school 135 to 100.

Materialists use Facebook more frequently, because they compare themselves to others, they objectify and instrumentalize others, and they accumulate friends.

The modern office chair, with wheels, was invented by Charles Darwin. Or so says a Wikipedia article.

The famous Harvard professor Clayton Christensen says that half of all colleges are bound for bankruptcy. (I am skeptical.)

We can rather easily multiply the lifespan of worms. We might ask though whether such actions actually slow down aging, or just prevent death. The former is true. Researchers found enhanced organ functionality in older, long-lived mutants.

I have always been fascinated by how poor synthesized speech sounds. For all the progress made so far, Siri still sounds like a machine. DeepMind had demoed better sounding synthesized speech a few months ago, but it was impractical computationally. They have now announced that they have a computationally practical version. Thus you can safely predict that within a few short years, synthesized speech coming out of most your devices will pass the Turing test: you won’t be able to differentiate voice coming out of your smartphone from actual human voice.

Echoing Tyler Cowen, Tim Hartford report on research research that found that

Companies still invest heavily in innovation, but the focus is on practical applications rather than basic science, and research is often outsourced to smaller outfits whose intellectual property can easily be bought and sold.

If you think that the solution is more government research grants… please consider that the surest way to be denied a government research grant is to propose a project that has a good chance of failing. Failure must not be an option if you want your research grant to be a success. It is that simple: governments are risk averse. And probably rightly so… when governments take chances, things often end poorly.

The data is in: coffee is healthy.

I keep telling people that scholars routinely cite papers that they have never read. I often get incredulous looks. Well. A made-up article got almost 400 citations.

About half of all men (me included) will suffer from male-pattern baldness. We still don’t know exactly what causes it and we do not have a handy cure for it. We have now explained 38% of the risk using genetic analysis.

Professor Kambhampati is a pacifist that does not support the campaign to stop “killer robots”. I share many of his thoughts.

Stress is good: Mitochondrial stress enhances resilience, protects aging cells and delays risk for disease.

Bees can be left or right handed.

The sex of the mice handlers seems to make a difference in drug experiments. (I am skeptical.)

How often do superior alternatives fail to catch on?

Many of us rely on a Qwerty keyboard, at least when we are typing at a laptop. It is often said that the Qwerty keyboard is inferior to clearly better alternatives like the Dvorak keyboard. However, this appears to be largely a myth backed by dubious science.

There is the similarly often repeated story of VHS versus Betamax tapes, when people would record video on tapes. The often told story was that Betamax lost to VHS despite being technically superior. But VHS tapes could record whole 2-hour movies whereas Betamax could not: so VHS was indeed superior.

It is often said that birds have far superior lungs than mammals. So mammals are failures compared to birds… However, bats (which are mammals) have superior lungs than either terrestrial mammals or birds. This suggests that mammals can acquire better lungs when they need it.

I fear that many of the stories about us being stuck with inferior products due to market failures, or about animals being stuck with inferior organs due to evolutionary dead-ends, might actually be weak or false stories.

Credit: This post was inspired by an email exchange with Peter Turney.

You are your tools

I believe that there are no miracle people. When others get the same work done as you do, only much faster, they are almost surely using better tools.

Tools are not always physical objects. In fact, most tools are not physical per se. For example, mathematics is a great tool. Word processors are another tool. Google is also a tool.

Intellectuals have tools to help them be productive. They have books. They have computers. They have software. They also have models, frameworks, and theories.

For example, I studied Physics, so I learned about how physicists think… and it is not how most people think. They have these tricks which turn difficult problems into far easier problems. The main lesson I took away from Physics is that you can often take an impossibly hard problem and simply represent it differently. By doing so, you turn something that would take forever to solve into something that is accessible to smart teenagers.

To illustrate what I have in mind… most people who have studied mathematics seriously, even teenagers, can quickly sum up all numbers in a sequence. For example, what is the sum of the numbers between 1 and 99. That sounds hard? So maybe you can look up a formula online. Maybe. But once you know the “trick”, you can do it in your head, quickly, without effort. There is no miracle involved. To sum up the numbers between 1 and 99, just pair up the numbers. You pair 1 with 99, 2 with 98… and so forth, up to 49 and 51. So you have 49 such pairs, and each pair, sums up to 100 (99+1, 98+2,…). So you have 49 times 100 which is 4,900. Then you have to add the remaining number (50), so that the sum is 4,950.

