Search This Blog

Follow adrianbowyer on Twitter

My home page

Monday, 18 November 2019


Note: there may be some flaw in the logic of this argument.  Or, alternatively, it may be an established result in genetics that my (brief...) researches have failed to find.  If either, tell me in the comments and I will amend as necessary. But if neither, I present it as a possible explanation of an important phenomenon.

There is a meta-analysis in Nature Genetics on 14,558,903 (!) partly dependent twin pairs that shows that the heritibility of a wide range of traits is 49%.  That is to say those phenotypical traits are 49% determined by a person's genes, and 51% by their environment.

This seemed rather close to 50% to me, and set me wondering if there is an evolutionarily stable strategy (ESS) at play that is forcing the figure to 50%.  The first such ESS that was discovered (by R.A. Fisher) is the one that gives a 50/50 sex ratio in a wide range of species.  An ESS doesn't have to settle at 50%; for example the ESS between hawks and doves in the human population (and many others) is heavily weighted towards the doves.

So.  The question to ask is, if you are a gene, in order to maximise your fitness how much of the phenotype you develop should you control, and how much influence should you hand over to the environment?  (Note importantly that "the environment" here includes all the other genes in the organism and any influence that they might have over that phenotype; and that those genes will be subject to the forces I am about to describe as well.)

Let's look at a simple specific made-up example.

Suppose you are a gene that controls everything about a fur-colour phenotype and you are a gene for green fur.  There is another, rival, genetic allele for brown fur.  In a verdant forest you, green-gene, will leave more offspring, and brown-gene will diminish in the population.  In contrast, in a brown-coloured savanah your brown-gene competitor will come to dominate.

Now suppose a mutation arises that causes neither green nor brown, but that turns the phenotype fur the colour of its surroundings when the organism of which it is a part is developing.  Clearly individuals posessing that new mutation will be able to colonise both the forest and the savanah, and the new gene will come to dominate (assuming the cost of its working is not greater than that of the other two).  This new gene has handed over some of the determination of its phenotype to the environment, and has thereby gained an advantage over its more dictatorial predecessors.

But a gene that has no influence at all over any phenotype, and which leaves the determination of the phenotypes to the environment (which includes the other genes, remember), has no fitness because it cannot influence its reproductive success.  It may get carried along for the reproductive ride as non-coding DNA, or it may just be eliminated altogether.

In our example, suppose the developing organism adopts the colour of the nearest object during its development, as opposed to switching between just green and brown.  And suppose it grows up in a nest surrounded by bright orange flowers.  Clearly, in this case, the gene will have given up too much control to the environment.

So in general we can see that it makes sense for a gene to allow some environmental influence over its phenotype to allow a versatile response to different conditions.  But it must not allow too much environmental influence lest it loses control and hence loses fitness.  We can show this graphically:

The X axis and the blue line are the proportion of the influence of the gene on the phenotype, which allows the gene to control things. The red line is one minus this - the proportion of the influence of the environment on the phenotype, which allows the phenotype to adapt to its surroundings in some way.  If we multiply those together, we get the yellow curve, which is the benefit the gene gets for a given proportion.  That is, the gene "wants" control, and it "wants" adaptability, but these are in opposition, so it can't have all of both; when one increases that increase necessarily drives the other one down. Unsurprisingly, given the symmetry of this example, the maximum - the most beneficial point for the gene - is at 50%, which is the figure we set out to explain.  But the symmetry seems a bit of a cheat.

So suppose we change the form of the blue line (and consequently the red one) so it is merely some (in general non-symmetrical) function, f, of the genetic proportion (call that g) along the horizontal axis.  As in the graph above for the simple case f(g) = g, the new general f(g) generates values in the interval [0, 1].  The yellow curve, the benefit the gene sees, B, is now:

B = f(g) . [1 - f(g)] = f(g) - f(g)2

(all other things being equal, the gene's fitness, ω, will be proportional to B) The maximum value of will be where dB/dg = 0:

dB/dg = f'(g) - 2f'(g)f(g) = 0

which gives:

f(g) = 1/2 .

So the watershed value of 50% for the maximum benefit to the gene doesn't change, regardless of the form of f(g).

There must, of course, be exceptions to this simple analysis.  But it does show how, for phenotypes that must adapt to their environment, the genes that control them would be expected to hand exactly 50% of their influence over the phenotype to that environment.

Sunday, 12 May 2019


I prompted a bit of a discussion on Twitter the other day by asking, "How deep a hole would you have to dig on Mars for the atmospheric pressure at the bottom to be 1 bar?"

