Search This Blog

Follow adrianbowyer on Twitter

My home page

Friday, 31 December 2010


As any gentleman who, being caught short, has nipped behind a hedge and inadvertently peed on an electric fence will attest, the experience is really quite memorable.  Indeed, so memorable is it that it is an almost perfect Pavlovian conditioner - once shocked, never forgotten.

It seems to me that the womenfolk of a household could put this to good use.  Many complain of poor male aim.  A wire ring powered by a small battery-driven inverter could be concealed invisibly under the flush-ring at the top of a lavatory bowl.  As aim drifted absentmindedly to one side, an automatic reminder would be experienced by the victim person performing an act of nitrogenous excretion.  The voltage could be adjusted from mild-tingly-admonition to contravenes-the-Geneva-Convention, ideally by a wireless remote that could be deployed from the other side of a door...

Friday, 24 December 2010


A neuron is (conceptually - if that's the right word) a fairly straightforward device.  Signals come in through its dendrites and are added up.  If the sum exceeds a value, the neuron sends a signal out along its axon, which is connected to the dendrites of lots of other neurons.  Otherwise it stays schtum.

The number of neurons you have (1011) is quite impressive, but nowhere near as impressive as the number of links each neuron makes with others (typically thousands).  Considered as a whole, it is that interconnectivity that makes the human mind the most fabulous object in the Universe.

But it is also annoyingly fragile.  After one has got past the grief when someone dies, a remaining irritation is the realisation that all that they knew has died too.  Who is that person in the photograph? you want to be able to ask them.  What do you think of Jim and Clara's buying that house?

An abiding and very old science fiction response to this mortal fragility is to propose some sort of brain scan that allows the recording of the entire structure and then the use of that data to create its emulation in a large computer.


But let's return to the inherent simplicity of a single neuron.  In a few years time we ought to be able to do the same in silicon.  (We could now, but for the very high connectivity.)  And we'll soon - apparently - have hordes of nanobots swimming round our insides fixing up molecular breakages.  If they also replaced about four million neurons (a tiny fraction) every day with their silicon equivalents, then, by the time someone was eighty, their brain would be entirely electronic.  And the life-long process would be so gradual and smooth that it would become part of the normal acquisition of wisdom with age.

But when someone's body died, you would be able to crack their scull, take out their silicon brain, plug in a five-volt power supply, a webcam, a microphone, one or two robot avatars and a network connection, and they would live forever.

And no one would ever get Alzheimer's disease again...

Thursday, 16 December 2010


Tube stations are plastered with advertising - it helps defray the cost of running the trains, and without it Oysters would need even more money put on them than they do now.

But in the tunnels the windows go black.  This is despite the fact that the passing train is throwing light out onto them.  Why not put ads there too?

The reason is that they would rush past in a blur.  Old-fashioned film would do the same if you just moved it through the gate of its projector.  Each frame had to stop at the gate, and a shutter had to open and close to freeze the frame for a moment on the screen.  Then there was a short time gap before the next frame.  You didn't see this gap because of your eyes' persistence of vision.

By making tube windows from liquid crystals just like the ones used for 3D movie glasses it should be possible to achieve the same effect.  As the train went past a set of identical pictures pasted to the tunnel wall the window liquid-crystal shutter would open and close for each one, maybe with an outward-pointing electronic flash to illuminate it extra brightly.  The opening and the flash could be triggered by a mark on the tunnel wall beside each picture to keep the whole thing in synchronization whatever the speed the train was moving at.

Identical pictures would give a still image.  But of course you could make moving pictures too.

Perhaps there should be ads all down one wall, and an idyllic sunlit country landscape with the trees moving gently in the breeze down the other...

Thursday, 9 December 2010


One of the advantages of being viviparous is that you know who your children are.  Or at least, you do if you are female - if you gave birth to it, then it's yours.  You know what your baby looks like; you know its smell.

This is the reason that brood parasitism is much more prevalent among birds than among mammals.  Brood parasitism is a very efficient reproductive strategy - you can have cared-for offspring without the trouble of caring for them.  But if you are a mammal, it's a lot harder to give live-birth in someone else's burrow and to get away with it than it is for a bird to lay an egg in some other bird's nest.

So why are brood parasites still comparatively rare, even among birds?  There is clearly an evolutionary arms race between the parasite and the parasitised.   To be effective, the parasite's egg must look as much like the host's egg as possible.  To a certain extent, the parasite can take advantage of the superstimulus effect to lay a bigger egg than the host's, so the resulting chick can be stronger and fight its corner better.  But there's a limit to this.  If parasitism were too common, the superstimulus effect itself would be selected against in the host and cease to work.

Bird eyesight is good anyway, so there's a strong selection pressure on the hosts to be able to recognise the individual speckle patterns on their eggs.  It might be thought that the host would also evolve an ability to count, but that is not much use: if you come back to six eggs when you know you laid five, which one do you smash?   Also, giving eggs a characteristic smell is not helpful because it nullifies one of the principal advantages of being oviparous in the first place: predators can't smell your young.

However, all this assumes that there are two species competing.  But if you are - say - a blackbird, it must also make sense for you to lay an egg in another blackbird's nest.  You gain the same advantage for your chick as the cuckoo and the cow bird normally have over you, plus your deception is a lot harder to spot.

This puts a female bird in a similar (but not identical) position to a male mammal: a man (before DNA testing) was never certain that the child he was rearing was his, though a woman was.   But the man can father other children that he does not need to care for.  By using conspecific brood parasitism both a male and a female bird can have chicks that they don't need to care for, but neither is certain that the chick that they are rearing is their own.