We don’t know yet what intelligence is. It is not something as simple as how many neurons you host in your neocortex… Dolphins have more such neurons than you do. It is probable that, in time, we will see that what defines intelligence is our ability to build upon new tools.

For some reason, the smartest among us have access to better tools. And that’s ultimately why they can run circles around you and I.

They can’t easily transmit their tools. It takes work, but it tends to happen. A few hundred years ago, most people could not read and write. It was widely believed that most people could never learn to read and write. Until fairly recently, the ability to read without speaking out loud was regarded as a very rare trait, a sure sign of high intelligence. We now expect even the dumbest kids in high school to read without speaking out loud.

Summing up the numbers between 1 and 100 in your head was, no doubt, a great feat once upon a day. Today it is something that all kids in Singapore know how to do.

You should be constantly trying to expand the number of tools at your disposal. It is a particular version of the growth mindset: the belief that you should always seek to better yourself, by acquiring new tools.

You might reasonably ask… “I have whatever tool that I learned to use, and it is good enough for what I do usually. Why would I invest in learning something new if I don’t feel any urgent need to do so?”

My answer is that acquiring new tools is the surest way to get smarter.

Further reading: Stop Using Excel, Finance Chiefs Tell Staffs at the Wall Street Journal.

Do relational databases evolve toward rigidity?

The Hanson law of computing states that:

Any software system, including advanced intelligence, is bound to decline over time. It becomes less flexible and more fragile.

I have argued at length that Hanson is wrong. My main argument is empirical: we build much of our civilization on old software, including a lot of open-source software.

We often build new software to do new things, but that’s on top of older software. Maybe you are looking at your smartphone and you think that you are using software built in the last 4 years. If you think so, you are quite wrong.

So it is not the case that old software becomes obviously less useful or somehow less flexible with time. Yet, to adapt to new conditions, old software often needs “rejuvenation” which we typically call “refactoring”. Old database systems like MySQL were designed before JSON and XML even existed. They have since been updated so that they can deal with these data types efficiently.

So old widely used software tends to get updated, refactored, reengineered…

Viewed at a global scale, software evolves by natural selection. Old software that cannot adapt tends to die off.

There has been a fair amount of work in software aging. However, much of the work is of an interested nature: they want to provide guidance to engineers as to when they should engage into refactoring work (to rejuvenate their software). They are less interested in the less practical problem of determining how software evolves and dies.

Software often relies on database tables. These tables are defined by the attributes that make them up. In theory, we can change these attributes, add new ones, remove old ones. Because open-source software gives us access to these tables, we can see how they evolve. Vassiliadis and Zarras recently published an interesting empirical paper on this question.

Their core result is that tables with lots of attributes (wide tables) tend to survive a long time unchanged. Thinner tables (with fewer attributes) die young. Why is that? One reason might be that wide tables covering lots of attributes tend to have lots of code depending on it. Thus changing these tables is expensive: it might require a large refactoring effort. Thus these wide tables tend to stick around and they contribute to “software rigidity”. That is, old software will accumulate these wide tables that are too expensive to change.

I believe that this “evolution toward rigidity” is real. But it is less of a general feature of software, and more of a particular defect of the relational database model.

This defect, in my view, is as follows. The relational model makes the important recognition that some attributes depend on other attributes (sometimes called “functional dependencies”). So if you have the employee identifier, you can get his name and his rank. From his rank, you might get his salary. From this useful starting point, we get two problems:

  1. Instead of simply treating these dependencies between attributes as first-class citizens, the relational model does away with them, by instead representing them as “tables” where, somehow, attributes need to be regrouped. So, incredibly, the SQL language has no notion of functional dependency. Instead, it has keys and tables. These are not the same ideas!

    Why did functional dependencies get mapped to keys and tables? Simply because this is a natural and convenient way to implement functional dependencies. So we somehow get that “employee identifier, name, rank” get aggregated together. This arbitrary glue leads to rigidity as more and more attributes get lumped together. You cannot reengineer just one dependency or one attribute, without possibly affecting a lot of code.

