No, actually that would be special relativity which has been tested to around 20+ orders of magnitude by cosmic rays as well as (arguably) tests of CPT symmetry which last time I checked (quite a while ago) was at about 18 orders of magnitude.

QED is 'only' at about 12-14 order of magnitude of accuracy (which is extremely impressive!). Indeed since QED incorporates Special Relativity it would be hard for it to be tested more accurately that SR since any test of QED is, by definition, a test of SR as well.

Models like this have been considered such as MOND (MOdified Newtonian Dynamics). These models were largely shot down by the aptly named Bullet Cluster. This is a system of two galaxies colliding at a high relative speed. The gas from the smaller "bullet" cluster collides with the gas in the larger cluster causing it to slow down, heat up and emit X-rays so we can see it.

So far so go. However you can also look at the mass distribution by seeing how it distorts the light from galaxies behind the cluster (this is called gravitational lensing). This shows that most of the mass of the smaller cluster has not slowed down and is now separated from where all the gas in the cluster is located. Effectively the collision has separated the matter from the dark matter because, unlike normal matter, dark matter has a tiny cross-section for interacting with itself or other matter. This is exceedingly hard to explain by modifying the behaviour of normal matter since you are observing a gravitational field where there is no normal matter.

It is only toxic because we have evolved to deal with air that is ~20% oxygen. Were the content 15% or 25% we would have evolved to cope with that. As for combustion enhancing firstly that does not really apply ~500 million year ago because all life was underwater where combustion is somewhat harder. Secondly the naturally combustable material today all comes from plants hence evolution would presumably have resulted in less combustable natural materials or better fire resistance because, as you point out, plants as combustable as our trees are today would rapidly die out due to fire and so would not evolve.

Unless there is some mechanism which is known to prevent evolution from being able to produce less combustable plant material I don't see the need to require a 20% oxygen atmosphere. There is also no way that lightning can "ignite the atmosphere". To get nitrogen (roughly 80% of the atmosphere) to react with oxygen requires a net input of energy so even if there were more oxygen you cannot get a runaway reaction.

Why would we have become extinct? Isn't one of the major results of evolution that life adapts to the environment in which it lives? I saw nothing in the article to suggest that it would have been impossible for life to adapt to cope with higher levels of oxygen.

I'm not sure that this is really true. The reactor appears to be able to burn already "spent" fuel rods from other reactors but this is not going to result in less radioactive waste but rather more. The dangerous waste is the fission products, not the remaining unburnt Uranium which is practically stable (half lives in billions of years). In this design they will be extracted from the molten salt and will then need to be stored somewhere resulting in an increase in the net waste stored since each fission generates 2 or more daughter nuclei and one common one is an isotope of Krypton, a noble gas, which will undoubtedly take up a lot more volume that than the original uranium fuel pellet it was made from.

Fukushima's failure was due to technology in that it relied on continuous power to provide essential cooling even after the reactor was powered off. Even putting the emergency generators on a hill would not help if, instead of a tsunami, the hillside they were on collapsed due to the earthquake. You would then be arguing that it was 'human stupidity' to put all the generators on a hill instead of in a basement. For me the 'human stupidity' factor was that they did not insist on flying in backup generators as a number one priority after the tsunami. However I would also argue that the technology itself is also flawed since it requires continuous cooling even after the reactor is subcritical.

Really? More unsafe than letting these people drive themselves because that is what we have now? Besides I imagine that it will be a lot easier to put restrictions on an autonomous car e.g. teenager can only go to destinations within a radius of x km from home, certain regions of a city can be blocked off as no-go areas, dementia patients could have destinations limited to certain fixed locations etc.

...and while arguing that autonomous driving systems, which replace drivers, are scary because there are idiots driving cars.

Sorry but this is simply wrong. Look up synchrotron radiation. For electrons this is a very important effect and your machine energy is limited by how much energy you can give to the electrons on each orbit of the machine. Even for the protons in the LHC this is a noticeable, but not energy limiting, effect.

Remember that the accelerator is in a tunnel usually through rock. Unless there is a fault line through the ring - which would be really stupid - the accelerator will be shaken and will need realignment but the damage should not be enormous. They have very successful accelerators in Japan.

True but a circular design for a electron-positron collider is far from the most efficient. At the energies needed to create the Higgs the energy loss caused by bending the electrons around in a ring means that the ring has to be far longer in circumference than a 'one-shot' linear collider would need to be. Worse if we find something even more exciting like Supersymmetry in our next run of the LHC starting this coming March you will never be able to increase the energy of a circular e-p machine to study it whereas with a linear collider you can extend it.

A circular machine only makes sense with heavier particles like protons but I question whether the cost savings of a single tunnel for both an e-p machine and a future proton machine will outweigh the massive increase in the cost of the magnets and accelerating cavities for the e-p machine.

A Darth Vader voice would be more fun for misbehaving kids:

"I find your lack of behaviour disturbing"

"I am your father"

"That was before you misbehaved, now I am altering the deal...pray I do not alter it any further."

"I hope so for your sake, your mother is not as forgiving as I am”

Sort of but I'm thinking that it is more about the process vs. result. Convergent evolution is about the end results: there appear to be only a certain number of basic eye designs which work and so evolution tends to converge on one of these solutions no matter where it starts. This result is talking about the fact that given the same starting point and the same environment identical organisms will evolve in the same way i.e. there are not just stable solutions which you arrive at but stable paths along which you travel to get there.

Anyway this is definitely the most interesting bioscience result I've seen in a while so I'll have to quiz my biological colleagues about it when I get the chance!

It does not appear to be quite the same thing as convergent evolution (but I'm a physicist not a biologist!). My understanding of covergent evolutions is that it is when two wildly different evolutionary paths end up with the same solution to a problem e.g. an octopus eye and a human eye are functionally very similar even though our last common ancestor certainly had nothing like it.

This is rather the claim that evolution is reproducible in the short term i.e. if you put the same strain of bacteria in the same conditions they will evolve in the exact same way and not find different evolutionary paths to the same goal. This means that evolution becomes predictable and you can then predict with some degree of accuracy how a virus, bacteria or cell will evolve. This has obvious applications for disease control and perhaps cancer too.

I understood that, or at least guessed that this was the reason. However when you are restricting the connection like that yourself you cannot go and claim that it is the fault of the other party. Either you need to admit that the limiting factor is your own internal network or you need to spend some of the large amount of cash that is flowing in from your subscribers to upgrade that network to handle it instead of using it to see whether you can break the record for executive bonuses.