In mammals the Y chromosome is asymmetrically passed exclusively down the male line.  Birds have the opposite system: their W chromosome is asymmetrically passed exclusively down the female line.  It would be interesting to look for similarities in the behavioural consequences of the possession of these two chromosomes given that egg-laying in the wrong nest puts female birds in a similar evolutionary position to a man fathering children by a woman other than his partner.

Indeed, maybe the Y and W chromosomal asymmetry itself is - in part - an evolutionary consequence of this similarity.

Thursday, 2 December 2010


When you phone virtually any large company or government department the phone will be answered by a robot.  The robot will tell you to press "1" for fifteen minutes of Vivaldi, or "2" for twenty minutes of Gerry Rafferty playing Baker Street.  From time to time the music will fade out, giving you the false hope that a real person will come on the line and actually deal with your query.  But of course it's just the end of the track - Vivaldi is replaced by Rafferty, or vice versa.  And so on.  For ever and ever.  World without end.

The reason that this makes you want to go and dynamite the head office of the organization concerned (surely no jury would convict?) is that they have imperiously decided that their time is much more valuable than yours, and that they have then implemented a Kafkaesque system that you cannot escape to impose that decision upon you.

I once recorded an endless loop of Greensleeves on my pocket dictaphone.   I faded the music down every thirty seconds or so and interspersed a message saying, "Your answer is important to us.  Please hold."  I would set this playing by the phone when the robot came on, so that the organisation's minion, when he or she finally did answer, would wait till I could be bothered to return.  I would then get on with something more important.

But what is really needed is some legal punitive measure.  How about a simple law that says, when you call somewhere and a non-human answers, you can press the * key and the call charges are reversed crediting money to your phone account until a human comes on the line?

This would at least redress the balance - you would be being paid by the organization for your time that they are wasting.

It might even become possible to make a comfortable income by calling up the worst offenders, pressing *, and then (some time later) wishing the person who finally answered a good day and hanging up...

Thursday, 25 November 2010


So you have some photovoltaic cells to generate electricity, and you have a bottle to store hydrogen.  You can now make the hydrogen by electrolysis, and so save the solar energy for use after dark, or in a different location.

But you need water to electrolyse.  In most places that isn't a problem.  But in deserts, or if you need to make a closed fully-independent system, what would be useful would be an automatic way of extracting the water you need from the atmosphere.

As anyone who has ever inadvertently left a few crystals of sodium hydroxide on a lab bench knows, deliquescence is an intriguing phenomenon.  Many soluble substances (like NaOH) have such a strong affinity for water that they will drag it out of the air and dissolve themselves in it, forming - in the case of the crystals - little droplets on the bench that it is unwise accidentally to lean on.

It ought to be possible to make a deliquescent solution that is also electrolytically appropriate for the production of hydrogen.  Along with the PV cells, this would take sunlight and make a stream of hydrogen from it continuously, extracting water from the atmosphere to replenish itself.  Having no moving parts, it should work forever, with no need for topping up and no need for maintenance.

Mass-produced generators would become possible that could simply be left in sunny places to produce hydrogen for ever...

Friday, 19 November 2010


Hybrid cars are a sort of half-way house between old-fashioned twentieth-century petrol cars, and proper electric cars with batteries.

We can't quite have the latter, of course, because the batteries aren't up to it yet (see my post on electric aeroplanes of a few weeks ago).

All the hybrid cars I have seen, though, use a conventional reciprocating engine to drive their alternators.   This is very silly - reciprocating engines are inefficient and unreliable.  They have one single advantage: they respond rapidly when you blip the throttle.  You need that if they are directly connected to the power-train, as in the old-fashioned cars I just mentioned.  But that advantage is completely pointless in a hybrid, in which the battery accommodates sudden changes in load.  For a hybrid you need a simple reliable engine that runs at constant speed and more-or-less constant power.

The picture is of a Garrett GTP 30-67 gas turbine.  It's about half a meter long and generates about 25 Kw of power.  That sort of engine would be perfect for a hybrid car.

Go build one, World...

Wednesday, 10 November 2010


It's not work that creates wealth.  It's wealth that creates work.

In buoyant economic times near-full employment comes about as a consequence of people flush with cash going out and spending it: someone has to make the jet-ski that they buy or the resort hotel in which they book a room.  And recessions are created by people hunkering down, sensibly hanging on to their money now that that rainy day has come, and thereby throwing the other people who would have been servicing their needs out of work.

Now, apart from buying lunch and a couple of commuter tickets, people actually at work are not spending a lot of the money that their work is earning for them.  The company for whom they are working may be spending as a result of their activities, but the amount available for the company to spend is comparatively low.  The total assets of the people who work for a company usually far exceed the assets of the company itself.

Almost all the money those people spend is spent during their free time.

Yet we have the idea that hard work and long hours are what makes for a strong economy.  Nothing could be further from the truth.  Hard work and short hours are what makes for a strong economy.  People with more leisure time spend more money on that leisure.  Further, as they are working fewer hours, there is more employment available to supply the goods and services that they are now buying.

As a consequence of all this I offer a five-word election manifesto to any political party that simultaneously wants to get elected and to improve its nation's economy.  It has such instant appeal to the electorate that it would guarantee an election win, regardless of how dismal the party's previous poll ratings were.

It is this: Add Friday to the Weekend.