1 bar is 1 Earth atmosphere, for non-metric people.  It turns out that the answer is 55 Km deep, which is rather inconvenient if you want to use this as a way to produce comfortable human living space...

But, as SCUBA divers know, every 10m deep you dive in water increases the pressure by 1 bar.  So how can we live on Mars under water?  We'll need a copious supply of water anyway that we would recycle, and the diagram above gives a rough idea how we could use that both to pressurise the living space and to create a radiation shield.

On Mars the 1 bar water depth would be at about 28m, because of the lower gravity.  But, if we were prepared to have a lower air pressure, that depth could be reduced to 20m.  (That would give the same pressure as the inside of a cruising airliner cabin - about 0.7 bar.)

So we dig a circular hole about 35m deep and about 40m in diameter.  We line and seal the walls to make them air and water tight.  Then we put in a transparent skin that forms the bottom of a water tank about 15m from the floor, and another skin at ground level.  We fill between the skins with water, while raising the air pressure under the bottom skin to balance the load.

When the structure is complete the water would be supported by the higher air pressure underneath.

The water would freeze at the top because of the Martian surface temperature.  The top skin is needed to keep the water/ice dust-free and to prevent loss by sublimation.  For safety, the bottom skin under the water would probably have to be made strong enough to survive both a loss of atmospheric pressure underneath it, and a loss of the water above it.  Alternatively, almost all the water could be allowed to freeze.  Then it would become a strong part of the structure, especially if it were mixed with transparent fibres with the same refractive index as ice.

As anyone who has dived in clear tropical waters knows, plenty of sunlight would be available in the living area at the bottom.  And water is an excellent radiation shield, so, together with the surrounding ground, the problem of Martian surface radiation would be eliminated.  If care was taken to control the freezing of the top layer of water to eliminate bubbles, it may even be possible to make the water roof optically clear, so the Martian settlers could see the sky.

Digging a cylindrical hole using a descending circular shield and lining the walls above the shield with a resin/rock-dust composite is a job eminently suitable for a robot.  It could work away for years before people arrived, preparing a hexagonal grid of living cylinders interconnected by short corridors at the bottom.  Another robot would be ferrying ice from the Martian poles to the site.  When people landed they could do the more fiddly job of fitting the skins and airlocks, gradually expanding to occupy more cylinders as needed.

In order to get all this working, we'd obviously have to try it out on Earth first.  Maybe we should do so at Coober Pedy in the Australian outback.  People there already live in artificial underground caves because of the heat.

And the whole project might run at a profit because of the Coober Pedy opals the robot dug up...

Monday, 18 March 2019


This is a post about heartburn and the origins of religious belief.  Run with me here...

You can give pigeons religion.  You observe them and, whenever one sticks its left wing out, say, you give it some food.  Soon the pigeons have a conditioned reflex that sticking out their left wing produces food. It becomes their "Give us this day ours daily bread" prayer.  Random coincidences can give the same effect without the intention of an experimenter: breaking mirrors and bad luck, and so on.

I have suffered from heartburn for a few years.  This year, I decided to do something about it.  So

  1. I changed the foods I eat.
  2. I changed the time of day of my main meal.
  3. I stopped drinking regular coffee.
  4. I started taking betaine hydrochloride (to increase stomach acid) and pepsin before every meal.
  5. I started eating a small amount of ginger (of which I am anyway fond...) after every meal.
Now.  I'm not a complete moron.  I know that the scientific way to do this would be to change just one thing at a time and to record the results.  But I wanted a cure NOW, so I just threw everything that might work at the problem at once.

And that everything did work.

But now I don't know which of those five changes made the difference.  I could just cut one out at a time and see when the problem returns.  But I really don't want the problem to return.  So I'll just carry on with my cure in ignorance of which bits of it actually worked and which are just snake oil.

And, of course, the snake oil bits are my religion.  They do nothing.  I half have faith that they work.  And I'm not prepared to subject them to empirical verification.

They are my pigeon's prayer.

Monday, 27 August 2018

The Support Shift of Sam McGee

There are strange things done 'neath the midnight sun
     By the bods who moil with code;
An embedded trace gives an endless chase
     When you compile in debug mode.
The panel lights have shown odd bytes,
     But the oddest they displayed
Was that night I thought, in User Support,
     That I'd do a sys upgrade.

From Seattle ground on Puget Sound,
     Where the Duwamish meets the sea,
The system spread like a wound that bled,
     But should not've passed the quay.
I was always told, by coders old,
     That it drained you like a spell;
But I had no choice; the boss's voice:
     "Linux? Rot in Hell!"

