Funnily enough, my need for controls in self-driving cars isn't the one you usually hear ("I need to be able to take control in an emergency") - as has already been amply researched, (1) the cars are likely to be much better at not getting into trouble in the first place than their cargo, and (2) suddenly handing over control to a human who's been dozing away watching the scenery, reading a book or whatever, simply is a really BAD idea - by the time they've worked out what's going on, it will be too late anyway, and anything they do is likely to make things radically worse, not better.

No, it's a far more pragmatic, human one: Not every car journey I make is a predictable trip from point A to point B, with no regard to what comes between. Sometimes I don't know what route I'll take, until I take it. Sometimes A and B are the same place. Sometimes I don't even know, when I set off, where B is. Sometimes my plans change in mid-journey, at very short notice. And sometimes, how ever good the traffic updates are, I'm likely to spot reasons why the car's preferred route isn't the one to take just then. So I need a reliable, simple interface that will give me enough control over my "autonomous" car to get it to take a particular route, pull over, slow down, speed up and whatever. Because, say, I want to pop into that little shop we just passed, that I've never seen before. Or I need a comfort break. Or the dog's been sick. Or I simply decide I want time to take in admire the stunning view on that windy little back road. What I don't want, and would think very carefully before buying, is a vehicle that takes away some of the freedoms that come with driving for myself.

I have little doubt that, if the time ever comes when these machines finally overcome the hurdles (not least, understandable human prejudices) and make it onto the market (and, frankly, I hope that's not so far away - it looks to this casual observer as though the technology is reaching a level of maturity wher ethe hurdles are legal rather than technical), such features will be there - because, frankly, they're obvious, and any offering that doesn't have them will lose out to the ones that do. But they're definitely needed. Google's prototype's lack of controls is a publicity grabber, pure and simple.

Passwords at home, I write down and file (with the exception of hyper-important stuff like bank access, where I choose passwords significant to to me and just write down clear hints that will help me get them but no-one else). I reckon that, if anyone gets access to those, I have bigger problems to worry about.

At work (softwear techie) I had, on average, 20-40 different password-protected access of various types. I (a) followed a theme meaningful to me (usually based on hobby things I'd been doing away from work); (b) used a single password on all systems; (c) guarded it carefully and changed it if I had the slightest suspicion it had been compromised; (d) changed it everywhere at the same time, regularly; (e) wrote down expired passwords so that I could recover any I accidentally failed to change; and (f) tried NEVR changed it immediately before going on leave. I found the combination of a password meaningful to me and the drill inherent in changing it multiple times in succession (and them using it regularly from that point on) meant that I never had a problem. Yes, I only had one password - one breach would have been a bigger exposure. But I NEVER had to write it down - and on the few occasions on which I had a brief memory glitch I could, in the worst case, give myself a big clue by looking back at my previous passwords to remind myself of my current "theme".

The IBM of today is a slow-motion train-wreck that too many business people haven't tumbled to yet. At the executive level it hasn't been about the computer business for a couple of decades; it's simply about screwing as much money as possible out of anywhere it can be found, and damn the consequences. If that means selling off the family silver, that's what will happen.

My first thought was that this sounds eminently sensible. My second was "Just who decides what 'reproducible' means?" My third is that it sounds like a recipe for delaying and tying up in the courts anything the far right don't like, on the grounds that there's "too much conflicting evidence and opinion" for it to pass the test. Frankly - it's a trap. Don't go there. And, just to lay my cards squarely on the table, I have no direct skin in this - I'm not a US citizen.

There's another well-known effect in which sound propagates asymmetrically - up- and down-wind at ground level. Viscosity effects produce a wind speed gradient (wind shear), with air nearest the ground travelling slowest. The result is that sound waves travelling downwind are refracted downwards (making them easier to hear), and waves travelling upwind are refracted upwards (making them harder to hear). If the wind's at your back, you can be quite close to a noise source yet unable to hear it - whereas someone much further away on the other side of the source, with the wind in their face, can hear it quite clearly.

There will undoubtedly be niche uses for such devices, but so far I simply don't see anything even remotely resembling the "killer app" that might make anything (that's appeared so far, at least) get widespread adoption; meanwhile, they're simply competing in niches that are already well-populated and quite a bit more mature.. That's not to say it's not possible, but in my book, until and unless something appears that makes these things massively and obviously useful in ways that existing tech isn't, this is just yet another technological "flavour of the month".

