Amusing Zahir effect: just today, Userfriendly republished the "metric to standard calculator: $15, telescope: $270, mars lander: $135 million, the look on the scientists' faces: pricess" strip.

At least, reboots are something Microsoft are very good at.

Speed? Is speed the only criterion to judge an ISP? I do not think so. Provided the speed is reasonable, with regards to the price and other similar offers, there are a lot more things that matter for an ISP. A few at random:
- How often do their systems break down and leave you without network access?
- If a router breaks down at 6 a.m. on a Sunday morning, do you have to wait until 9 a.m. on Monday to hope someone will fix it?
- And do you have any information during or after the breakdown, or are you left wondering it it will happen again any moment?
- If you call the hotline, do you get a nice music and an incompetent droid reading a checklist, or a competent technician?
- Do they offer cool services, like native IPv6 or reverse-DNS customization (and IPv6 reverse-DNS delegation)?
- Can you get someone's attention for unusual problems, like your IP range getting into a spam blacklist?

As for me, I am happy to pay a little more and have a little less max bandwidth to be on the good side for most of these points.

What about Dirac, the HD codec developed by the BBC? It has at least two Free implementations, and there are probably no patents?
Does anyone here have a pointer to a comparison taking Dirac into account, both for low and high bitrates?

That completely depends on the mathematicians, and the kind of mathematics they do. For proofs that rely only on calculations, you do not need even to understand the low dimension case, just do the computations right.

But proofs with computations are rarely elegant. Some mathematicians prefer a more geometric approach, and for that, they need to see, un to a certain level, the objects in higher dimensions.

Furthermore, the 2D or 3D spaces we have direct access to are really limited. There are lots of phenomenas that only happen starting with dimension 4 or 5. For example, think of this 2D property: "two lines perpendicular to a common third line are parallel"; if you try to take it as is in higher dimensions, you get something false; fortunately, you can think in 3D and see that it is false. There are similar examples in higher dimensions. Curvature, for example: curvature of 2D surfaces in 3D spaces is misleadingly simple, compared to curvature of higher dimensional spaces.

Sometimes, there just is not space enough to build the objects you need in 3D space. For example, if you want to study circles drawn on a sphere, the object you need to make the properties apparent is a 3D hyperboloid in a 4D space. If you settle for a 2D hyperboloid in a 3D space, you end up studying pairs of points on a circle, which is rather boring.