Friday, 5 November 2010


Artificial intelligence researchers are very fond of the idea of a Turing test: a computer program converses with a person who can't see their fellow interlocutor.  If the person cannot tell that they are talking to a computer, as opposed to another person, then the program must be intelligent.  (Or at least as intelligent as the person's friends...)

Turing's brilliant proposal cuts the Gordian knot of defining what is or is not intelligent behaviour with subtle elegance:  if people can't tell that they are not talking to other people through the wall when they actually talk with the computer, then the computer must be at least as intelligent as people expect people to be.

Of course, no one has written a program that will pass the test yet.  But programmers often simplify things by restricting conversation to a narrow area, such as fashion, or physics.

Others try to write programs that do non-conversational intelligent things that people do: the world chess champion is now a computer program, and poker-bots attempt to fleece real people on online poker sites.

But how about a program to pass human intelligence tests?  These tests are highly structured, and so it should be relatively easy to write a program to interpret the input.  And they usually require a rather restricted range of pattern-matching skills to solve.  Finally, the score can be directly compared with that of the large groups of people who also take the tests.

It might be rather salutary if computer programs were to turn out to be both good at IQ tests and quite simple to write...

Friday, 29 October 2010


The two most immediate dangers the Earth faces are anthropogenic climate change and the ongoing current mass-extinction event, also anthropogenic.  There are bigger dangers, of course, but those two are the most close at hand.  They are also two that we can do something about.

For climate change, how should we set about reducing atmospheric CO2?  It seems to me that if we are going to save the planet, we should harness the single most powerful force that we know and set it to tackling the problem.

The single most powerful force that we know is human greed and stupidity.

World yearly plastic production is currently about 300 Mt.  That plastic contains about 250 Mt of carbon.  World CO2 production is about 30,000 Mt.  That CO2 contains about 8,800 Mt of carbon (oxygen is a lot heavier than hydrogen).  So the world's consumption of plastics uses up about 3% of the carbon that we emit into the air - a useful dent.  Except it doesn't use it up, of course, because most of those plastics are made from oil that came from underground.

But there are plenty of plant-sourced plastics available.  They do use up atmospheric carbon - that's how plants grow.  The trouble is that the plastics industry tries as hard as it can to make those bio-sourced plastics also biodegradable, to enhance their green credentials.  But every plant-sourced plastic item that biodegrades returns the carbon that was locked up in it to the carbon cycle.

What we should be doing is to make non-bio-degradable plastics from plants, and then to encourage people to consume them and waste them, throwing them into landfill.  The economic churn would throw positive feedback into the mix, people would be getting more and more of the consumer goods that they crave, and the carbon would be being locked up in the ground far more securely than with any current carbon-capture scheme.

Greed is good, as Gordon Gekko said.  Consume, and don't recycle.  Throw stuff away to save the world.

Thursday, 21 October 2010


Last week I was named as the 103rd most influential person in British science by the Times' Eureka science magazine.

So - only 102 people to bump off, then...

I am, of course, very grateful to the Times for considering me, and for the party that they organised in the Science Museum in South Kensington for us all.

In addition to lots of my fellow scientists and engineers, there were quite a few people from the Times at the party, and the conversation turned to News International's paywall.  Behind this their online newspaper content now resides.  There was much rather anxious discussion about whether or not it was going to work.  I said that it ought to be straightforward to predict if it will or not using evolutionary game theory, and on the train on the way back home, I worked it out.

Here's the result.  The Y axis is the income that a news site gets, which is made up from advertising and - for non-free sites - the paywall subscriptions.  The X axis is the proportion from 0 to 1 of news sites that are behind a paywall.  The blue curve is the income a paywall site gets, and the red curve is the income a free site gets.

The actual Y-axis numbers are arbitrary.  What's interesting is the shape of the curves.  As more and more sites go behind paywalls the income per site drops because the paying customers have more choice, and there is less advertising income per site.  The free sites' incomes rise, because all the remaining newsreaders go to them, and so - consequently - does most of the advertising.

With the numbers I used the evolutionarily-stable proportion of paywall sites is about 0.2 of all news sites, with 0.8 of them remaining free.  If you change the assumptions the curves change and so does the crossing point, of course.  But the principles remain the same.  If the number of paywall sites is to the right of the crossing point, you make more money if you run a free site; if it is to the left you make more money with a paywall.  That's why the crossing point is evolutionarily stable.

This graph must be the reason that News International broke ranks and went paywall first.  That means that they are operating at the left-hand end of the graph on the blue curve.  If you are a big news organization it makes sense to occupy as much of that left-hand end as possible.  Indeed, if the total of all your own paywall sites just makes your proportion of all the news sites in the world the crossing point, then you get all the paywall income that the world has to offer, and your rivals will do better to stay free...

Tuesday, 12 October 2010


We all want sustainable energy these days, what with fossil-fuel-induced climate change and the need for nations not to have to rely on dodgy foreigners to sell them the gigajoules they need.

Why don't we use the power of the wind?  I don't mean the pathetic flow of air over our own planet that we already capture with giant versions of children's toy windmills.  I mean an altogether more serious wind:

And I also don't mean capturing its kinetic energy.  That would be far too difficult, and it's not the main source of energy in the solar wind anyway.  The main source of energy in the solar wind is helium-3.   This isotope of helium is particularly suited for use in fusion reactors.  Fusion reactors are attractive because they are potentially very powerful, and because anyone can make one for a few hundred dollars.  Admittedly we haven't yet made one that generates more energy out than you have to feed in, but an abundant source of helium-3 and the combined genius of lots of garden-shed tinkerers might well change all that.

