Not so stupid
Three months ago, you reported on the French National Academy of Science’s forthcoming report on the health effects of low doses of ionising radiation (This Week, 2 September). A radiation biologist from London called the report “stupid”. Is that so?
The International Commission on Radiation Protection based its assessment of risk from high doses of radiation on epidemiological surveys. The ICRP gave particular emphasis to survival after the atom bombs dropped on Japan, in which the dose rates were high. In other data with lower dose rates the risk is smaller. Moreover, the main reason why the ICRP increased its estimates of the carcinogenic effect of radiation in 1990 was the introduction of risk projection models. Data have been recently published which are inconsistent with the model chosen by ICRR
The extrapolation from high to low dose is also controversial. Firstly, cell repair is much more effective in cells which are not dividing. Also, when a high dose of radiation kills a significant proportion of cells, the surviving cells begin to proliferate and this interferes with cell repair. Furthermore, whereas misrepairs are frequent after a high dose or high dose rates, recent papers on repair of DNA lesions suggest that repair mechanisms are far more effective at low doses.
In an editorial in Science, (September 1994, p 1507) Philip Abelson wrote: “The current mode of extrapolating high-dose to low-dose effects is erroneous for both chemicals and radiation. Safe levels of exposure exist.” Most correspondence following that article supported this view and a similar conclusion was reached by Dan Koshland (Science, December 1994, p 1925).
Recent epidemiological surveys give no evidence in favour of the new ICRP cancer risk coefficient. No survey, including those on atom bomb survivors, detects any effect for doses of less than 200 millisieverts in acute irradiation.
The dose due to natural radiation varies in most countries between 1.5 and 6 mSv per year and many surveys have failed to detect any difference in cancer incidence between regions with high and low background radiation. The ICRP’s dose limit of 1 mSv per year may give rise to unjustified anxiety in populations living in areas with high natural radiation levels. Discussion about the carcinogenic effect of low doses of radiation should be based on hard scientific facts and not on prejudices.
Conscious machines
Stuart Sutherland reviewed my Artificial Minds (Review, 7 October) as a chronicle of the artificial intelligence (AI) debates, though it was intended to promote a new paradigm of mind. Too bad.
I completely agree with much of what Sutherland says. AI people, including me, are certainly optimistic about the future of machine intelligence. Too many of them (not including me) believe that intelligence is all that’s needed for mind. I must, however, clarify a few points.
The first AI debate pits the boosters, who expect to see real machine intelligence, even consciousness, against the scoffers who find this idea ridiculous. Sutherland argues as a scoffer, pointing out the ineffability of a person’s sensation of red and concluding that it “could not be captured by any artificial device”. Yes indeed. Nor could it be captured by any other natural one for that matter, not even by another human. This is no way precludes an “artificial device” having its own sensation of red, as do other humans and many animals.
He further argues that “programs could never be conscious in the sense that humans are”. Of course. Neither could apes, dogs, bats or dolphins, all of whom are conscious in their own unique ways. Some AI researchers expect machines eventually to become conscious in their own unique ways.
The second AI debate focuses on artificial neural nets. The symbolic AI side maintains that such nets can, at most, implement symbolic AI structures, and thus can add nothing substantive to AI. The connectionists on the other side claim that their brain-like structures are more suitable for implementing intelligence. Sutherland sides with symbolic AI, asserting that “connectionist AI cannot learn to execute tasks that involve rules, such as mathematics”.
Not so. Whole conferences devote themselves to the study of neural nets learning to act according to rules. Several such have been organised by Lee Giles at NEC-Princeton. John Anderson at Brown University experiments with neural nets that learn to do arithmetic. And, as Max Garzon and I have shown, anything that can be computed (with a Turing machine) can be implemented with a neural net. The issue fuelling the debate between symbolic AI and connectionsism is more subtle. I obviously didn’t make this clear enough in Artificial Minds.
In response to the possibility of systems employing both symbolic AI and connectionism, Sutherland returns to the first debate, asking: “鈥� how could even a combined AI system have a sudden ‘change of mind’ like a human?” The copycat system of Hofstadter and Mitchell, discussed in detail in chapter 13, goes through just such a change of mind in the example presented.
The third AI debate is over the necessity for mental representations. In an attempt to trivialise the debate Sutherland asks, “if visual images, smells, sounds and memories are not representations, what are they?” While one day-to-day usage of the word “representation” might well provoke this question, the debate at hand concerns representation in the sense of a sign or symbol representing something.
The sign or symbol should be more or less arbitrary, and not casually connected to the something being represented. In this sense, the length of the mercury column in a thermometer measures temperature but does not represent it. Also, in the relevant sense, visual images, smells and so on are consequent on entities in the world but are not representations of them. Again, I’m sorry my explanations were not more clear.
Africa appeal
I have belatedly been reading your special issue on Africa. I agree and sympathise with a great deal of what you have written. Africa is not the basket case it is so frequently assumed to be.