  2. Functional dependencies are nice, but far more inflexible and limited than it seems at first. For example, some people have more than one name. People change name, actually quite often. Some information might be unknown, uncertain. To cope with uncertain or unknown data, the inventor of the relational model added “null” markers to his model, and some kind of three-value logic that is not even consistent. In a recent paper with Badia, I showed that it is not even possible, in principle, to extend functional dependencies to a open-world model (e.g., as represented by disjunctive tables).

So I would say that relational databases tend to favor rigidity over time.

There are some counterpoints that may contribute to explain why the sky is not falling despite this very real problem:

  • Programmers have a pragmatic approach. In practice, people have never really taken the relational model to its word: SQL is not a faithful implementation of the relational model. Ultimately, you can use a relational database as a simple key-value store. Nobody is forcing you to adopt wide tables. So there is more flexibility than it appears.
  • There are many other instances of rigidity. Constitutions change incrementally because it is too hard to negotiate large changes. Biology is based on DNA and it is unlikely to change anytime soon. Mathematics is based on standard axioms, and we are not likely to revisit them anytime soon. So it is not surprising that we end up locked into patterns. And it is not necessarily dramatic. (But we should not underestimate the cost: mammals have lungs that are far less efficient than the lungs of birds. Yet there is no obvious way for mammals to evolve a different lung architecture.)
  • We have limited the rigidity when we stopped relying universally on SQL as the standard interface to access data. In the web era, we create services that we typically access via HTTP requests. So the rigidity does not have to propagate to the whole of a large organization.

Credit: Thanks to Antonio Badia and Peter Turney for providing me with references and insights for this post.

Science and Technology links (November 17th, 2017)

Josiah Zayner, a biochemist who once worked for NASA, became the first person known to have edited his own genes (…) During a lecture about human genetic engineering that was streamed live on Facebook, Zayner whipped out a vial and a syringe, then injected himself. Now, following in his footsteps, other biohackers are getting ready to take the plunge and tinker with their own genes. Zayner’s experiment was intended to boost his strength by removing the gene for myostatin, which regulates muscle growth. A similar experiment in 2015 showed that this works in beagles whose genomes were edited at the embryo stage. He injected himself (…) to remove the gene. (Biohackers are using CRISPR on their DNA and we can’t stop it)

Human beings definitively do not have the largest brains:

We found that the long-finned pilot whale neocortex has approximately 37.2 × 109 neurons, which is almost twice as many as humans, and 127 × 109 glial cells. Thus, the absolute number of neurons in the human neocortex is not correlated with the superior cognitive abilities of humans (at least compared to cetaceans) as has previously been hypothesized.

We can make old mice smarter by tweaking just one gene according to an article published in Nature:

This study demonstrates for we believe the first time in vivo that 6 months after a single injection of s-KL into the central nervous system, long-lasting and quantifiable enhancement of learning and memory capabilities are found. More importantly, cognitive improvement is also observable in 18-month-old mice treated once, at 12 months of age.

I stumbled on an older post (2015) by Marcel Weiher about his views on where software is headed:

(…) for most performance critical tasks, predictability is more important than average speed (…) Alas the idea that writing high-level code without any concessions to performance (often justified by misinterpreting or simply just misquoting Knuth) and then letting a sufficiently smart compiler fix it lives on. I don’t think this approach to performance is viable, more predictability is needed and a language with a hybrid nature and the ability for the programmer to specify behavior-preserving transformations that alter the performance characteristics of code is probably the way to go for high-performance, high-productivity systems.

He is arguing that engineers have made it hard to reason about performance, and to design software for the needed performance. When we are stuck with unacceptable performance, we are often stuck… unable to know how to fix the problems.

We found a single gene that makes you live seven years older on average than your peers.

Exercise increases the size of your brain.

Fast exact integer divisions using floating-point operations (ARM edition)

In my latest post, I explained how you could accelerate 32-bit integer divisions by transforming them into 64-bit floating-point divisions. Indeed, 64-bit floating-point numbers can represent accurately all 32-bit integers on most processors.

It is a strange result: Intel processors seem to do a lot better with floating-point divisions than integer divisions.