There is almost no helium-3 on Earth.  It has been proposed to mine the Moon for it, as the solar wind embeds it in the lunar regolith.  But the problem with that is that the Moon is a gravity hole that you have to expend energy getting down to, and then expend more getting up from.

Why not put a solar-powered satellite that is, essentially, a big electromagnet at one of the Lagrange points?  The solar wind is ionised, and so it would stream to the satellite's poles, like a mini aurora.

Of course, helium-3 is only a minor component of the solar wind, so it needs to be separated.  But here things start playing to our advantage.  The satellite would use the same technique that mass spectrometers use to produce a clean stream: when you pass an ionised plasma through a magnetic field, the ions deflect through different angles depending on their mass.  All you have to do is to put a collector at the correct angle to the magnetic field for an atomic mass of 3, and the helium isotope you want just falls into it.

If we got everything right, we could provide thousands of amateur fusion researchers with a fuel with real potential, and leave them to collaborate and to compete in the usual crowd-source way to come up with the best design.

And, when they give us our first energy-positive fusion reactor, we would have a non-polluting power source for the entire world that, as it would have been developed on-line by a collective, would be free of patent encumbrance.

Finally consider this: people working on fusion power are, of course, working on building clean power stations.  But — in some cases without realising it — they are also building the human race's first starship engine...

Thursday, 7 October 2010


The number of people with autism-spectrum disorders is increasing faster than the number of those with any other single neuropsychiatric condition.  Without doubt, this is partly discovery through better diagnosis techniques.  But there is significant real growth too.

People with autism are more than twice as likely than people without to have fathers and grandfathers who are engineers.   Further, it is known that autism is strongly heritable.

The Industrial Revolution made engineering one of the most secure of human professions.  Engineers only occasionally get rich and famous, of course, but they almost always live comfortable lives.  They are often among the last to find themselves unemployed when times are hard, and economic pick-up sees them in the vanguard of those back to work.

And a comfortable life is an optimal basis for the raising of children.  Perhaps, since the Industrial Revolution, simple Darwinian selection has acted to increase the proportion of descendants of engineers in the population.

Perhaps further, a side-effect of this increase is the increase in autism-spectrum disorders that we are now seeing.

Thursday, 30 September 2010


Ice is flat. Or rather, to be pedantic, ice forms a surface parallel to the geoid, an approximate sphere roughly 13,000 kilometers in diameter.  This is true of both ice on rivers and lakes, and also ice in ice rinks.

Making an ice rink is fairly straightforward.  You construct a shallow dish and lay plastic pipes across the bottom.  You fill the dish a few centimeters deep with water, then pump refrigerant through the pipes.

Skateboard parks, in contrast, are anything but flat.  This allows skateboarders the freedom to play with the interchange between their kinetic and potential energy, with - in competent hands (or rather, feet) - spectacular results and  broken bones.

Skateboard parks are made by pouring concrete into formers and letting it set.

But you could do just the same thing with ice.  The formers could contain the refrigerant pipes.  You would fill them with water, not concrete, then turn on the refrigerator.

The result would be like a skateboard park, but for people wearing ice skates.

Thursday, 23 September 2010


Pharmaceutical companies spend a lot of their time, and ill people's money, patenting medicines.  They then have a legal monopoly on those medicines, which allows them to charge a great deal for them.  For example the anti-cancer drug trastuzumab (Herceptin) costs about €30,000 per year for one breast cancer patient.

Trastuzumab is a monoclonal antibody, so it is not as easy to make as making carbon dioxide by adding sulphuric acid to chalk, but it isn't that hard either.  You fuse cancer cells and spleen cells in a mouse that has been injected with whatever substance that you want the monoclonal antibodies to bind to.  In the case of trastuzumab, that's human epidermal growth factor receptor 2, which is a protein that makes breast cancers grow.  (In that particular case, you also can't use ordinary mouse cells as people react to mouse proteins, you have to use cells that make the human version.) Finally, you clone the cells to increase their number then use them to make the antibody.

These days all this sort of messing about is done by robots with pipettes, not by the pharmaceutical company employee above with the conical flask.

How much would such a robot cost? I build robots all the time, and I estimate that such a robot might - if mass produced - cost a couple of thousand Euros at most.

Here patent law (in Europe, not America) gets interesting: any private individual can make any patented object for their own use without infringing the patent.  They can't sell it, but they can, for example, eat it or inject it.

This means that national health services could loan patients a drug-synthesizing robot that cost less than a few-months supply of some drugs.  The patients could make their own drugs and thereby treat their disease with the latest medicines.

The health services would save a fortune.

Thursday, 16 September 2010


One standard way to make a fractal curve is to take a straight line, cut it in half, and move its mid-point a random amount. You then do the same with the two halves, but move their midpoints by a smaller random amount. If you keep going you end up after a few times with something like the picture above.

This is a way to generate artificial coastlines. If you do it with a triangle instead of a line and move the midpoints of its edges out of the plane of the triangle, you make an artificial fractal landscape (shown here flooded):

Here's a new way to do this sort of thing.

Take a box (blue) with a single diagonal line in it (black). Split the line in half and move it as far as you can without a disc on the split-point hitting anything. You get this:

Then repeat this on the two halves, with the exclusion disc reducing in radius at each stage and being imagined to be on every vertex of the curve:

That's it after six steps. Note that, although it looks random, it is entirely deterministic and its shape is set by that simple rule: split, and move as far as you can without the straight lines overlapping the discs.

Here it is after splitting ten times.

If you start with the four edges of a square (filled in red here so you can see what happens):

And let it run you get this:

For more on this idea see here.

