Coordinating crime
The “anti-social behaviour coordinator” mentioned by Feedback (16 April) reminds me of the police officer in Rockhampton, Queensland, who has the paradoxical title of “central Queensland crime coordinator”.
Nukes aren't the answer
Alan Berman posits that nuclear power is the only means to a low-carbon future (16 April, p 28). While the energy density achieved by nuclear reactors is attractive at first glance, it still entails approximately one-third as much CO2 emission as gas-fired electricity production. The rich uranium ores required to achieve this reduction are, however, so limited that if the entire world’s electricity demands were met by nuclear power, these ores would be exhausted within three years. Using the remaining poorer ores in nuclear reactors would produce more CO2 emission than burning fossil fuels directly, as researchers in the Netherlands have pointed out ().
Another problem with nuclear power that is addressed neither by Berman nor by Steuart Campbell in his letter, is that of nuclear weapons proliferation. As researchers at the US Institute for Energy and Environmental Research recently noted: “In order to fuel one thousand 1000-megawatt nuclear plants (a common reference case in many nuclear growth scenarios), a global uranium enrichment capacity roughly nine to ten times greater than that currently operating in the United States would be required. If just one percent of that capacity was instead used to manufacture highly enriched uranium (HEU), then enough HEU could be produced every year to make between 175 and 310 nuclear weapons. With an expanded trade in the specialized materials required to build and operate gas centrifuge and other enrichment plants…diversion of supposedly ‘peaceful’ technologies will become harder to identify” ().
Society would be made much safer and debt reduced by looking to non-nuclear means to a low-carbon future.
Whose eyes are they?
Gleefully and unrepentantly quoting out of context, I note from the section on the eye in “Evolution’s greatest inventions” (9 April, p 28) that “It was once believed that the eye had evolved independently on up to 65 different occasions. But new genetic evidence suggests that it only happened once,” and “The first eyes appeared…in a group of trilobites”.
All these years of mourning the extinction of the trilobite, only to discover that I’m a direct descendant of one.
The editor writes:
鈥� It does seem contradictory to state that the genetic evidence suggests a single origin for eyes, and then to say that the first animals to evolve eyes were trilobites. However, the two claims can be reconciled. It all depends on your definition of “eye”.
A few years ago geneticists discovered that eye development in animal embryos is almost always initiated by a gene called pax6, which like many developmental genes is found in many different lineages. That led to the idea that eyes all date back to a common ancestor with a prototype eye. That is probably true, but the prototype was very simple 鈥� one photoreceptor cell and one pigment cell. Various descendants of this organism 鈥� including trilobites 鈥� went on to discover unique ways of improving on this prototype to create fully fledged eyes with lenses. So in a sense, eyes evolved both once and on numerous occasions.
Millennium forum
Your article on the Millennium Ecosystem Assessment (MA) correctly notes that it does not have a direct link to any governmental process (2 April, p 9). But there was some confusion regarding comments I made about the difficulty of seeing how it would link to the key biodiversity-related conventions. Throughout the MA process there has been careful coordination with these five conventions: migratory species (CMS); biological diversity (CBD); endangered species (CITES); wetlands (Ramsar); and desertification (CCD). The findings of the MA will, hopefully, be integrated into future discussions.
The difficulty for the MA 鈥� which many might claim is a weakness 鈥� is not the link to environmental conventions, but the lack of an intergovernmental process to ensure that results and recommendations are converted to policy and action. Whether the MA is the view of one scientist or 1300 scientists, without a process to link science to policy the substantive results of the MA might well “raise the bar” of the debate, as William Clark noted, but it would remain only an academic debate.
The UN Environment Programme (UNEP), the Global Environment Facility and the World Bank are therefore drafting a proposal to make certain, among other things, that the MA results are discussed in an intergovernmental forum, where policy-makers can debate the recommendations and implications and develop a “sense of ownership” over the scientific assessment. UNEP 鈥� along with the various conventions that it sponsors 鈥� is willing to devote a considerable amount of time and effort to bringing the important results of the MA to the attention of governments and to facilitate policy discussions.