But it has been through a terrible time. Thinking of the economic chaos caused here when GDP simply failed to grow, one has to ask how we would cope with a 40 per cent fall in GDP as happened in Zambia with the collapse in copper prices. Equally, it is hard to see us dealing with the ending of virtually the whole of our export trade, as happened to Niger because of the disappearance of the uranium market.
As you point out, education in science and technology is essential for economic progress throughout the continent. But lack of educational materials is a major problem in many places. A few years ago, I was taken to visit a secondary school in Zimbabwe. The library had nothing but a few magazines and the collected works of Enver Hoxha. I was full of sympathy but it was not something I felt I personally could do anything about.
Reading your articles, however, it strikes me that there is a small and not particularly costly step which I and many of your readers could take. It is simply to send our back numbers of New 杏吧原创 to an African school or college. The package may take some time to get there but being a few months late with the news on what is happening in science and technology is an awful lot better than depending on Mr Hoxha’s pearls of wisdom.
Many readers already send back issues of New 杏吧原创 and text-books to African countries via the charity Book Aid International, which responds to requests for material from universities and research centres. Details are available from 0171-733 3577 鈥� Ed
Electrocrimes
With regard to the Orwellian points alluded to in your editorial, it hardly needs pointing out that in Orwell’s vision the activities that today’s city centre surveillance systems are set up to catch would actually have been ignored (Comment, 4 November). Muggers would have been left to mug, rapists to rape, and vandals to vandalise 鈥� so long as the equipment and Inner Party property or members were not on the receiving end.
In fact Big Brother did not watch most people, just those who “thought too much” and might have become politically active in a manner threatening to the Inner Party. The rest were left to entertain themselves with sadism, xenophobia and lynching, general drinking and violence, and being blown up a bit.
The other aspect you mention comes closer to an Orwellian image than you seem to think. There was almost a fashion at one time to call anything that now gets called “jargon”, “newspeak”. But this derogatory stab at what was considered confusion by the use of arcane words got it completely wrong. The purposes of Orwell’s newspeak was to have a language that everybody understood. And only it. Newspeak got rid of words, and did not take up new ones because they might have dangerous ideas associated with them.
Electronic bureaucracy might not set out with such aspirations of intellectual pruning, but at the end of the day “syntax error” may close an option as surely as “crimespeak”.
Harmful or benign?
The National Radiological Protection Board chooses to ignore a considerable body of independent evidence regarding the effects of electromagnetic radiation on natural organisms (Letters, 11 November). The importance of low level background electromagnetic fields (EMFs) to life is emphasised by NASA putting Schumann wave generators on spacecraft, where lack of natural terrestrial EMFs causes immune problems and weight loss. The question is: are some frequency-amplitude electromagnetic windows harmful instead of beneficial?
Two simple studies will very easily demonstrate possible effects of EMFs generated from our electricity supply system. The first is a correlation of households in which someone is or has been a cancer sufferer against distance from substations and power lines. The second is looking for possible variations caused by microwave radar shadows (direct sight and behind hills in undulating topography). Both these could be done as small scale pilot studies in less than one year, given a Geographical Information System and some cooperation between a local health authority, local GPs and a good statistician.
First with the chip
I read Tim Hunkin’s comments on The Microprocessor: A Biography with interest (Review, 4 November). I can certainly sympathise with his experiences of early “development kits” and the difficulty of actually doing anything useful with microprocessors.
As producer and presenter of a BBC/Open University radio series on the history of electronics, I interviewed Gordon Moore of Intel, and Ted Hoff. Hoff certainly acknowledged the part played by Fredrico Faggin in the development of microprocessors, as well as many other software and chip fabrication engineers. Hoff himself believed Faggin would have been the only person able to describe the role of every logic gate in the 8080 design from his head.
I’m sure Intel is big enough to speak up for itself, but my understanding is that Hoff’s own role (and his claim to be the inventor of the microprocessor) is his suggestion (based on his design work) that one general purpose, programmable chip could fulfil the brief of the client, the Japanese Busicom company. As I understand from Moore and Hoff, what Busicom actually wanted was a set of 13 different bespoke chips 鈥� the multipurpose chip solution was suggested to Busicom by its supplier, Intel.
The subsequent design, the rights to which were bought back by Intel from Busicom, became the 4004 chip 鈥� but I have heard arguments that the 8008 (the first chip to have a hardware interrupt line) or even the later 8080 (in performance terms) was the first “real” microprocessor. Who is the inventor? The person who first has the idea, or the one who lays out the logic design, or the one who fits this into silicon?
Hunkin complains of the difficulty in understanding the complexity of modern chips 鈥� who would disagree? But I wonder if our bafflement is any different from that of people in the 1920s, faced with the dawn of the wireless age? Many concepts which today seem simple 鈥� the tuned circuit, the working of the thermionic valve, the existence of radio “sidebands” 鈥� were themselves the subject of complaint and debate.
Rambling rabbi
Your “Complete scientific history of the Universe” (28 October) has Leonardo Fibonacci introducing Arabic numerals to Europe in 1202. Had you specified “successfully”, your history would be less incomplete.