Indeed. There's a phrase that goes back many decades - "pushing the analogy too far" - which is what's going on here. The "rubber sheet" picture is still fine for conveying the basic idea of geodesics, so it's still an excellent way of explaining "bent" spacetime to someone who's struggling with the idea. The fact that the analogy may be deeply flawed in other ways is irrelevant.

(As for models versus analogies, though - a model isn't usually perfect either; it's normally a simplification of the way things actually are, that is hopefully a close-enough fit to be useful. The more accurate the predictions you need to make, the better the model needs to be. In that sense the rubber sheet analogy is a model as well; just a very poor one.)

Except the poor beggar on the other end of the line is simply doing a job, and probably for mediocre wages because it's what they could get. There's no reason not to do them the courtesy of staying civil.

Agreed. My response is more terse, along the lines of, "I object on principle to having my potential internet access censored."

Position 2 is not atheism. Atheism is the absence of belief, not the belief in absence.

How someone in the US or UK might pronounce her name in the absence of other information is irrelevant anyway. It's her name, so her answer is definitive - even if she says it's pronounced "Farquharson". Any other pronunciation is plain wrong.

...the old, hoary "science has been at a loss to explain.." hyperbole. Really? In reality, I suspect any number of science students could have made a pretty good guess on that one.

If accepted, this contains the idea (and evidence) that simply measuring a particle results in the observer becoming entangled with it. I've been pointing out for some time that so-called "spooky action at a distance" potentially melts away if the observer becomes entangled with the particle he measures, as follows:

• - One observer measures one of a pair of entangled particles, and becomes entangle with it (and, by implication, with the other particle)
• - At some other, indeterminate time and location, a second observer measures the other particle, becoming entangled with it (and, by implication, with the first particle and the first observer)
• - The two observers now compare their (entangled) results - which, being entangled, cannot help but correlate

Viewed in that way, no "probability wave" collapsed, nor did any signal travel at greater than the speed of light - let alone (as seen from some frames of reference) backwards in time. Entanglement forced, as seen by the entangled observers, two matching results.

Philosophers have nothing of use to say on matters of this sort; all they can do is play elaborate, aggressively argumentative word games, shifting their ground whenever they need to. Personally, I'm still waiting for a clear, SCIENTIFIC definition of "free will". Give me that, and I'll answer the question.**

It's a bit like the old saw about whether a tree falling in a forest makes a noise when there's no-one there to hear it. Define in scientific terms precisely what you mean by "a noise", and you'll find there's nothing to argue about. If a noise is a particular type of vibration in air or another medium, the answer is yes; if your definition's more complex, involving the detection and recognition of those vibrations in a human auditory system, the answer is no. Something inbetween will depend on your precise definition - but it will still be unambiguously answerable. The point is, definition is everything. Fail to define what you're arguing about - such as "free will" - and people will simply swap backwards and forwards all day between two or more conflicting definitions (without admitting or even necessarily recognising that the definitions are indeed conflicting), using each to argue that you're still wrong.

**For my money, by the way, the answer is almost certainly "no". The human brain is a hugely complex mechanism, true, but it's still just a mechanism, and I can't personally think of a definition of "free will" that isn't constrained by that. There's plenty of experimental evidence, for instance, to show that it reaches decisions, and that an observer with the right equipment can even detect those decisions, well before the individual becomes consciously aware of them, No - feed the identical information in under identical circumstances, get the same results from any randomisation mechanisms (such as quantum states) involved along the way (totally undo-able in reality, but we're talking theory here), and you'd get the exact same result out. So, even though I feel like I'm sitting here typing this because I decided "of my own free will" to do so, in fact it's simply a predictable consequence of my current mental state (combined with all my sensory input., etc., right now). Personally, I don't have a problem with that.

Myst was released 20 years ago, and the fact that Cyan feel the need to trade on the reputation of a property that old ought to be at least as much a warning sign as it is a cause for excitement. Yes, Cyan may technically be the company that produced it - but, whilst companies can theoretically live for ever, people move on. It's an absolute certainty that any new property would be produced by different people, under different management, with potentially different attitudes and values (and I've seen at first hand just HOW much of a difference 20 years of change in a company can make), using a different engine, to vastly different market expectations (Myst was good in its day - but if it launched today, I suspect it would barely make the discount shelves).

Good luck to them, by all means - new, quality products in the market are always welcome. And if they get to market, and it's any good, I may buy it. But this is no different to any other company coming along and saying "We plan to develop a new game, and it's going to be great, so give us the money to let us do it".