Artificial psychosis
Like most articles on strong artificial intelligence, yours focused on the negatives and ignored the positives (23 April, p 32). Strong AI offers enormous, almost unlimited benefits for the future of humanity.
At the same time, your article still misses the real danger posed by AI 鈥� system infiltration. Intelligent machines are uniquely vulnerable to hacking. A few commands at its top level of control could turn almost any AI into a psychotic killing machine, create an ultimate terrorist weapon, or in the most extreme example even create an army of robot soldiers.
The first task of any designer is to make sure their machines are as invulnerable to infiltration as possible: if a machine is invaded it should destroy itself. The good side of this is that as long as it is not infiltrated by a hacker, no AI will ever be a threat to humanity. The catch is that only an expert in machine minds can make this judgement, and there are no experts in this field.
From David Pavett
Before debate in the area of artificial intelligence gets bogged down in the technological gee-whizzery of genetic algorithms and neural networks, we would do well to reflect on one of the deepest theorems of computer science: what one computer can do can be done by any other (even if it might take a bit longer). This means that if any computer can be “intelligent” then so can the one on your desktop (even if it might need to tap into a larger memory).
There are likely to be fewer takers for the idea that computers might one day think when people realise that it implies that even the lowliest computer is a potential thinker, given the right program and enough memory.
Isleworth, Middlesex, UK
From Nicholas Argyris
I wonder how a computer would understand “bares fruit” in the first sentence of the final paragraph of your article: peeling apples, perhaps?
Trivial in itself no doubt, but the underlying point is that human intelligence allows us not to notice, or to take no notice of, many errors of spelling, grammar, punctuation and syntax and to get to the meaning. What are the chances of a computer learning to do this any time soon?
Forres, Morayshire, UK
Harvesting fog
Farmers here on Lanzarote value dew, but do not agree that the stone mulch they use works by feeding dew directly to crop roots (16 April, p 52). Quick examination confirms that even the heaviest dew only moistens the surface of the mulch and that this rapidly evaporates during the day.
Deposition of dew (unlike fog or mist) on a surface is greatly limited by the need for the surface to lose the heat gained as water condenses, normally by radiation into a clear sky. Plants, especially in drought conditions, can take in water directly through their leaves, but the quantities available are of limited significance to a crop with a closed canopy. A stone mulch can, however, be so effective at preventing weed growth that it becomes practical to space the crop plants widely. Not only is this appropriate to Lanzarote’s rainfall conditions (equal to the nearby Sahara) but it also allows other parts of the plant to absorb dew directly through the leaves, measurably improving growth.
For further details of this deceptively subtle pre-scientific technology see
Don't blame Pinker
The splendid Elaine Morgan attacks Steven Pinker for believing in market forces and for placing the wrong emphasis on altruism and cooperation compared with aggression and competition (23 April, p 50). Having just led a 10-session course on Pinker’s book The Blank Slate, I do not recognise these propositions as attributable to Pinker.
Morgan says, “To Pinker and those who share his view, the positive emotions are problematic and secondary 鈥� at best self-deceiving, at worst hypocritical.” Since Pinker does not hold this view, I assume the “sharers” may not have it either.
Words unspoken
You describe a new microphone with non-acoustic sensors that detect speech via the speaker’s nerve and muscle activity, rather than sound (9 April, p 21).
In 1960 I bought a De Havilland Chipmunk aircraft that was equipped with a throat microphone. This was strapped around the neck with two vertical sausage-shaped devices resting against the sides of the throat. It felt strange at first, but I got used to it. After a few weeks I found that I only had to mouth the words, not speak them.
I did hear, however, of people whose throat microphones shorted out, causing nasty burns on the throat. I suppose this is a reason that the microphone never became popular.
And what is fear?