Recall the numbers that I got for the throughput of division operations:

64-bit integer division25 cycles
32-bit integer division (compile-time constant)2+ cycles
32-bit integer division8 cycles
32-bit integer division via 64-bit float4 cycles

I decided to run the same test on a 64-bit ARM processor (AMD A1100):

64-bit integer division7 ns
32-bit integer division (compile-time constant)2 ns
32-bit integer division6 ns
32-bit integer division via 64-bit float18 ns

These numbers are rough, my benchmark is naive (see code). Still, on this particular ARM processor, 64-bit floating-point divisions are not faster (in throughput) than 32-bit integer divisions. So ARM processors differ from Intel x64 processors quite a bit in this respect.

Fast exact integer divisions using floating-point operations

On current processors, integer division is slow. If you need to compute many quotients or remainders, you can be in trouble. You potentially need divisions when programming a circular buffer, a hash table, generating random numbers, shuffling data randomly, sampling from a set, and so forth.

There are many tricks to avoid performance penalties:

  • You can avoid dividing by an arbitrary integer and, instead, divide by a known power of two.
  • You can use a divisor that is known to your compiler at compile-time. In these cases, most optimizing compilers will “optimize away” the division using magical algorithms that precompute a fast division routine.
  • If you have a divisor that is not known at a compile time, but that you reuse often, you can make use of a library like liddivide to precompute a fast division routine.
  • You can reengineer your code to avoid needing a division in the first place, see my post A fast alternative to the modulo reduction.

But sometimes, you are really stuck and need those divisions. The divisor is not frequently reused, and you have lots of divisions to do.

If you have 64-bit integers, and you need those 64 bits, then you might be in a bit of trouble. Those long 64-bit integers have a terribly slow division on most processors, and there may not be a trivial way to avoid the price.

However, if you have a 32-bit integers, you might have a way out. Modern 64-bit processors have 64-bit floating-pointer numbers using IEEE standards. These 64-bit floating-point numbers can be used to represent exactly all integers in the interval [0,253). That means that you can safely cast your 32-bit unsigned integers as 64-bit floating-point numbers.

Furthermore, common x64 processors have fast floating-point divisions. And the division operation over floating-point numbers is certain to result in the closest number that the standard can represent. The division of an integer in [0,232) by an integer in [1,232) is sure to be in [0,232). This means that you can almost replace the 32-bit integer division by a 64-bit floating point division:

uint32_t divide(uint32_t a, uint32_t b) {
  double da = (double) a;
  double db = (double) b;
  double q = da/db;
  return (uint32_t) q;

Sadly, if you try to divide by zero, you will not get a runtime error, but rather some nonsensical result. Still, if you can be trusted to not divide by zero, this provides a fast and exact integer division routine.

How much faster is it? I wrote a small program to measure the throughput:

64-bit integer division25 cycles
32-bit integer division (compile-time constant)2+ cycles
32-bit integer division8 cycles
32-bit integer division via 64-bit float4 cycles

These numbers are rough, but we can estimate that we double the throughput.

I am not entirely sure why compilers fail to exploit this trick. Of course, they would need to handle the division by zero, but that does not seem like a significant barrier. There is also another downside to the floating-point approach: it generates many more instructions.

Regarding signed integers, they work much the same, but you need extra care. For example, most processors rely on two’s complement notation which implies that you have one negative number that cannot be represented as a positive number. Thus implementing “x / (-1)” can cause some headaches. You probably do not want to divide signed integers anyhow.

I plan to come back to the scenario where you have lots of 64-bit integer divisions with a dynamic divisor.

This result is for current Intel x64 processors, what happens on ARM processors is quite different.

Fast software is a discipline, not a purpose

When people train, they usually don’t try to actually run faster or lift heavier weights. As a relatively healthy computer science professor, how fast I run or how much I can lift is of no practical relevance. However, whether I can walk the stairs without falling apart is a metric.

I am not an actor or a model. Who cares how much I weight? I care: it is a metric.

I could probably work in a dirty office without ill effect, but I just choose not to.

So when I see inefficient code, I cringe. I am being told that it does not matter. Who cares? We have plenty of CPU cycles. I think you should care, it is a matter of discipline.

