Editor's pick: Combining bad algorithms will lead to injustice
Many ethical questions need to be answered before facial recognition technologies can be used as part of criminal investigations, as Donna Lu rightly points out (7 September, p 14). She notes that combining inaccurate facial recognition algorithms with the new algorithm to deblur faces leads to the possibility of the wrong person being identified.
This will certainly happen. Combining two inaccurate algorithms dramatically increases the false positive rate 鈥� which, in the case of the best face-recognition algorithms working with clear faces, is over 80 per cent.
People's choices are already being led by AI
You suggest that humans are less likely to accept a choice attributed to an artificial intelligence when it concerns a social problem (24 August, p 38). Many people make decisions that are influenced by their social media interactions, which are governed by AI. Fake news spreads between users on Facebook, who may change their vote because of it . YouTube’s algorithm has been shown to recommend more extreme videos, affecting people’s attitudes (13 July, p 14).
It appears that humans already make decisions suggested 鈥� perhaps covertly 鈥� by AI.
Richard Dawkins, history and lived experience (1)
Richard Dawkins tells Graham Lawton that he doesn’t comment on topics such as fake news and the promotion of gut feelings over facts, or issues relating to the environmental crisis, because none of these are in his area of expertise and his amateur opinion would be no more interesting than anyone else’s (21 September, p 38). I am familiar with Dawkins’s studies in zoology and his eminent career in evolutionary biology and ethnology, but I seem to have missed learning of the period in which he studied theology and religious history.
Richard Dawkins, history and lived experience (2)
Dawkins has usefully promoted scientific truth through his writing on evolution. We need to remember, however, that there is also historical truth, which differs from scientific truth in that history can’t be repeated.
It is important because it helps us to understand the cultures and countries around us and, we hope, to avoid serious future political mistakes and wars.
There is also a third form of truth: that of experience. The best example is love, a vital ingredient in families, which can’t be measured, only given and received. Enough people experience it that it is significant.
A bigger study shows no benefit of organic food
Ann Wills mentions a study that compared consumption of organic food with cancer rates in a sample of nearly 70,000 people (Letters, 14 September). She doesn’t mention a much larger UK study of 625,000 women (British Journal of Cancer, ). This compared those who only ever ate organic food with those who never did and reported on the incidence of 16 different cancers over a nine-year period.
In that time, 50,000 women developed cancer, but there was no statistically significant difference between the groups, except for small increases in breast cancer. There was a lower rate of lymphoma in the organic eaters, although the number of lymphomas was too small to judge statistical significance. As with the breast cancer, this may be due to chance variations. Organic associations promote the idea that their food is healthier, which isn’t scientifically sustainable. Trying to get people to eat their five-a-day by recommending more expensive organic produce is inevitably counterproductive.
ET lacks the time, space and persistence to reach us (1)
There are many suggested solutions to the Fermi paradox (there is a high probability of alien civilisations existing, so where are they?) that Sarah Rugheimer discusses (31 August, p 42). A glaring omission from the list you present is also one of the simplest: space is too big. A journey by a biological organism to another inhabited world would take an implausible amount of time or energy.
Some assume aliens possess technologies beyond our current understanding. These would require at least one gaping hole in our understanding of physics. What if our physics is essentially correct and no such holes exist?
ET lacks the time, space and persistence to reach us (3)
The Drake equation that estimates how many alien civilisations exist that are capable of signalling their presence includes the period in which each actually transmits across the galaxy.
We have existed for several hundred thousand years, but have broadcast in this way for only about 100 years. We are now moving to less noisy laser and cable signals, so that period is unlikely to be much greater than 1000 years.
A circle with a radius of 1000 light years covers less than 0.04 per cent of the area of the disc of our galaxy; 1000 years is a tiny fraction of its 13-billion-year age and it is likely that it will exist for as long again.
We have been looking for these signals for less than 100 years. This is like shining a tiny torch into the Grand Canyon on a pitch-black night. The probability of seeing anything is surely very low, however much has been, is and will be out there.