The apparently enigmatic reference embedded in a hacker’s text mentioned in Feedback 鈥� “And what is fear, but a marbled grasshopper?” 鈥� may have a simple explanation (2 April). “A marbled grasshopper” is as succinct description as I have ever come across of the New Zealand weta (Hemideina) a very large and ferocious-looking insect of the grasshopper family.
To wake up and find an adult specimen calmly clinging to the headboard of your bed in the middle of the night will give almost anyone a good shot of adrenalin. What is fear, indeed, but a marbled grasshopper?
First-class research
For the past year CSIRO has been fielding media attacks that stem from a small number of disgruntled “former insiders” who neither understand nor appreciate the leadership and strategic direction of CSIRO. I am disappointed to see New 杏吧原创 recycle this blend of speculation, innuendo and inaccuracy without seeking any input from the organisation (2 April, p 49).
CSIRO’s divisions have been under rolling review over as many years as I can remember, and changes in divisional structure occur because CSIRO wants to make the best of its science. For example, the merger of marine and atmospheric research will bring together teams of scientists working on terrestrial and marine carbon cycles, boosting critical mass and Australian climate and Earth systems science. Furthermore, CSIRO ranks in the top 1 per cent of world scientific institutions in 12 of 22 research fields, and our citations per publication are 30 per cent above the global average. There is no evidence to support the assertion that there is slippage in CSIRO’s cutting edge science.
As for the suggestion that CSIRO is “reinventing itself as a consulting firm”, you should know that consultancies represented just 4 per cent of CSIRO’s total budget in 2004/2005 and are viewed as an ancillary facet of CSIRO. While talk of breaking up CSIRO was rife in some circles some years ago, no one talks sensibly about that option today.
Our communications team is every bit as effective as it was in the past, and Donna Staunton is not only a highly competent communications executive, she has the full confidence of the senior management team. CSIRO has a positive relationship with the media and despite some unpleasant reporting and strained relationships with some journalists, we are working hard to keep all channels of communication open.
As for Geoff Garrett’s reappointment, that is a matter for the minister and the board, but Garrett certainly has the full support of the leadership team. He will continue in his role and CSIRO will continue to deliver great science.
Dogs' hidden costs
Peter Radvan’s suggestion that a museum sniffer dog would be cheaper than a 拢10,000 “e-nose” (23 April, p 31) is romantic but doesn’t consider the dog’s total cost of ownership.
A dog must be fed, housed, exercised, immunised and given veterinary checks. Someone has to remove its waste. My back-of-the-envelope calculation gives a running cost of 拢3000 per annum. Over a 10-year working life that’s three times as expensive.
And the dog has to be trained. Here in New South Wales, training a police sniffer dog costs A$90,000 (拢37,000) 鈥� and I doubt that smelling mouldy paintings is easier than sniffing out a joint.
Besides, no one who has had a dog chew a favourite shoe or a vital document would ever let one near valuable artwork.
Long life, but a dull one
Ray Kurzweil’s quest for immortality with a quadrazillion supplements and intravenous infusions (9 April, p 38) reminds me of a sprightly octogenarian on Have a Go, the 1950s BBC radio talk show hosted by the great Wilfred Pickles. He said that if you stayed away from tobacco, sex and alcohol, you didn’t actually live longer, it just seemed longer.
The Elgin effect
Feedback talks of hearing a name a second time shortly after hearing it in a different context (16 April). In our family this is referred to as the Elgin effect.
There are millions of things happening around us all the time. We only notice that many of these are repeating after someone has drawn our attention to a particular instance. I came up with the name while discussing intuition with a colleague: I took “Elgin” as an example and assured her that, although she believed she hadn’t heard of the place in two years, she would hear it again within 48 hours.
Later that day she contacted me to report that she had overheard a colleague talking about his holiday in Morayshire. Sure enough, in the next sentence he described a “small place just outside Elgin”.
Calling Kierkegaard fans
I’m not quite sure why New 杏吧原创 chose to review a theologian-philosopher’s biography (16 April, p 55) but the review did spark a memory.