Yes, only about 1% of all code being written really matters. Most people write code that may as well be thrown out.

But then, I dress cleanly every single day even if I stay at home. And you should too.

I do not care which programming language you use. It could be C, it could be JavaScript. If your code is ten times slower than it should, I think it shows that you do not care, not really. And it bothers me. It should bother you because it tells us something about your work. It is telling us that you do not care, not really.

Alexander Jay sent me a nice email. He reviewed some tricks he uses to write fast code. It inspired me these recommendations:

  • Avoid unnecessary memory allocations.
  • Avoid multiple passes over the data when one would do.
  • Avoid unnecessary runtime inferences and branches.
  • Avoid unnecessary performance-adverse abstraction.
  • Prefer simple value types when they suffice.
  • Learn how the data is actually represented in bits, and learn to dance with these bits when you need to.

Alexander asked me “At what point would you consider the focus on optimization a wasted effort? ” My answer: “At what point do you consider being fit and clean a wasted effort?”

There is a reason we don’t tend to hire people who show up to work in dirty t-shirts. It is not that we particularly care about dirty t-shirts, it is that we want people who care about their work.

If you want to show care for your software, you first have to make it clean, correct and fast. If you start caring about bugs and inefficiencies, you will write good software. It is that simple.

China is catching to the USA, while Japan is being left behind

I have previously reported that there has been a silent revolution in science where countries like China, India, South Korea… that previously contributed few research articles… have started to catch up and even exceed the productivity of the western world.

The National Science Foundation (NSF) in the US has published a report on Science and Engineering Publication Output Trends. Their subtitle is “Rise of Developing Country Output while Developed Countries Dominate Highly Cited Publications”.

The report echoes my earlier observations: lots of new countries are become scientific powers in terms of publication output, with China leading the charge. The major claim made by the report is that developed countries continue to dominate in terms of highly cited publications.

I should preface any further analysis by reminding us that such numbers should be interpreted with care. They count the number of research articles produced per country along with other metrics such as which fraction of those is highly cited. Counting citations and research papers is an imperfect way to measure scientific output. Bibliometrics is a field that is full of methodological issues. So we should not try to compare countries on this basis only. Furthermore, the report does not seem concerned with “per capita” analysis. For example, though the European Union appears to fare quite well according to this report&emdash;i.e., they indicate that it publishes more than the US—we know that it is no longer true once you factor in the number of citizens which is higher in the European Union (500 million people versus a bit over 300 million people in the USA). So it might be misleading. Generally, the report does not shy away from comparing small entities to large ones, and that is a problem.

Back to the main claim of the report: though China is rising in output, countries like the US still have a larger fraction of their research articles that are highly cited.

If you accept that Japan is a “developed country”, and you should, then you are forced to view it as an exception from this observation. What is surprising to me regarding the numbers is the fact that Japan seems to underperform compared to what we might expect. They have barely expanded their publication volume and their citation patterns seem to be closer to India than the US. Back in 2008, I already observed that Japan produces far fewer research articles per capita than other rich countries, so it is nothing new. It reminds me of an August 2017 article in “Nature: Budget cuts fuel frustration among Japan’s academics:

young researchers, facing unstable employment conditions and economic uncertainty, are forced to aim for results that can be accomplished in the short term and in which true originality and creativity are difficult to realize.”

Furthermore, the data presented in the paper clearly indicate that countries like China are bridging the gap with respect to how impactful their work is (see Fig. 2 in the report), on top of bridging the gap with respect to the total volume (see Table 1 in the report). It should be viewed as remarkable especially if you consider that most scientific papers are written in English and published by American or European publishers and that most of the most prestigious universities are in the US. And of course, China lags behind the US considerably in research spending, both in absolute dollars and in a percentage of GDP.

A better title for the NSF report might have been “China is catching to the USA, while Japan is being left behind”.

Does any of it matter? Many people believe, or assume, that great output in terms of research articles should cause economic prosperity and innovation. I have post entitled Does academic research cause economic growth? that makes the contrary point. That is, though China is catching up in terms of scientific output, this may be a consequence of their prosperity: they can now afford to have their very best minds work on producing research articles. It is much easier for rich countries to fund people so that they can publish in Nature. So being rich will allow you to catch up.