As I sipped my brew the screen went blue,
     And then the helpline rang:
"My Word doc's gone! It's almost dawn.
     “You're the one that I'll harangue.
"I've a meeting at ten. Must I use a pen?
     “You're supposed to make it work."
I could tell from his tone at the end of the phone
     That this one was a jerk.

But he had a point: in this hardware joint,
     The server's meant to serve,
With an uptime that, quite unlike FAT,
     Would every bit preserve.
I set down my cup, took the backup,
     Then mounted it in the drive,
And thought to myself, "If the link were ELF,
     “I'd have it up in five."

The drive-LED flashed. The head then crashed.
     My tea soaked round the keys.
So I cursed an oath at the undergrowth
     Of the open-plan tubbed trees.
Then I recalled the machine installed
     To test a new release.
Maybe that would run better than none
     And finally give me peace.

I plugged in a mouse, and keys unsoused,
     And a postcard-sized green screen,
Then I hit reset, with my brows knit,
     Hoping that release was clean.
My luck was in. Beta for the win.
     It booted to a prompt.
I ran the scripts and checked the MIPS;
     It wouldn't end up swamped.

The Post-it note that I wrote
     Had brief words of advice.
"Admins", it said, "The machine is dead.
     "The disk has failed us twice.
"But the spare server is a life preserver
     "That'll run till half-past three.
"When the next shift, if you catch my drift,
     "Takes over" - Sam McGee.

Wednesday, 4 July 2018


It is pretty easy to add a rectangular fly screen to a sash window.  But the problem with sash windows is that the maximum they can open is half, and the sash mechanism is less reliable than a simple (or complicated; see the picture...) hinge.

So how about an elastic concertina fly screen for a hinged window that folds away into the surround?  It has a magnetic strip like a fridge door that attaches it to the three opening sides of the window frame, and, as the window is opened, it un-concertinas (if that's a verb) to fill the gap.

When the window reaches a certain point (say open about 20 cm) the concertina is fully extended.  Then the magnetic strip pulls off the frame and folds itself away again into the surround using its stored elastic energy.

The magnet eventually re-attaches when the window is re-closed.

It should be simple to make, and could probably come as a retro-fit kit for existing windows, as well as being an option on new ones.

Thursday, 14 June 2018


This is a couple of Augusta Westland AW609s.  They are vertical take off and landing aircraft that rotate their engines and propellers when up in the air to fly horizontally.  There are quite a few other VTOL aircraft that use this principle.

If you look, you can see that the propeller blades twist like a helix (all propeller blades do this; it compensates for the fact that the tip is moving faster than the middle).  The blades can also be twisted as a whole, which is called variable pitch.  

In a helicopter with just one rotor, variable pitch is essential for forward flight because the blade that is moving forward with the direction of flight is going fast into the air, and so generates more lift, whereas the one on the other side of the rotor that is going backwards relative to the air generates less lift.  Without the blades twisting every half-rotation using their variable pitch to give more lift on the back stroke, the helicopter would simply tip over and fall out of the sky.

But this effect is neutralised with two rotors like the AW609, one on the left and one on the right of the forward direction, as long as one rotates clockwise and the other rotates anticlockwise.  Then the forces balance, and the blades don't need to flap with each half-revolution.

The problem with planes like the AW609 is that the propellers need to be big to act like helicopters, but that makes them very inefficient in horizontal flight, limiting both the plane's speed and range.  What would be ideal for VTOL planes like this would be a propeller that could also shrink to a small radius in horizontal flight, and expand to a big radius when helicopter-style vertical flight was needed.

Given the lack of need for variable pitch, this could be made to work with four-bladed propellers (rather that the three you see in the picture), or, indeed, propellers with any even number of blades.  The blades would be hollow, with one very slightly smaller that the other.  To reduce the propeller diameter the blades would be drawn through the hub and the smaller one would slide inside the slightly larger one opposite.  They would also have to twist as they did this, to accommodate the helical blade shape.

There are a few problems with this idea, but I don't think they are insurmountable:

  1. All current blades are not a constant-pitch helix.  This would be needed for them to fit inside each other.
  2. Careful thought would need to be applied to balancing the propellers given the slight difference in the sizes of the pairs of opposite blades.  The masses need to match, obviously, but so too would the moment of inertia, lift and probably drag.
  3. The blades could not be variable pitch, except when fully extended.
  4. The blades would have to have a constant cross-section.
  5. Your [it-won't-work-because] goes here...
I don't know if the aerodynamic compromises needed to accommodate the above list (plus the things I haven't thought of) would nullify the increased speed and range that would come from having a more-or-less conventional sized propeller for horizontal flight.