Nearly 30 years ago I lived in inner-suburban Melbourne and often travelled by train to visit a friend in Blackburn (then an outer suburb, now not). From the train, somewhere between Blackburn and Hawthorn and on the southern side of the track, I used to see the most wondrous graffito: S脴REN KIERKEGAARD’S EXISTENTIALIST PHILOSOPHY.
It was painted in big black block capitals, over two feet high, along a very long paling fence visible (and probably only visible) from the railway. I knew enough about the subject at the time to know that it was all spelled correctly (I think it even had the slash through the O) and made some sense, and to know that the graffitist and I were both therefore members of a tiny minority.
It has nagged at me ever since, because the train was always moving too fast for me to read any more or to be sure there wasn’t more. The graffitist may have wanted to be enigmatic, may have run out of fence or paint, or may have continued ‘IS INSPIRATIONAL’ or (this being Australia, after all) ‘IS BULLSHIT.’ I just don’t know.
It’s a long time since I lived in Melbourne, but perhaps NS readers can help. Is it still there? Is that all there is or was? And finally (a very long shot, I know), who painted it?
Circular transportation device
I am concerned that Paul Marks may have misled or at least confused readers about the nature of Australia’s patenting system (2 April, p 26). Australia, like many other countries, has a two-tier patent system, and the author’s reference to “the country’s new patenting system” is to the innovation patent, which is a utility model system.
The innovation patent system was introduced in 2001 to provide a patent option that was relatively inexpensive, quick and easy to obtain, and suited inventions with a short commercial life or those incremental or lower-level inventions that would not necessarily meet the inventive threshold required for standard patents. The innovation patent is granted after the patent application passes minimum formality requirements. However, before the innovation patent can be enforced it must be certified. Certification involves search and examination of the patent to ensure that it meets the patentability requirements, including newness and innovation step.
It was under the innovation patent system that an application for a “circular transportation device” was granted an innovation patent after passing the minimum formality requirements. It is important to restate that at this point in the lifecycle, the innovation patent has not been searched or examined and is not legally enforceable intellectual property right. To that end, had a request for certification of the “circular transportation device” been made, the innovation patent would have been revoked because at examination it would certainly have been found not to be new or to contain an innovative step.
I trust that this explanation of Australia’s innovation patent system clearly severs the article’s suggested link between quality of the innovation patent at a stage in its lifecycle where it has not been subject to search and examination and is not an enforceable intellectual property right, and electronic systems used by intellectual property offices for searching prior art.
Symbol grounding
Justin Mullins seems to have missed the real reason why many AI researchers are unhappy about the claims of the creators of Cyc. Many modern fields within AI, such as behaviour-based robotics, artificial life and evolutionary robotics, were developed as a response to the problems of symbolically representing the world 鈥� problems that are typified by the traditional rule-based techniques of Cyc.
Cyc may manipulate the symbol “house” and statistically link it to the symbols “garden” and “roof” within its database, but as the program has no functional relationship with a house, it can never know what a house actually is. This issue in AI is highly contentious and is referred to as the symbol grounding problem.
Although it is true that Joseph Weizenbaum’s program Eliza was highly influential in AI, it is worth pointing out that he used it not to demonstrate that computers could be intelligent, but to highlight both the symbol grounding problem and the inadequacy of the Turing test to indicate intelligence.
DNS poisoning
Interference with domain name system (DNS) server programs is a problem (23 April, p 25). The biggest threat, however, is not to the DNS servers run by major internet service providers. It is to the itty-bitty servers on the edges of companies’ sub-networks. Business users running Microsoft network software are most vulnerable, since many of them do not patch or upgrade properly. I conclude that office workers who, for example, access their personal bank accounts on company time are most at risk of being diverted to malicious sites through “DNS poisoning” of their company’s server.
From Nik
Instead of promoting BIND, the author would serve the community better by mentioning a secure alternative: djbdns. See
Los Angeles, California, US