Thursday, 9 September 2010


We sweat to dump excess body heat.

Sweating is an elegantly efficient mechanism that takes advantage of the fact that our most readily available liquid - water - carries an exceptionally large quantity of energy away when it evaporates. The enthalpy of vaporization of water is 2257 kilojoules per kilogram. This means that if you are exercising at 500 watts for half an hour (which is a good aerobic workout) you only need to evaporate 400 grams of sweat to carry away all the heat you generate. That's one large glass of water.

In practice, it works even better than that because you also lose heat by radiation and convection as well. But just because something marvelous has evolved, it doesn't mean that we can't improve on it.

As anyone caught in a rainstorm in light clothing knows, a wet shirt is very cold indeed, especially when the wind blows. It is sweating for you when you don't need it to.

We also deliberately get our clothes wet once each time we wear them: we wash them afterwards.

Poly-(N-IsoPropyl-AcrylAmide) is hydrophobic above 37oC and hydrophilic below 32oC. This means that at low temperatures it holds onto water, and at high temperatures it expels it. There are other materials with similar properties.

Suppose we were to make clothing incorporating these materials? If we were to wash them on a warm wash, they would absorb some of the final rinse water and then retain it when cold, feeling dry to the touch. But when we exercised in them, or simply found ourselves in the warm sun, they would start to give up their stored water.

Our clothing would keep us warm and dry when it was cold, then sweat for us when it was hot.

Thursday, 2 September 2010


Most electric aeroplanes tend to look like this. Make no mistake, the achievements of electric aero researchers are impressive. But the results always tend to look a bit fiddly and - both more importantly and more understandably - lightweight.

That example (Helios) is solar powered. Sunlight has a density of about 1 Kw per square meter, and conversion efficiencies don't get much above 30%. This means that you don't have a lot of power to spare, given a reasonable wing area. That is why the plane looks like it does.

The obvious solution - replace the sun with batteries - stumbles because batteries don't give a lot of power either. Worse, they are heavy.

Suppose we want to carry 300 people like this:

The source of power here (oil) has a specific energy of about 45 megajoules per kilogram. But the battery in your car only has about 0.15 megajoules per kilogram.

However, the battery in your car is 100-year-old technology. The specific energy of batteries has been increasing, and increasing more rapidly in recent years. The best we have today (lithium ion nanowire) gives about 2.5 megajoules per kilogram.

And batteries don't have to get to the energy density of oil to compete in aircraft propulsion. This is because electric motors are much more efficient than gas turbines. Typically, you can't get more than about 40% out of a gas turbine, but you can easily get 80% from an electric motor.

So when batteries get to ten times the energy density of the best now, it will start to make sense to power aircraft with them. And with the rate of progress in battery development that will probably take less time than most would imagine.

But there is a research gap. As far as I can find out, no one is trying to develop an electric aircraft turbine for passenger aircraft. Given the lead times for this sort of technology, this seems short-sighted to me. We are in danger of having the batteries for planes, and then having to take ten years to make engines to go with them. Engine development doesn't need to have the batteries available, as all the testing can be done on the ground.

A final point. At half the energy consumption of an equivalent gas turbine, an electric turbine will generate half the heat. This will have to be got rid of in the airflow, where it can be put to good use expanding the air and generating extra thrust.

Thursday, 26 August 2010


Having a picture framed is an expensive business - if the frame is going round a print it can cost three or four times as much as the print itself. And even if the frame is for an original painting it can cost a significant fraction of what the painting cost (unless you are the sort of person who needs to re-frame a Caravaggio).

It's also a difficult job to do yourself. Even the most careful amateur carpenter can have trouble mitering the corner of the moulding to get a perfect 45o cut.

My father was a painter, and he would frame his own pictures for sale. The machine on the right used to live in our kitchen when I was a child. It is a Morsø mitering machine, and it makes the cutting of a perfect 45o miter easy; anyone can get a good result with one of these. It guillotines the moulding to form two mitres simultaneously.

But it obviously makes no sense for everyone to have such a machine for themselves for use on the one occasion every two years when they want to frame a picture.

There are many products that are designed to break once cleanly. Think of the perforations around postage stamps, or of the ring pull on a Coke can. Why not make picture frame mouldings that have a series of 45o indentations along the back every few millimeters designed to snap cleanly? The indentations would have to be symmetrical about the middle of a length of moulding with -45o to the left and +45o to the right. You could then buy two lengths a little too long for the width of your picture, and another two a little too long for the height, and make your own frame of any size you liked.

It ought to be possible to do the design with added clips that would fit into the remaining indentations to hold the corners together, so you wouldn't even have to use a hammer and nails.

Wednesday, 18 August 2010


Deaths of passengers from airline terrorism are, of course, fabulously rare. The worst decade was the 1980s, when about 150 passengers per year were being killed. The figure has fallen steadily since then, and is now around the same level as it was in the 1960s - about 40 deaths per year.

Today there are many more flights than there were in the 1960s, so the relative figures are even better than those numbers imply. The reduction in deaths is due to a combination of factors, the two principal ones being better airline security and the general fact that terrorism - airline and otherwise - is much rarer now than it was.

The world's airlines fly about two billion passengers each year. If each passenger spends an extra three-quarters of an hour getting through airport security over what they would have spent in the 1960s, then 2,300 human lifetimes are taken up in airport security per year.

So airport security takes about sixty times more human lives than the threat that it protects against...