But it would be interesting to do some experiments and calculations...

Postscript 3 March 2019

A similar alternative that I thought of after I wrote this article is to have each blade telescopic and retractable inside itself. This would allow odd numbers of blades and probably be simpler overall (and certainly more symmetrical). 

Wednesday, 11 April 2018


This is an edited version of a letter that was published in the London Review of Books Vol. 39, No. 11, 1 June 2017.

Driving speed is easily controlled by self-funding radar cameras and fines; in contrast, MP3 music sharing is unstoppable.

Every technology sits somewhere on a continuum of controllability that can be adumbrated by another two of its extremes: nuclear energy and genetic engineering. If I want to build a nuclear power station then I will need a big field to put it in, copious supplies of cooling water and a few billion quid. Such requirements mean that others can exert control over my project. Nuclear energy is highly controllable. If, by contrast, I want to genetically engineer night-scented stock to make it glow in the dark so it attracts more pollinators, I could do so in my kitchen with equipment that I could build myself. Genetic engineering is uncontrollable.

We may debate controllable technologies before they are introduced with some hope that the debate will lead to more-or-less sensible regulation (if it is needed).

But it is pointless, or worse damaging, to debate an uncontrollable technology before its introduction.  Every technology starts as an idea in one person’s mind, and the responsibility for uncontrollable technologies lies entirely with their inventors. They alone decide whether or not to release a given technology because - if they put the idea up for debate - its uncontrollability means that people can implement it anyway, regardless of the debate's conclusions. (Note in passing that - all other things being equal - an uncontrollable technology will have greater Darwinian fitness than a controllable one when it comes to its being reproduced.)

In my own case I classify technologies I invent as broadly beneficial or damaging. The former I release online, open-source. The latter I don’t even write down (these include a couple of weapons systems at the uncontrollable end of the continuum); they will die with me.

I may be mistaken in my classification, with consequences we may regret. Other inventors may act differently: we may regret that too. But we shouldn’t make the mistake of indulging in (necessarily) endless discussion of what to do about a technology if it is uncontrollable. The amount of debate that we devote to a technology should, inter alia, be proportional to how controllable it is.

Technological changes have unforeseen and occasionally negative social and political consequences.  This is inevitable when something powerful impinges on things that are relatively weak like regulation; the same applies to the benefits. Fortunately the vast majority of people are well intentioned, and technology amplifies the majority along with its complementary minority. Much happens faster and more spectacularly, but the ratio of more good to less bad stays about the same.

Monday, 12 March 2018


Castaway, the first British reality TV show nearly two decades ago, dropped a group of about thirty people on the remote Scottish island of Taransay and filmed them as they argued with each other and fell out brutally and in a psychologically damaging way over the following weeks.

I watched the opening episode, which had the whole group in a room in London before they set out discussing what they would do and how they thought things would work, and I predicted to anyone who would listen (i.e. my family and the cat) that the whole thing would be a social and emotional disaster for most of them.  And so it was.

The problem was that - in that London room - they were all talking with each other excitedly and at length in a friendly, convivial, and engaging way.


Think of two island fishermen in their fifties who have known each other since childhood.  On a Monday their total day's conversation as they pass each other on the quayside might be:



And similarly every day of the week, with - perhaps - on the Friday:


  "Morning.  Storm's coming."


They, and the rest of their island community, have evolved a peaceful system of friendship and cooperation an essential component of which is not annoying each other with their personal views, history, random thoughts, and chatter.

Our two friends sit together all evening in the pub in silence, their pints of beer in front of them on the table, taking a sip every minute or two and thinking their own thoughts.  If something needs to be communicated (like a storm) they mention it, then shut up.  Occasionally the whole community all gets very drunk and sing and play the pub piano and talk nonsense for hours then, the following morning, their hangovers enforce a return to their normal reservation.

A lot of folk anthropology consists of just-so stories about how we are adapted to life in a hunter-gatherer village and how we carry that inheritance over to modern global civilised life.  Sometimes, it is claimed, conflict results; one obvious example is xenophobia.  But one thing we have certainly not carried over is the circumspect reservation that we can observe today in isolated small communities.  Every communications technology we have created - printing, the telephone, radio, television, the internet, social media - works against that reservation, and we embrace them all with delight.

And we wonder why we don't get on as well as the two fishermen.