But Japan shows that you can be a very rich country and choose not to produce many great research articles. In the least, this establishes that you do not need to produce many great research articles to be prosperous.

Science and Technology links (November 11th, 2017)

Read the following quote from the New York Times:

Business is taking an interest in artificial intelligence, or A.I., and some professors, are forming or joining companies to capitalize on the expected boom. But the new move toward commercialization is disrupting the academic community and provoking fears that university research will be hurt.

Some researchers welcome the business interest. Others, however, complain that corporations are outbidding the campus for scarce personnel, and that work is being diverted from long-term research to short-term problems with immediate application.

It was written in 1982.

The Spectator has a piece arguing that moderate drinking is good for you and that this fact is being suppressed by health professionals. (My thoughts: I think it is far from certain that alcohol is good for you, but it is probably not what will kill you.)

We all know that human activity is wiping out some animal species. According to professor Chris Thomas… a different story can be told

The biological processes of evolutionary divergence and speciation have not been broken in the Anthropocene, they have gone into overdrive. Come back in a million years and we might be looking at several million new species whose existence can be attributed to humans. In the end, the Anthropocene biological revolution will almost certainly represent the sixth mass genesis of new biological diversity. It could be the fastest acceleration of evolutionary diversification in the last half-billion years.

In a short talk, the famous polymath Eric Weinstein argues that Physics is lost to theoretical nonesense… that it is stuck away from reality. He suggests that it might be time for outsiders to come in an disrupt the discipline.

You can spot cancer cells because they burn through sugar very fast, generating heat. Canadian researchers have designed a very low-cost device that cools the skin and then measure the temperature to reliably spot skin cancers. If not for regulations, this could be available cheaply at Amazon in a few years?

A boy was suffering from a terrible genetic skin disease. Doctors took skin samples, corrected the genetic anomaly, regrew the skin and put it back in place. It worked. The vast majority of his skin had to be replaced.

You may have heard that chocolate was a health food? According to a story in the Atlantic, there is some really bad research practices involved:

If you look at the most recent version of their clinical trials registry, it was published in January 2015, three months after they published their Nature Neuroscience article. “So they went back after article was published in Nature and changed their clinical trial registry. There is no mention of this in the trial report,” Drysdale added.


“The bigger concern is that people are trying to do a better job of selling the research itself and not just telling what the straight out answer is,” University of Toronto nutrition researcher Richard Bazinet said. This study only showed that over a period of three months, in a small group, according to a very narrow test that taps a very specific region of the brain, cocoa supplements enhanced cognition. That became “chocolate fights Alzheimer’s” — a message Mars surely appreciated.

As we grow older, we accumulate senescent cells. These are old cells that won’t replicate or function normally. It is strongly associated with several diseases of old age. Currently, there are a few initiatives to develop safe drugs that can kill senescent cells. Amazon’s Jeff Bezos is funding one such initiative. Unexpectedly, some researchers have done breakthrough work that shows we can possibly reverse the senescent state:

The researchers applied compounds called reversatrol analogues, chemicals based on a substance naturally found in red wine, dark chocolate, red grapes and blueberries, to cells in culture. The chemicals caused splicing factors, which are progressively switched off as we age to be switched back on. Within hours, the cells looked younger and started to rejuvenate, behaving like young cells and dividing.

The discovery has the potential to lead to therapies which could help people age better, without experiencing some of the degenerative effects of getting old. Most people by the age of 85 have experienced some kind of chronic illness, and as people get older they are more prone to stroke, heart disease, and cancer.

This week, I learned that molecules inside our cells move really, really fast:

You may wonder how things get around inside cells if they are so crowded. It turns out that molecules move unimaginably quickly due to thermal motion. (…) a molecule will collide with something billions of times a second and bounce off in a different direction. (…) As a result of all this random motion, a typical enzyme can collide with something to react with 500,000 times every second. (…) In addition, a typical protein is tumbling around, a million times per second. Imagine proteins crammed together, each rotating at 60 million RPM, with molecules slamming into them billions of times a second. (…) enzymes spin at up to 700 revolutions per second, which is faster than a jet engine.