Thursday, 12 August 2010


Loctite Blue glue is remarkable stuff. In the presence of oxygen it is a liquid, and it only sets hard when oxygen is excluded. This means that when you put it on a bolt thread it is easy to screw the bolt onto its nut. But once the bolt is tight (which excludes the oxygen) the glue locks up, preventing the thread coming undone.

Even cleverer, the reaction is reversible. Thus, when you loosen the thread a bit with a spanner, the ingress of the air releases the bond, and the bolt can easily be undone the rest of the way. Rather neatly, the bottle in which the glue is sold is permeable to oxygen, so the glue stays easy to pour.

There are often photochemical reactions that are equivalent to this type of reversible bonding. So how about a glue that sets solid in the dark and is liquid in the light?

You would paint it on a surface, where it would stay happily liquid (well, it would till dusk...). But when you put down the other surface to be stuck on top of it, the glue would instantly set solid as the shadow fell, holding both fast.

If one of the two items to be stuck were translucent, but covered (except for the sticking surface) in an opaque paint, the glue would still work. But you could release it simply by opening a window in the opaque paint. It would be possible to set up all sorts of complicated simultaneous sticking and release mechanisms that were worked by allowing light in and excluding it.

It might even be possible to have the material respond to different wavelengths if the no-stick chemistry were only triggered by photons of a specific energy. Different coloured lights could then be used to release different bonds.

You could also have an electric release mechanism: simply bury LEDs in one of the surfaces...

Thursday, 5 August 2010


Tobacco, notoriously, is not very good for you. Consequently there is an ongoing battle between the people who sell it and - more-or-less - the rest of the human race. The former want to push more cigarettes; the latter want to stay alive. The battle is quite instructive in a game-theoretic way. In particular, the sellers carefully manipulate the level of the addictive component (nicotine) to maximize the number of cigarettes that they sell.

When there is a large number of smokers, it makes sense for the tobacco companies to reduce the nicotine level; that way they sell more (and so make more money) as people have to buy more to get the level of nicotine in their blood up. But as the number of people smoking drops, the companies raise the nicotine level in order to create more addicts among people (particularly teenagers) who only try two or three cigarettes. The reduction in the number of smokers in recent years (at least in the developed world) is the reason that the companies have been cynically upping the dose. (See, for example, this article in The Washington Post.)

Those on the other side who are concerned with keeping us all alive deprecate this, and advocate controls on the maximum nicotine levels allowed. Jack Henningfield and Neal Benowitz in a British Medical Journal editorial, say: "Possible strategies to be overseen by the Food and Drug Administration could include ... restrictions on the amount of nicotine in tobacco products..."

A single cigarette contains about 10 mg of nicotine. It is one of the more poisonous substances known - about 60 mg will kill a non-smoker and about double that will kill a smoker. It is a particularly effective insecticide, which is why tobacco plants have evolved a metabolic pathway to make it, of course. But nicotine is one of the less harmful ingredients in tobacco. The things in there that really kill people are the tars, benzene, formaldehyde, and so on in the smoke, and the carbon monoxide that is consequent on the combustion that creates it.

So it may well be that restricting nicotine levels in cigarettes is the wrong way to go. The authorities should require them to have a minimum nicotine level. That level would be set so high that non-smokers starting would immediately throw up, thus putting them off, and so that smokers would only need to smoke one cigarette a day to get their dose, thus much reducing their exposure to all the substances that really harm them.

Thursday, 29 July 2010


Glass is a wonderful material - you can see through it, but it lets nothing material pass.

It does let heat through, of course. But with coatings and double glazing that can be minimised.

However, that very see-through-ness is also a shortcoming; sometimes you want your windows not to be transparent. That's why you have curtains or blinds. Othertimes you want to be able to see out, but to deny people the ability to see in. That's why you have net curtains.

There have been schemes to make glass windows that are liquid crystals, so that you can switch them electrically between transparent and opaque. But they are horrendously expensive.

What's needed is a cheap simple mechanism to allow a window to be switched from transparent to shaded to opaque and back again that requires minimal electricity (such as that that could be generated by putting PV cells around the frame).

Why not pump coloured liquids between the inside two of the three panes of a tripple-glazed window? You could have any colour you want. You could use milky liquids to achieve opacity. You could even have fancy effects like a liquid with a suspension of neutral-density glitter to scatter the incoming sunlight all round a room rather than just dropping a quadrilateral of it on the floor.

The pump would only have to run when a change was needed, so the electrical requirements would be minimal. It would obviously be essential for the glass to drain dry when the liquid was pumped out to achieve total transparency (or before a colour change; you probably wouldn't want the alternative liquids to mix over time). To get that, you just give the glass a highly hydrophobic coating, like that achieved by lotus paint.

In a big deserted office block at night, you could even make a coarse-pixeled display of a whole glass curtain wall just by changing the colours and turning the room lights on and off under computer control...

Thursday, 22 July 2010


Stylometry is the battery of techniques used to compare the authorship of documents. Stylometrists do things like counting the frequency of the uses of "and" and "the". Similarity in a whole group of such measures and related statistics can say with reasonable certainty whether two documents were written by the same person.

Stylometry is usually carried out for academic or forensic purposes: literary researchers want to identify the authorship of interesting anonymous works; the police want to identify the authorship of the ransom note.

But the web is a vast collection of samples of writing. It ought to be possible to run a crawler over it that logs the stylometric measures of every piece. A search engine could then be constructed that allowed anyone to submit some text for analysis. The engine would report back the most similar pages on the web in terms of writing style.