Although it did not catch on, an earlier introduction was actually made by peripatetic polymath Abraham ibn-Ezra (1093-1167) of Spain (the same Rabbi Ben Ezra of Robert Browning’s poem), who travelled and taught in the cities of Western Europe and the Mediterranean, including London.
He used the first nine letters of the Hebrew alphabet for the digits one to nine, and invented a special symbol for zero, allowing the place value notation where shifting a digit to the left multiplies it by ten.
Sinister planes
I think your correspondent will find that if Second World War fighter pilots did have a tendency to turn left under stress, it was almost certainly because of the bias of the aircraft and not any inherent bias in the pilot (Letters, 4 November).
Any pilot of a single-engine propeller-driven aircraft will tell you that engine torque is always a consideration and the larger the engine/propeller combination the greater a consideration it becomes. Consequently, fighter planes had enormous torque.
Mostly, when viewed from the cockpit, they had clockwise rotating propellers, which meant that the whole aircraft tended to rotate anticlockwise, or left wing down. That is, the aircraft tended to turn left naturally so, when you were attacked, it made sense to turn in the direction the aircraft would move quickest 鈥� that is, until your attacker got wise to it.
Apparently the Red Baron cottoned on to a related phenomenon at an early stage, resulting in his great success as a fighter pilot. Some early British fighter aircraft had rotating radial cylinder blocks which, being massive and rotating at high speed, acted like a gyroscope, so that when a sharp nose-up force was applied to evade attack, the gyroscopic precession would snap the nose to the left as well as up. The Red Baron learnt quickly to anticipate this, aim off and wait until the victim flew into his line of fire.
Transgenic cuisine
Reading Stephen Day’s cover story “Invasion of the shapechangers” (28 October), I was very much reminded of Stephano the drunken butler in Shakespeare’s The Tempest, when he stumbled across the monster that spoke from both its ends.
What indeed can be made of a mouse that has the spine of a primordial jawless fish?
Perhaps I could suggest a more user-friendly direction which might put some meat on the bare hones of this line of research: namely, why not try to genetically regress a sheep to taste like a lobster 鈥� a meat that is far older and more expensive than that of mammals?
This would not only improve the health of many a mutton eater and lessen overfishing, but may also ease the burden of Australia’s foreign debt, by exporting this transgenic cuisine to that Brave New World that hath such wonderful creatures in it.
Good vibrations
In our early teens in the mid-1950s, a friend and I used to cadge junked mains radio receivers from repair shops, refurbish them in the cellar at home, and barter them for ex-army two-way radios. While doing the repairs we worked on the exposed circuitry with the sets plugged into a two-pin non-earthed 240-volt mains supply.
As well as learning a lot of practical electronics (such as how not to be electrocuted by 300-volt DC and 240-volt AC) we also noticed a rather uncanny effect. If we stroked a finger lightly along the metal radio chassis, the surface felt as if it were vibrating. Yet we knew it could not really be vibrating because the “vibration” ceased the instant the finger stopped moving.
There was no need even to switch on the radio to feel the effect. Just a single pin of the two-pin plug pushed into the socket was enough. We got the best effect using the softer skin on the back of the hand. Eventually we decided that the tiny 50-hertz current flowing through the capacitance between our body and ground was probably modulating the friction between the finger and the chassis in some way.
A bit later I found a description of a pre-thermionic-valve loudspeaking telephone receiver where the varying direct current of the telephone circuit passed between a flat metal strip and a rotating chalk cylinder impregnated with brine. The strip linked a diaphragm to a spring and rested on top of the cylinder so that the tangential friction tended to stretch the spring and reduce its pull on the diaphragm. Apparently the electric current modulated the frictional force causing the diaphragm to reproduce the speech.
During the intervening forty years I have kept a lookout for any other references to these effects, but in vain. Has anyone else noticed our “vibrating chassis effect”, and does anyone know if it has been used for anything besides the mechanical amplifier in the telephone? It seems too nice to be ignored.
False effect
Susan Blackmore is correct in saying that there are numerous studies on anomalous physical phenomena associated with random number generators which have been interpreted in terms of psychokinesis (Review, 11 November). However, many of these studies ascribe similar properties to randomly seeded pseudo random number generators. This should be sufficient to allow anyone to smell a rat.
Alien lampshades
If the extraterrestrials allegedly visiting us are so smart (This Week, 4 November) why do photographs usually show them driving 1950s lampshades or pie dishes supported by strings?
My hypothesis is that, rather than render themselves invisible, they deliberately disguise themselves as domestic kitchenware etc, in order to heap ridicule on anyone who reports the inevitable accidental sighting.
Letters to the Editor
Write to: Letters to the Editor, New 杏吧原创, King’s Reach Tower, Stamford Street, London SE1 9LS, or fax to 0171 261 6464.
Please include a daytime telephone number, and cite the date of the articles mentioned. We reserve the right to edit longer letters. Your letters may also be published in New 杏吧原创 newsletters.