Survival of the freest
Terry Deacon proposes that relaxation of selection pressure can lead to innovations that would not otherwise have evolved 鈥� “something for nothing” (27 September, p 40). This looks quite credible if we compare it with a simple physical model such as a ball rolling in a hilly place.
The ball, attracted by gravity and reducing its gravitational potential as much as possible, rolls downhill until it stops in a hole 鈥� this is the “optimised” solution, a local optimum on our simplified “fitness landscape”. The ball could reach a much lower-energy state were it not hindered by the hole’s wall. If the walls flatten out, the ball can go further downhill, taking a position even more optimised.
If you take the ball to represent a species; its gravitational potential this species’ potential for survival; and the slopes and walls to represent the external constraints on the species 鈥� then the model says that the evolution of the species may be stuck in a local optimum. It is then well-adapted to live under the prevailing conditions, but may not get further. In Deacon’s bird example, it may not learn new songs.
If the walls flatten out 鈥� if the prevailing constraints relax 鈥� the ball can roll somewhere further: evolution can then explore a much larger playing field and can reach a new optimum.
So-called “genetic programming” techniques (6 March 2004, p 28) are already using this principle to find fast solutions to far more complicated optimisation problems with many variables. To avoid getting stuck in a local optimum far from the most valuable optimum, these try out random combinations of intermediate solutions. And the process in which chromosomes that bear some genetic mutation recombine at random during sexual reproduction is a similar and familiar route to organisms which may be fitter than those with the original set of chromosomes.
Slanted cycling record
You report a new world top speed record for bicycles of 132.4 kilometres per hour (4 October, p 23). But the rules under which that record was set allow for a 0.66 per cent downward slope. For a typical mass of machine and rider at 113 km/h this is worth approximately 190 watts, and even more at 132.4 km/h.
An average person has trouble producing that amount of power even for a moment. That allowable down slope is equivalent to using a hidden motor. So sadly, this is a merely a downhill record, and I argue that it is not a legitimate top speed record.
A simple solution would be instead to require that the finish line must be the highest point on the record-setting course. Within a few years, the top speed record would be just as high and, more importantly, trustworthy.
Excellent supervision
You comment on the investigation into a study published in 2001 in Blood of which Catherine Verfaillie was senior author and mention “inadequate supervision” as one of the factors that permitted publication of data falsified by Morayma Reyes (11 October, p 8). As her former trainees, we want to make some comments regarding our personal experience working under her supervision.
We all received part of our scientific training at the University of Minnesota under her skilful guidance. During the time we were working with her we were mentored closely, having the opportunity to meet her on a weekly basis to discuss in detail our experiments and projects.
Indeed, despite her busy schedule and the large number of people in her lab, Verfaillie was always willing and wishing to help all of us in any aspect of the science, of the design of experiments or teaching us how to write manuscripts or grant proposals. She also had a genuine concern to help us personally.
This thorough personal commitment, complemented by her rigorousness, her tireless capacity for work and drive to help us to pursue our aims with the integrity that is necessary in scientific practice, provided a most valuable learning experience for every one of us.
She spoke openly and honestly about experimental results, and repeatedly emphasised the critical importance of honesty and integrity in science. At no time did she ever insist on receiving particular results, nor make her students feel obliged to produce predetermined findings. On the contrary, she accepted the results we produced and worked with us to better understand and validate our findings, however surprising or unexpected. She frequently asked us to repeat and re-repeat our experiments to ensure our findings were reproducible prior to publication.
Most of us have also been trained elsewhere and have moved on to other academic institutions or into industry. We are all agreed that Catherine Verfaillie’s mentorship and management is at least as appropriate, dedicated and careful as that in other laboratories. From this personal experience we most certainly disagree with any general suggestion that she was or is a poor mentor. Rather, we believe her to be an outstanding scientist and mentor, and consider ourselves extremely fortunate to have had the opportunity to train in her laboratory.
Fantasy modelling
No doubt it is possible to find a mathematical model to predict stock market developments (27 September, p 5 and p 8) 鈥� but only on the assumption that your actions do not influence the market. If you are a significant player in the market your actions 鈥� based on your model 鈥� do influence it, and your choice of model will be part of the market, and another model will be optimal.