Would this be intrusive or useful? Probably both - people could check whether the reference they have just been given was actually written by the referee, or by the job candidate. Ghost writers of celebrity (auto)biographies would be exposed. It would just be interesting to know that Ian McEwan's style is closer to Margaret Atwood's than to Martin Amis's. And you could take any piece of writing from anywhere and find out who in the world is most likely to have written it.

Thursday, 15 July 2010


A curious aspect of human psychology is that it seems impossible to design a machine voice that is acceptable. There is the dimension with hectoring at one end and unctuous at the other. Patronising is a good way along that, and no point on it seems right. And in an orthogonal dimension there's Stephen Hawking's deliberately retro Dalek at the left and the uncanny-valley at the right. That latter voice is almost too perfect, and yet - or and so - you still know that it is a robot.

Of course, anything with a voice immediately passes the gut-reaction Turing test: if it talks to you, it must also understand you when you swear at it. And you swear at it with more gusto and elaboration than you do at the mere screwdriver with which you have just stabbed your thumb. Sure, the screwdriver is out to get you too, just like every other inanimate object in the universe, but it doesn't simultaneously assault you and adopt airs above its station.

In a forlorn attempt to escape from this, some GPS manufacturers offer the option of celebrity voices, allowing you to have Arnold Schwarzenegger (or Stephen Hawking) telling you to turn left in three hundred meters.

Let's accept that that problem is insoluble and go - so to speak - down a different road: why not use the voice to convey more information? The GPS knows (we hope) where it is. So it could speak to you in the appropriate regional accent.

In Bristol it would talk to you in completely different tones to those it would use in Bermondsey. What's more, with blending techniques similar to those used to morph a computer graphic of one face into another, the accent could change smoothly from west-country to cockney as you drove East up the M4. This would be interesting, would help to preserve those accents against the onslaught of electronic cultural uniformity, and would (in Glasgow) be incomprehensible to all but people who knew their route anyway. You could even have a freeze button, so that, when you hit an accent you particularly liked, you could lock the machine to speak that way until you decided to release it from your spell.

Thursday, 8 July 2010


This is a zero-emission car that runs on compressed air. The air engine is made by a French company called MDI, and the car's claimed top speed and range are about 100 kph and 250 km respectively. An obvious advantage of this over an electric car with batteries is that it's very easy to re-charge - you just plug in a compressed air line for a few seconds, and then off you go.

I am old enough to remember steam railway locomotives from when I was a child. Around the age of eight I saw a very curious one like this:

Smug and thermodynamically-aware, even at eight, I realised that - if it had no funnel - it couldn't have a firebox. And - if it had no firebox - it had no source of power. Yet it was happily shunting wagons up and down the tracks with lots of hissing and steam.

My father explained that the locomotive was carefully insulated, and that all they did was to fill it with very hot water under very high pressure from a stationary boiler beside the track in the morning. Then it would shunt all day as its store of water boiled off to steam.

This has potential for cars too. With modern materials like aerogels we should be able to make an almost perfectly-insulated pressure tank.

The compressed air for the compressed-air car comes from an electric compressor. Here all that would be needed would be an electric kettle element embedded in the car's insulated tank. You would fill the tank with water in the evening when you got home, plug the car in, and switch it on. The water would be heated to a few hundred degrees under high pressure overnight, and the car would be ready to go in the morning.

And the simplicity of re-charging that the air car achieves would be retained by the stored-steam car too. At the filling station you'd just refill the car from the filling-station's boiler.

Friday, 2 July 2010


Democracies and dictatorships share a shortcoming: those in power want to be there. Since before Plato we have known that the most important disqualification from government should be a desire to govern, and there have been many proposals over the years for rule by juries of conscripts. My personal favourite would be to retain the idea of elected representatives from constituencies, but to deny those representatives the right to vote. They would be able to propose and to debate, rather like courtroom lawyers, but the conscripted jury (who would be allowed to do neither) would get to make the decisions.

But no such scheme will be implemented, of course, because both the politicians currently in power and their opponents who want to be in power next year have no interest in volunteering to be usurped by a jury of Ordinary People.

So how could we move a little way in the direction of government by jury?

One difficulty prevalent in democracies is a byproduct from the need to establish consensus to get anywhere. This good thing leads to a bad thing: the social cohesion required of the ruling group and the desire of its members for preferment cause those members to vote with the group even when they know it's doing something silly (and these are politicians remember, so silliness is pretty inevitable).

Secret votes in parliaments and senates would eliminate that difficulty completely. But that would lead, in turn, to a democratic problem: constituents need to know how their representatives voted in time for the next election, so they can decide if those representatives kept their promises and acted well.

Fortunately the inevitable bureaucracy of general-election nominations can be used to resolve this problem quite simply. Election candidates have to put in candidature forms before a deadline that is some weeks ahead of an election. Therefore we set up a scheme whereby sitting parliamentarians and congresspersons vote secretly in their parliament, and their votes are recorded. Then, the day after the deadline for the election-candidature forms, the voting records of all sitting members are released.

That way people in parliaments can vote according to how they really think, and the electorate also gets to know exactly how they voted before deciding whether or not to re-elect them.

Thursday, 24 June 2010


The Moon is all very well, and is quite pretty in its way.

But how about an installation consisting of a set of rings round it?

We'd need material for the rings, and we'd probably need to put some shepherd satellites in orbit to keep the rings sharp. This implies moving a few asteroids about, a skill that it may well be vital to have developed anyway if a rock looks to be heading our way.

The shepherd satellites would also be needed to keep the rings tilted relative to the plane of the ecliptic so they weren't edge-on. That way we'd get the full impressive effect.

