I don't think the sky will actually go dark. As far as I know, the aerosols reflect infrared, but let through visible light, so life should go on as normal, while reducing the temperature. With volcanic eruptions, which also cause a cooling effect, a great deal more than SO2 goes into the atmosphere, e.g. smoke and dust, and the sky does go dark, so killing plant life if it lasts too long.

I just find the interface terribly awkward and error-prone. I am glad I don't drive, because the amount of concentration needed to prod the screen in the right places would be dangerously distracting. Dedicated buttons and knobs can be operated by feel, with maybe just a glance to see where things are on the console.

I use some electronic test instruments at work, and the modern ones require some touchscreen input, in addition to the physical knobs and buttons. This can be a right pain. One thing that happens is that when I am pointing out some feature on the display, I accidentally touch the screen, and Something Happens. I have to reset all the measurement parameters to get back to where I was.

I don't get why people think mobile phones are a modern convenience. They just make life more complicated than it needs to be. I don't mind a bit of texting, but I am terribly slow with a touchscreen, compared to using the keyboard on my laptop. The main problem is the error rate. I keep on hitting the wrong "key", so I have to delete (if I can hit that correctly), and try again. Sometimes, the button I hit flatly refuses to respond. Did I hit it? Dunno.

I am pretty sure touchscreens save manufacturers a great deal of money. That would apply particularly to cars and complex electronic equipment. What would be an expensive mechanical design and plastic moulding job, which is difficult to modify, becomes a software job, with virtually no hardware costs once you have a touchscreen to interface to.

I don't know if much of that went on in our IT department. In my department, which is electronic and mechanical design, we did have some pretty useless new engineers doing that. Maybe that is what they used to do in their previous job. When results failed to materialise, they were booted out.

You can waste a lot of time trying to persuade the code generator to generate the code you want. The time would be better spent just writing code, Chaps in the IT department at work, who ought to know better, put together a stock control system using Python, plus some clever framework. Development zipped along, until we tried to implement the detailed requirements for our business. The developers ended up fighting the framework. Some things couldn't be done, or if you could do them, would be dangerous, as in corrupting data in unrelated work areas. We dumped it, in favour of an online service where the people actually understand stock control. Seems to work quite well so far.

When I used to commute from Birmingham to a place a few stops out, I found the smart Virgin trains actually quite inconvenient. The space seemed to be so cramped. If you have standing passengers, they are hopeless. The clunky box things run by London Midland (as was) are far more practical. I presume that the streamlined Virgin trains come into their own over a longer distance, such as going all the way to London. Though why anybody would leave a good city like Birmingham to go to that den of thieves is beyond me. I wouldn't mind going to Edinburgh by high speed train. I used to do that trip when I was at uni. There is a lot of scenic countryside, but really nothing but sheep, so the faster the better, I reckon. Edinburgh is a fine and elegant city, and not infested by dodgy oligarchs, as far as I know. But it was a long time ago.

Economics is currently my favourite nerdy topic.

Unfortunately, it tends to be the people who are prudent with their money to bail out the spendthrifts.

Seriously, a large amount of consumer debt makes things worse for everybody when times get hard. Widespread destitution, made worse by debts, is a national problem. What do you do about people who are chucked out on the streets because they can't pay the rent, just let them starve to death?

Computing science could be classed as math, but most programming does not require knowledge of computing science. I learned math without even having a computer. It is worth noting that quite a few numeric algorithms were developed well before computers existed. Doing the calculations by hand was a frightful slog, but that is what people had to do in the old days.

For example, there is the Euler method, which is an iterative method for solving differential equations, still much in use today. Wikipedia says this dates between 1768 to 1870. No computers in those days. But does anybody learning programming learn the Euler method? In order to understand it, you would have to know what a differential equation is, and to understand that, you have to understand calculus. Now that really is math.

You can write programmes without understanding the underlying math, because you have libraries implementing algorithms and data structures. This is not at all good for learning math. In fact, treating all this stuff as magic black boxes that crank out results given inputs could result in serious errors and inefficiencies, if misused.

I used to calculate Fourier series with pencil and paper and an electronic calculator. When I came across the Fast Fourier Transform algorithm, I thought it is really cool, and one the first programmes I wrote was an implementation of the FFT. It was a very bad implementation, but it did work. What this exercise did teach me was better programming.

In the case of numeric algorithms, programming is a way of doing something faster, but don't actually add anything to the math.

is that when somebody steals it, you're stuffed. I am not talking about physically stealing someone's face, but hacking the computer that contains the facial recognition data, so it recognises someone else (the crim) is you. How would you prove that had happened?

A big problem these days is people trusting computers rather than good old paperwork. I heard a documentary this morning and a nationwide system in the UK, to connect up lawyers, courts, prison services, and other criminal law stuff. This would theoretically save time on all the paperwork. The trouble is, the computer system frequently made errors, such as failing to send a warrant to the prison when a prisoner is eligible for release. What the legals did was to keep notes, to check up on the computer system. This actually made the job harder, not easier. The people developing and running the system swore that the system was working, so they ignored the lawyer's paperwork. The documentary did not get to the bottom why the system was making so many errors. It could take years. Bear in mind that if courts are wired into the system, it is a wee bit difficult proving your case in court.

In the early days of voice recognition, I read about a lawyer who thought it would be cool to dictate his letters directly into his computer. First, you have to train the algorithm to recognise your voice. This lawyer rather liked his liquid lunches, and did the voice training in the afternoon. Trouble is, the algorithm would not recognise his sober voice in the morning.

The first thing I thought about was that producing ethanol by fermenting sugar or starch generates quite a lot of CO2. That is based on brewing and baking. Then you have to distil the ethanol from the fermentation product, which requires energy. I am not sure what processing is needed to get fermentable sugars from corn starch. Starches such as barley and wheat are digested only very slowly by yeast. Brewing beer normally relies on enzymes produced by malted barley. Sourdough bread culture relies on bacteria to break up the starch into sugars, which can be digested by yeast. Another thing I note from brewing is that it uses massive quantities of water.

As far as I know, standard building safety regulations say you should build to withstand a once a century event, e.g. high winds and their effect on tall structures. What I don't think you are required to do is withstand all events in recorded history. I would say, though, that nuclear power installations should be held to a higher standard than regular structures. What seems to be the problem with nuclear power is that failures tend to be catastrophic. There is so much stuff to go wrong.

I enquired of a nuclear safety engineer whether you could build in some kind of fuse, that would shut the whole thing down when stuff gets out of control. One problem is that such a shutdown effectively scraps the reactor. You would have the problem of false alarms costing billions. In my power supply designs, a fuse blowing would probably cost only hundreds, due to loss of service and a callout to fix the power supply.

What actual alternatives does Japan have? I think they have to import all their fossil fuels. I am not sure if the geography of Japan suits wind power. Maybe somebody can let us know.

Yes, there is historical data. But I don't think there is a fixed upper limit. I would like to see the distribution of tsunami heights. Bigger tsunamis are presumably less likely than smaller tsunamis, but I don't think there is some number where there is no bigger tsunami.

From what I understand about Fukushima, and its lack of tsunami protection, the weak point was the auxiliary power, which is needed to run cooling pumps and so on. With the benefit of hindsight, the auxiliary power could have been made more secure. As you say, you put critical stuff in higher locations, so it is more likely to survive a tsunami.

But where do you draw the line in making safety provisions? Japan is subject to earthquakes, so it is prudent to plan for that. But the magnitude and location of the earthquakes does not have definite boundaries. If an earthquake happens, and the planned provisions did not withstand it, then was that bad planning?