If your competitors use the same model as you, yet another model would be optimal, and the likelihood of instability is very great.
From Fenton Robb
You rightly point to the limits of statistical modelling of financial markets (27 September, p 5). Surely similar caution should be applied to all statistical models attempting to anticipate future states of complex systems.
Complex systems are known to be prone to influence by “butterflies” 鈥� very small and possibly remote perturbations that may have disproportionately large effects 鈥� and also by “black swans”, which are perturbations arising from hitherto unknown, ignored or discounted factors. Add the uncertainties about the position of the thresholds for state-changes in component subsystems, and we should be very wary of any predictions about the behaviour of such complex systems.
But doesn’t this apply to climate-change modelling just as it does for modelling of financial markets?
Eyemouth, Berwickshire, UK
Radio retrospect
Marcus Chown’s article on telescope design is a classic example of what happens when a theorist researches a report by talking to other theorists (27 September, p 36). When writing about telescopes, it would be worthwhile talking to the people in windowless labs who build telescopes for other astronomers to use to collect data upon which theorists, who have offices on the top floor with windows, can build their theories.
In practice, the idea of a telescope operating at all wavelengths from radio to optical is a non-starter. Put simply, what works at one wavelength does not work at another. Consider receiver noise, cost, accuracy of surfaces, antenna design, bandwidth-smearing and computer speed, and it becomes apparent that, for example, there is good reason for those narrow-band filters used in current telescopes.
If the experimentalists are interviewed, you might have less theoretical excitement, but you’d be able to discuss crash-landings in Antarctica, what it’s like to work at an altitude of 5000 metres, and the crucial role of duct tape in holding experiments together.
From Tony Foley
Tegmark’s new approach to telescopes has some advantages. It offers high sensitivity and economies on computing power, but at the cost of a loss of flexibility in antenna placement and angular resolution. This combination is very good for some observations 鈥� but not for all. It allows for a very wide bandwidth, but there is no obvious route to observing “all wavelengths” in the next 30 years.
Your article should have acknowledged, as Tegmark does, other innovative radio telescope designs. See, for example, the , the and the . These can be seen as pathfinders for alternative designs for the , which has been under consideration for decades.
Dwingeloo, The Netherlands
From Jet Foncannon
You attribute the discovery of the fast Fourier transform algorithm to James Cooley and John Tukey, as many do. The great German mathematician Carl Friedrich Gauss mentioned the algorithm in 1805, and it has been rediscovered several times, in various versions, since then.
Philadelphia, Pennsylvania, US
Excellent supervision
You comment on the investigation into a study published in 2001 in Blood, of which Catherine Verfaillie was senior author, and mention “inadequate supervision” as one of the factors that permitted publication of data falsified by Morayma Reyes (11 October, p 8). As her former trainees, we want to make some comments regarding our personal experience of working under her supervision.
We all received part of our scientific training at the University of Minnesota under her skilful guidance. During the time we were working with her we were mentored closely, having the opportunity to meet her weekly to discuss in detail our experiments and projects.
Most of us have also been trained elsewhere and have moved on to other academic institutions or into industry. We are all agreed that Catherine Verfaillie’s mentorship and management is at least as appropriate, dedicated and careful as that in other laboratories. From this personal experience we certainly disagree with any general suggestion that she was or is a poor mentor. We believe her to be an outstanding scientist and mentor, considering ourselves extremely fortunate to have had the opportunity to train in her laboratory.
鈥� The full text and list of 22 signatories is online
Electric caveats
You mention General Motors junking its EV1 electric vehicles in 2003 (20 September, p 26). My understanding is that its range at temperatures of -20 掳C was something like 12 miles (20 kilometres). It would be interesting to know what the ranges of the Tesla Roadster, the Toyota Prius and the Honda FCX Clarity are, in strictly electric mode, at such temperatures, which are common in North American winters.
Bouncing burgers
Your question “Meat or not?” (13 September, p 28) need not be such a dilemma, at least in Australia. There’s no need to wait for lab-grown meat. There’s already a solution bouncing around. Many commentators 鈥� including Ross Garnaut, author of the governmental report 鈥� believe that switching from cattle and sheep to kangaroo could lead to a significant reduction in greenhouse gas emissions.