Waymo, the Alphabet [i.e., Google] self-driving car company, now has cars driving on public roads in the Phoenix metropolitan area with no one in the driver’s seat.

In a timid study, a small amount of blood plasma from young people was transfused into people suffering from Alzheimer’s. The study failed to show massive benefits:

Nine patients with mild to moderate Alzheimer’s got four once-weekly infusions of either saline (as a placebo) or plasma from 18- to 30-year-old male donors. After a 6-week break, the infusions were switched so that the patients who had gotten plasma got saline, and the patients who had gotten saline received plasma. Another nine patients received young plasma only, and no placebo. (…) The remaining patients who completed the young plasma treatment performed no better overall on objective cognitive tests given by medical staff. However, on average their scores improved slightly—4.5 points on a 30-point scale—on a caregiver survey about whether they needed help with daily activities such as making a meal or traveling. The patients’ average scores also improved modestly on another survey that asks caregivers how well patients can perform simple tasks like getting dressed and shopping. (…) Wyss-Coray agrees that not much can be concluded from the small trial, but says, “It’s tempting to feel hopeful about the improvement in functional scores.” Because the treatment seemed safe, Alkahest now wants to launch another trial that will use just the fraction of the blood plasma that contains growth factors, but not coagulation factors and other components that may do more harm than good. In animals, this plasma fraction was more effective at improving cognition in the mice with an Alzheimer’s-like condition than whole plasma, Wyss-Coray says. Alkahest also wants to test a range of doses and include patients with more severe Alzheimer’s.

Recall that we do know that parabiosis, the transfer of blood between two animals, can age or rejuvenate, but the best evidence suggests that we simply acquire aging factors as we age: there are no magical youthful factors. So it might be a total waste of time to add a bit of young plasma in the blood of older individuals. Instead, we need to find a way to normalize the composition of the blood of older people. This problem may prove very hard with our current technology. Or not.

It is usually assumed that peer review, as practiced by science journal, improves research articles. It may not be so simple:

Peer reviewers fail to detect important deficiencies in reporting of the methods and results of randomised trials. The number of changes requested by peer reviewers was relatively small. Although most had a positive impact, some were inappropriate and could have a negative impact on reporting in the final publication.

Alzheimer’s disease may not start in the brain. Without kidding: there is some evidence that unhealthy teeth and gums could lead to Alzheimer’s.

A young girl that was doomed as a baby due to a spinal muscular atrophy was rescued due to gene therapy.

I always assumed that prestigious journals had better peer review. That might not be so:

If you take all journals and rank them according to prestige,the most prestigious journals publish the least reliable science (at least when you look at the available evidence from experimental fields).


In the field of genetics, it appears that errors in gene names (and accession numbers) introduced are more common in higher ranking journals


statistical power has been declining since the 1960s and that statistical power is negatively correlated with journal rank (i.e., a reproduction of the work above, with an even worse outcome). Moreover, the fraction of errors in calculating p-values is positively correlated with journal rank, both in terms of records and articles

What might explain this effect? Maybe high prestige attracts certain types of people with goals the differ from that of science.

A breast cancer drug appears to cure arteriosclerosis (a condition responsible for heart diseases) in mice.

Systematic reviews of the state-of-the-art are often produced in medical research to summarize the research. I always expected this work to be top-notch. It may not be so, as these studies fail to provide the necessary information to make their analysis reproducible:

the data needed to recreate all meta-analytic effect estimates (…) in only 65% of SRs. Only 30% of SRs mentioned access to datasets and statistical code used to perform analyses.

It is just not possible to check whether a review paper has done its work properly. You cannot double check the computations.

According to the New York Times, the best college majors according to wages are engineering, economics, computer science and nursing. These worst are education, social work, humanities, philosophy, liberal arts, psychology, English, and biology. Business and accounting are somewhere in the middle.

The hole is the ozone layer is shrinking.

Amazon’s Jeff Bezos has funded a company that is buidling a gigantic indoor farm in Seattle.

Lifting weights twice a week is the best routine to keep your muscles as you age. Lifting weights is better than cardio to stay fit as you age.

Sleep deprivation messes with your neurons in a bad way, according to Nature.