In my mind it centers around this question:

Assume an arbitrary community of 100,000. Leaving out natural deaths, let's say you have an annual death rate of 10 by gun, 3 by knife, and 5 by other (beating by trout, scissors, whatever), for a total of 18.

Then remove all the guns. Does the death rate stay at 18, with just the distribution of causes changing, or does it go down, because some who would have died from gunshots now don't (from e.g. mass shootings, gun accidents).

Annual number of handgun-related deaths per 100,000 people by selected country (from http://en.wikipedia.org/wiki/List_of_countries_by_firearm-related_death_rate)

Australia: 1.06
Canada: 2.38
Germany: 1.24
Israel: 1.87
Japan: 0.06
Netherlands: 0.46
United Kingdom: 0.25
United States: 10.3

Actually that's not as big a contrast as I expected -- I thought the US was 20-50 times higher than the norm, but it's significantly less than that for most western countries. The worst mostly in Central America, but Mexico is only slightly higher than the US at 11.17.

50% more usable

Ok, you're right.

Perhaps I should have said "included in the cover charge". But the cover charge is largely paid by the healthy, capable members of society, while the services are needed by, and provided to, the more needy and less capable members. Whether one considers this to be better or worse than the "user pays" model depends on your politics.

That being said, it could be worth a reasonable payment if one could get it done in three minutes at the local pharmacy rather than taking 15 minutes at the clinic (which is just above said pharmacy).

You must live in the USA to have to pay for blood work. It's free in all the civilized countries.

Very interesting. Now can we please have those rules properly expressed in BNF http://en.wikipedia.org/wiki/Backus-Naur_Form ?

How about a new law explicitly insisting that ALL government data, except for personal data covered by privacy laws, must be made public within (e.g.) 24 hours. Exceptions only on a similar basis to the way warrants are done now: apply before a judge and get an x-month publication ban (x not to exceed say 72 months, except for real state secrets, where it could be longer).

The usual tariff is based on a concept called "weight-measure", which works like this:

- For cargo less dense than water, a given tariff is per cubic meter.
- For cargo denser than water, the tariff is per metric ton (one cubic meter of water weighs one metric ton).

If you think about it, this makes perfect sense, because anything heavier than denser than water has to be accompanied by enough air (i.e. empty space inside or outside the container) to make the average density of the shipment equal one, and anything lighter than water takes up just as much space in the ship as heavier cargo would. The result is that if you have e.g. a 2000 TEU ship, and each TEU is 35 cubic meters, a full ship will always generate 70,000 tariff units, whether it be laden with cotton candy or iron pellets.

Of course, shipping companies play both ends against the middle and can, with optimization, get better than 100% billing (e.g. by using fluffy stuff like household goods to provide the airspace needed to compensate for containers full of car engines).

In a previous incarnation I was a Systems Designer at a major container shipping company.

Consider the possibility that only the British and their colonies feel English hard cheeses worth celebrating. The rest of Europe seem to get their hard cheeses from the Netherlands or Switzerland.

Reading some of the comments at Gizmodo ( http://www.gizmodo.co.uk/2013/06/bbc-home-page-clock-to-be-pulled-after-one-complaint/ ) I begin to see the BBC's point. Apart from (possibly) being part of the template code and therefore requiring extensive site-wide testing, it might well probably break their (reportedly extensive) page caching.

Some there speculate that the old technique uses a client-side function call that is happily cached, while the new version either has to include client-side code that maintains and corrects for the local error offset (= extensive testing requirements due to multiple client platforms), or needs to include a server-generated time specifically customized for that client (= impossible to cache).