Yes. Always cherry pick first before trying to get a broader perspective. This is the best way to get off on politics rage. This is why we are talking about this right? Right?

Also, the article's source seems to be down so I don't even figure out how in world they are claiming "$7 million for each additional household served".

The parent hit it right on the head. Stanford CS still believes in teaching the people -- who are likely to professionally program -- using Java and C++. Seriously, the only reason this is being discussed like this is because of Stanford's unusual course numbering system. This is not the usual meaning of a "101" course. This is not a required intro course and most technical people will not bother taking it. (Knowing Stanford students, "techies" would probably consider it beneath them.)
The first programming class most people [who can hack it] take is the 106 series. The 106 series are the classes you expect all of the engineers, and most of the hard scientists (bio, chem, etc.), some of the economics majors, etc. This also includes the CS majors that don't skip to 107. Stanford has had an introductory non-majors class in Javascript for a while, CS 105. I only knew a couple of people who took 105 (both econ majors), and more than I am going to bother to count who took 106. (I hung out with techies.)
Just to give some perspective, lets check the numbers. Enrollment for 106A in 2009-2010 was 1087 over the year . The number I am pulling up for CS105 was approximately 300 a year in 2007. Final FYI, next year CS101 is only offered 1 quarter and CS105 is offered for all 3.
tl;dr non-event becomes headline due to misleading name

That is as far as I got. Every alarm bell in my head was ringing. Quite unpleasant. So I went and found something of possible value: http://www.cra.org/resources/crn-online-view/undergraduate_cs_degree_production_rises_doctoral_production_steady/ . Long read, but take a look at those charts. I think I am willing to step out on a limb and say, "Man people like the idea of getting rich for not much work, and during the dot-com boom CS was the place to pan for gold." I need to find longer term data though. Oh and the claim about women in the above quote is not born out by this data.

Current computer science graduate student here. My anecdote is that I have directly turned one conference into one summer internship. Networking can pay off, who knew?

Yeah outrage number 2...

Nevermind that the US has never stopped making things.

(citation).
Lets go out and find some more of those fact thingies shall we. Sstatistics for 2002 and 2007. (for pretty pictures of 2002)
Only two data points, but it beats the pants off of the article's zero. Lets see total manufacturing sales beat beats inflation. Salary per worker roughly matches inflation. The first derivative on manufacturing jobs is negative.
And by the by, if you want to increase manufacturing jobs in the US, you are going to need to increase the [already high] worker productivity given the rather high labor cost in the US. This is a form pf innovation. Just saying.

Your definition of NP-completeness is a bit off. An NP-complete problem is a problem that is in NP and any problem in NP can be reduced to it in polynomial time. (This eludes some technicalities, but you can look them up if you really care.) So any problem in NP can be reduced to a 3-SAT query in polynomial time. If this result is correct, any problem in NP can be reduced to 3-SAT in polynomial time, which can then be solved in polynomial time by this algorithm. This results in a polynomial time algorithm for any problem in NP. Hence NP is a subset of P, and P==NP.

Quick correction. The integer factoring problem is in NP, but is not known to be NP-hard. Here are a couple of explanations:
http://cstheory.stackexchange.com/questions/159/is-integer-factorization-an-np-complete-problem
http://en.wikipedia.org/wiki/Integer_factorization

For more information about what experts think about P =? NP, check out Bill Gasarch's wonderfully entertaining poll: http://www.cs.umd.edu/~gasarch/papers/poll.pdf

...there is a Guardian copy editor who is going home proud of themselves for a job well done.

Estimate of US health care cost in 2009: $2.5 trillion.
CBO estimate of the 2001-2017 total cost of the wars in Iraq and Afganistan: $2.4 trillion .

This is also kinda depressing.
But back to space, NASA's annual budget $18.724 billion.

Psuedoscience is I believe the word you are looking for.

Both do the following: they detect when some cores are being unused, they then give less power and decrease the clock speed to some of less used cores, and the power/clockspeed is then increased on the rest of the cores. The AMDs have do this by having 2 different of modes of operation with fixed power distribution/clockspeed settings for each mode. Intel does something more dynamic and on-the-fly.

Now I might have mis-summarized the article, but shouldn't that have been the article summary instead of a rephrasing of the article's lead?

Being forced into showing some humility over this is a nice thought, but...
Felix Ortiz is from Assembly 51. Lets take a look at the 2008 election results:

State Assembly - District 51 - General
80 of 80 Precincts Reporting - 100%
Name Party Votes Vote %
Ortiz, Felix (i) Dem 14,648 86%
Garcia, Luis GOP 2,134 13%
Coen, Grace Con 282 2%

I believe he would be fairly safe claiming 1+1=1 with those numbers.

If you read (or at least scan) the paper being referenced by the article, this is only studying sobering up faster (or more accurately the time until BAC is effectively 0). They are not testing headaches and hangovers. The only mention of headaches and hangovers is a throw-away sentence in the discussion section at the end.

I kept thinking that they just took the 20 min. running time and divided that by 1 day. This is roughly 1.38 percent. Make an assumption about energy consumption being constant while the machine is running, and viola instant buzzword compliance. But then I read the article,

*the JuGene supercomputer at Forschungszentrum Julich requires about 52800 kWh for one day of operation on the full machine, the IBM demonstration required an estimated 700 kWh

Which is roughly 1.32 percent. So no big surprises.