ET lacks the time, space and persistence to reach us (2)
To answer the question “where is everybody?” we must consider the time period in which a technologically advanced civilisation maintains a broad coalition with the desire to put an immense amount of resources into interstellar travel. I can’t see anything in human history that indicates we are capable of that kind of long-term commitment. If the course of human development to date is a model of how other civilisations develop, then we will never meet any others.
Prospecting for metal with plants goes way back
David Hambling reports that a company is using trees to find gold deep underground in Australia (17 August, p 12). Plants have been used for such purposes at least since modern metal prospecting began. Possibly the best known is Ocimum centraliafricanum, the , which has been used in the Central African Copperbelt since the 1920s to confirm the presence of copper in soil in clearings where the absence of trees is thought to be due to the poisoning of their seeds by copper.
More recently, there have been attempts to recover nickel from the ash of trees that concentrate it (22 March 2014, p 46).
Keep those fossil fuels until we really need them
Tom Chivers discusses solar technologies that could be used instead of fossil fuels, which would help address the increasingly urgent requirement to control global warming (10 August, p 34). There is a second reason to do this.
All understanding of Earth’s long-term cycling tells us that, on a much longer timescale than the current warming phase, a new colder spell within our present ice age will eventually loom.
If humanity is still around then, it may need fossil fuel reserves to burn to counter that threat. Long-term atmospheric management in both directions of temperature will be necessary for a truly long-term future for our species.
Engineering obstacles to electrolysing seawater
Why can’t we use seawater to make hydrogen, asks Albert Lightfoot (Letters, 21 September). We can, but corrosion of the anode by chloride ions in seawater is a problem. It isn’t insurmountable: you could use anode materials like gold or platinum, but they are expensive. Researchers at Stanford University are working on exactly this issue: see .
I wonder whether we could use graphite for the electrodes. I don’t think it would be corroded, but its relatively low conductivity would mean the cells would have to be bigger for a given rate of hydrogen generation.
Lightfoot suggests that there might be useful by-products, but sadly, rare earth metals and cobalt aren’t present in significant amounts in seawater. Lithium only forms 160 parts per billion by weight of seawater and the hydrogen production process wouldn’t help much in its extraction from that water.
Stroppy teenagers are just how they should be
It was good to read your interview with neuroscientist Dean Burnett and his conclusion that teenagers are how they are because it was evolutionarily useful (14 September, p 56).
I’ve been going on about our adventurous species and the drive of the young towards risky behaviour and exciting experiences 鈥� looking to see what is over the next hill and so on. Now, I can quote Burnett as an authority on the subject when my listeners’ eyes glaze over.
How can my circadian rhythm be so precise?
Jessica Hamzelou reports that boosting circadian rhythms can help relieve perinatal depression (21 September, p 15). This prompts me to wonder how such rhythms work. How is it that I always wake at 7.20 am, plus or minus 30 seconds? How can the wetware of the mind be so precise? I get no external cues, visual or aural, and this routine has survived three house moves. It even resets for daylight saving time after about three weeks.
Do take into account the effects of the effects
We read with interest your snippet reporting that sleep loss is worse for young people’s mental health than social media (24 August, p 17). Taking into account sleep, physical activity and cyberbullying, the effect of frequent social media use in causing unhappiness and anxiety was found to be insignificant.
We suggest a story along similar lines. Cancer is worse for you than tobacco: people who smoke tobacco frequently tend to be more prone to breathlessness and lung cancer than those who smoke less. But taking into account tar, nicotine and carbon monoxide, the effect of frequent smoking was found to be insignificant.
A proposition about the use of the word 'theorem'
Leah Crane says that because an estimate by physicist Maximiliano Isi of the mass and spin of a black hole is based on the no-hair theorem, which holds that no information about a black hole beyond its mass, spin and electrical charge is visible beyond its event horizon, this suggests that the theorem is correct (21 September, p 6).
A theorem is a mathematical proposition not self-evident but proved by a chain of reasoning, a truth established by reason and based on accepted axioms.
I suggest that this “theorem” would more accurately be called a conjecture.
The editor writes:
We have to call things what they are universally called and this is the language cosmologists use.