I'd rather have more accurate models than more precise models.

Bad models don't get any better by adding decimal places.

I expect that accurate modelling of something as complex as climate is really, really hard.

In that case it's lucky the method in the article doesn't gain root access, and can do nothing beyond what the "shell proxy" app can do.

1) Andrews & Arnold Ltd don't have 4 million numbers. They have fewer than 100,000 geographic numbers, plus a few tens of thousands of non-geographic numbers, assigned to them by the UK telephony regulator. I suppose it's possible that they could have agreed to use more through another provider.

2) Trapping a few telemarketers and tormenting them for entertainment purposes is fine, as is making money for receiving these calls, but what will happen in practise is that they will answer a lot more "wrong numbers" from regular people who have mis-dialed. If they search their existing CDRs for rejected calls to their unused numbers they will almost certainly find that there are a few numbers that already receive many call attempts because the number actually dialed is similar to some other genuine number. Recording and using mistaken calls from "your mum" for entertainment purposes and charging her for the privilege is somewhat immoral in my opinion.

3) The correct behaviour is to reject unused numbers with an NU indication. Anything else is antisocial and profiteering, but they would be welcome to do this on their freephone numbers (where they are charged for the calls).

Note: I work for a telephone company that does have millions of numbers assigned, including many premium rate and pay-per-call numbers. We could make a significant amount of money from caller's mistakes, but that would not be right.

Sure. Bubblewrap too.

It was kinda hoping nobody noticed, but it clearly heard one of the other robots saying "Butter Phingers".

You still need all the mechanism of the machine to transport hundreds of feet of film past the scanning head at a constant speed without breaking it and keeping it nicely spooled. If the 35mm film had sprockets perhaps they could have used the mechanism from an existing 35mm film projector instead of having to make their own constant speed mechanism for the sprocketless film.

The phototransistor (photodiode, CCD etc) method is a long established technique for playing back an optical analog sound track from film http://en.wikipedia.org/wiki/Sound-on-film and I can't see any particular need to reinvent the wheel for this. If they considered that the film was too fragile to pass through the machine more than once, it would not be difficult or expensive to have 8 phototransistors so that all the tracks could be played back and recorded digitally at the same time.

Poor reception means that the phone has to transmit at higher power to reach the cell base station.

The Sony warranty is worth nothing to them. It may even have negative value. Anything sent back for warranty repair/replacement would be updated to the useless version of the firmware.

The time shown on the screen isn't the same as the time in the internal hardware clock. It shouldn't practically matter what time the hardware clock is set to provided it always increments one second per second. You can't tell from the time on screen if the hardware clock is correct.

More likely they have the faulty chip, but the internal clock is wrong. This would be more or less the same "fix" as removing the RTC battery for a while.

I would expect a small number of owners to have the problem for 24 hours at some random time. There have probably been some PS3s that have done this before March 1st.