We could rent out projection advertising space on them on a strictly one-day-a-month basis, that day being a new-moon day when they'd probably be at their least naturally spectacular. That should pay the bill, and leave 27 days each month for us to enjoy them...

Thursday, 17 June 2010


SETI - the Search for Extra Terrestrial Intelligence - is rather like religion or catwalk fashion: all of them seem simultaneously both profound and profoundly silly.

The SETI crowd concentrate on scanning the sky for alien radio signals. To a certain extent this is a case of looking for your lost contact lens under the street-lamp, rather than in the shadows where you dropped it. This is because radio (and visible) electromagnetic radiation is what makes it all the way down through the atmosphere to us. What's more, the Earth radiates in the radio spectrum like a small star. Everybeing with a receiver in a sphere 200 light-years in diameter (and that's about 60,000 stars) will know that we are here, and will be uncomprehendingly trying to deduce the relationship between The Sopranos and Bugs Bunny.

If we were looking at us using radio, we would see us. But we would not be seen because we are trying to communicate beyond the Earth, but because of leakage from our inefficient broadcasting system. We can imagine that, with technical advances such as the one on which you are reading this, in but a few decades our skies will become silent again even as our communications become ever more feverish.

We have also sent a few deliberate transmissions, with clever clues as to how to decode them into pictures (the lengths of the sides were prime numbers). Of course, any intelligent listeners may have no concept of a picture. But - much more importantly - they will have to be listening in the right direction at the right time. Radio signals flash past at the speed of light. Blink, and you've missed them.

For a moment consider the matter from the perspective of the Things doing the transmitting. Like our prime-product-picture senders, they have something to say. What characteristics would their ideal transmission medium have? I contend that they are these:

  1. The medium should selectively seek out places where life, and hence receiving intelligence, is most likely;
  2. It should travel at the speed of light.
  3. It should last a very long time when it arrives, avoiding the blink-and-miss problem.

In short, and in twenty-first century human terms, what a transmitting civilization needs to do is to drop - at the speed of light - a DVD of Wikipedia and a few Bach partitas down on every habitable planet; a DVD that will remain playable for hundreds of millions of years. The trouble is, of course, that it requires infinite energy to accelerate a DVD to the speed of light, and they don't last more than a few decades, let alone hundreds of millions of years.

But what about a bacterial spore with the message encoded in its junk DNA?

Again, it needs infinite energy to get it to the speed of light. But - with a mass of 10-15 Kg - it only requires 7x10-15 joules to get it to 99% of the speed of light. And that will convey it to its destination almost as quickly as a radio signal. It would have to be coated in something to protect it - probably a microscopic metal labyrinth from which it could escape, but which shielded it from every direction. This is because, when you move at 99% of the speed of light, ordinary starshine pours down your throat as hard gamma rays; it is Doppler shifted. The protection would raise the mass a bit, and hence the launch energy. And, of course, most spores would never make it.

However, those that found a convivial home would start to reproduce, spreading the message and copying it down the aeons. The message would become a little corrupted with time, and the bacteria would change through evolution. But by comparing the DNA from many of its descendants it ought to be possible to reconstruct the message.

Instead of looking to the skies, we should be examining junk bacterial DNA for the first hundred digits of the base-4 expansion of pi.

It's possible, given long enough, that the message itself might evolve into a species capable of understanding it. So maybe we should be looking at our own junk DNA too...

Thursday, 10 June 2010


Holograms are extraordinary things. Just by using constructive and destructive interference, a two-dimensional photographic plate recreates the entire pattern of light in space bouncing off a three-dimensional object.

The diagram of the way they are made is a staple of every physics text book:

(That one is courtesy of Wikipedia.)

A hologram is a rather complicated diffraction grating. Imagine leaving the three-dimensional object out, changing the angles in the diagram a bit, and just recording the interference between the raw illumination and reference beams. When you do the interference sums (which for this case require no more than school trigonometry) you find that that interference pattern forms a - this time simple - diffraction grating. If you re-shine the reference beam at that grating, trigonometry again shows that the angle of the diffracted light that it gives off is just right to recreate the original unhindered illumination beam.

The hologram of the three dimensional object is just this process going on all over the place to recreate the much more complicated light pattern bouncing off the object.

For someone used to the Roman alphabet like me, walking round an Oriental city can be a confusing experience. You have to work out every sign by consciously and deliberately comparing bits of geometry. And heaven help you when the font changes. The same must also be the case for someone used to reading, say, kanji hiragana and katakana, when they find themselves in Europe or America.

What you would like to be able do is to buy a pair of glasses for a couple of dollars from a concession stand at Tokyo (or London) Airport with a British (or Japanese) flag printed on their cardboard mount. You put them on, and suddenly all the signs are in your own language.

(Image adapted from this original.)

One of the things that a hologram must recreate is light polarisation.

So - you start by making a sign for Ueno Station in Japanese that reflects polarised light in one orientation. Simply placing a sheet of polaroid in front of the sign should do the trick.

You half-expose the sign to create a white-light hologram. You then do the same for the sign in English with a different angle of polarisation and do the other half of the exposure. The result should be a sign that, when viewed through polaroid sunglasses with the appropriate angle, lets you read the sign in just Japanese or just English. The other language should be (almost) invisible. It may even be possible to do more than two languages at once, depending on how finely the polarisation angles can be distinguished.

Of course, it doesn't just work for station signs and street names. It should be possible to implement the whole scheme for free in any city, simply by insisting that advertisers who use the technique also pay for some of the public signs...