Gay abandonment
I am startled to read an incredibly offensive paragraph in Mary Midgley’s review of Gary Greenberg’s The Noble Lie (13 September, p 46) 鈥� a tacit endorsement of the notion that homosexuality is a disease which was mistakenly deleted from the Diagnostic and Statistical Manual of Mental Disorders (DSM) due to the campaigning of gay rights activists. Midgley also seems to share Greenberg’s lament that it was “the first time in history that a disease was eliminated by the stroke of a pen”.
The editor writes:
鈥� Greenberg examines the cultural history: homosexuality was removed from the DSM for sociopolitical reasons. That does not mean it should have been in there in the first place. It had been added for sociopolitical reasons, not scientific ones, and at that time many homosexuals were glad of this because it shifted the focus away from criminal law and the idea 鈥� perpetrated largely by religious groups 鈥� that it was some kind of moral choice. Both changes were “noble lies” in that they were decisions based on moral attitudes, disguised as science. Neither Greenberg nor Midgley thought that the deletion was a mistake.
Science and philosophy
Mark Buchanan spoils an otherwise interesting article on emergence (4 October, p 12) by describing the work of Mile Gu and colleagues as providing “a promising scientific boost” for the topic, “which tends to get swamped with philosophical arguments”. If science and philosophy are seen as two rival “camps” in this way, it can only be because some scientists are not interested in the philosophical implications of their work. Luckily, there are philosophers of science who are happy to take this on. Surely, in order to fully describe and understand the world, we need both.
Whose big ideas?
In his commentary about “celebrating the real big ideas” (27 September, p 49), Lawrence Krauss once again shows how little understanding many elite scientists have of normal human nature and the “real questions” that move humanity. Ask a person in the street what the deepest, most important questions are and they will certainly not answer: “Is there a multiverse?” or “Are the laws of nature unique?”. Even those with a sound scientific background are more likely to ask: “What is the meaning of life? Is there more to life than what we see? Why so much suffering? Is there hope in this life?”
Look at meeee!
You report that people who commonly upload online content may be “just attention seekers” (4 October, p 23). The team from Hewlett-Packard that carried out the study base their conclusion on the observation that people whose web pages get more hits are more likely to upload more videos. Surely this is exactly the pattern one would expect if the motives of people uploading these videos were genuinely altruistic. Uploading content that no one watches plainly doesn’t help the community. Might an alternative explanation simply be that the more successful contributors (those with the most hits) have more evidence that the community benefits from their efforts than the less successful ones, and that this motivates them to do more work?
Self-control
I was impressed with the possibility of letting complex systems like traffic lights self-organise (9 August, p 28). But this ignores something that urban planners in North America are also guilty of, which is dealing with road traffic and pedestrians separately, even though intersections are shared by all.Any rules for self-organising lights should allow for pedestrians crossing. All too often, pedestrians are stuck at intersections with 10 seconds to cross eight lanes, or they have to wait so long for the crossing signal that they give up and cross anyway.
Autism origins
Maia Szalavitz gave a fascinating account of some recent ideas in autism (20 September, p 34). Rather surprisingly, however, she omitted to mention the work of Laurent Mottron, Jake Burack and others in Montreal on the “Enhanced Perceptual Functioning” model of autism. Originally put forward in 2001, this seems to have anticipated key concepts in the “Intense World” hypothesis which Szalavitz discussed.
Bad atmosphere
Richard Lovett claims that a warmer atmosphere can hold more water than a cooler one (27 September, p 34).
But this is what I call the “windbag argument”, which was disproved by John Dalton in 1802 (see The Last Word, 12 February 2005). Water vapour acts independently of the other gases in the atmosphere.
In fact, it behaves exactly the same in a vacuum. All one can say is that the higher the temperature of the air, the more water vapour there will be in it (but not because the air has held any of it).
Richard Lovett writes:
鈥� This boils down to a semantic argument over the definition of “atmosphere”. To me the atmosphere includes the water vapour. Dalton’s Law does hold that in a mix of gases, each gas’s partial pressure behaves independently of the others. Warmer air means warmer water vapour, which means a high partial pressure 鈥� and on a hot day you get more evaporation than on a cool one.