It's the so called Voight-Kampff test in the Sci-Fi movie Blade Runner to test whether a human is real or a so called "Replicant".

The "profile" of one "best" person is not a statistically reliable indicator of what type of people to look for. From your description this person looks like a statistical outlier; there are many thousands of people out there who "barely gives a crap about the job" and are really lousy at what they do.

If you want to establish more reliable criteria for people to look for, you should establish a population of maybe the 10 best programmers or even the 20-50 best programmers you have run into during your 20 years in the business and look for commonalities among them, their driving forces, their passions and so on.

I have run into a few geniuses in the field where I'm at and I have found only two common factors among them; they are very intelligent and they are good at what they are doing. Apart from that, I have found them to be very different from each other when it comes to personal traits, interests and political opinions. I could not guess that they would have the level of skill that they had. Only in some cases the person was "socially awkward" and/or had an "exotic" taste for clothing where I could figure that either this is a total fool or an utter genius and the latter turned out to be true. In another case one guy was dressed as a "goth" with piercings and makeup, it was easy to dismiss him but after a while I realised that there was more to him than his style and looks.

I generally don't delete files by mistake but there was one day when I did. I had a directory with .rar files that I wanted to remove while doing other things that involved repetitive tasks. So I issued rm *.rar in the directory I was at and removed the files that I wanted. Since the tasks were repetitive I issued commands from the command history and when I was in another directory I mistakenly brought up and entered the rm *.rar command again. Not funny.

I think it would be useful if there were some sort of prefix (like I enter !rm, !chmod, !format instead of rm, chmod, format) that prevents the given command to be either stored in the command history or be prevented from being invoked from the command history again.

For examples of using vocal fry in singing look at the following clip that is "I Don't Wanna Miss a Thing" by Aerosmith:

http://www.youtube.com/watch?v=Vo_0UXRY_rY

pay particular attention at 0:31, 0:39, 0:48, ...

Or look for example att "Angels" by Robbie Williams:

http://www.youtube.com/watch?v=Fl4q2WBCywg

at 1:30. So vocal fry is also used for accentuation and is not merely a trick to compensate for lacking vocal abilities in singing. Vocal fry is also used to activate the vocal cords so as to help building up a vocal technique with a smoother transition between what is called "chest voice" and "head voice" without disconnecting or bursting into falsetto when reaching higher registers or "slamming" into chest voice when transitioning from higher registers into lower.

Well then, there are four ALUs in each core and not two cores. Just because the said performance is higher per core in the next generation CPU doesn't make it right to double the core count the way AMD has done. Hyperthreading or SMT as other CPU manufacturers than Intel call it can allow for more than just two threads per core. The implementation AMD has done is merely an enhanced version of SMT, no more no less.

Sure everything is nice and dandy in theory whereas in practice things are not so shiny and the Bulldozer family has failed miserably to impress even with their line of Opteron 6200 CPUs on server benchmarks. As the per-core licensing scheme is becoming the standard licensing scheme among servers, the Bulldozer family of CPUs looks very unattractive right now.

Yes, this is what I think they should have done.

I totally agree with you, their strategy don't make any sense at all. They probably didn't realize that this would be the consequences.

Yeah, it's like AMD's new Bulldoer family of CPUs. They have all of a sudden doubled the core count of the CPUs so that their line of quad cores is now "octocores" and their octocore Opteron 6200 family of CPUs now have a whooping 16 cores.

What they did was implementing a form of hyperthreading by throwing in an extra integer ALU into each core. They changed the name "core" to "module" and now falsely claim that each "module" now has two cores. If that were true then the old Sun UltraSPARC T1 would be 32 core, the Pentum D would actually be quad core (as it is also specified to have two ALUs per core), and all Intel processors that support hyperthreading can double their core count.

Tests done by Anandtech and other people indicate an underwhelming performance on these CPUs so I was a little confused as to why they would resort to such a cheap and fraudulent marketing trick, but I have now figured out what this is all about. As many people state, the Bulldozer is mainly targeted at the server market with their Opterons (that also has shown abysmal results in server benchmarks) and when it comes to servers, not only the workload is different but also the software licensing. A lot of server grade software is actually licensed on a per-core basis, i.e. the license you pay for a certain piece of software is based upon the number of cores you intend to run it on and not the number of systems as is the case with PC grade software. Microsoft used to charge their server software on a per-CPU basis or per-chip basis but they are already transitioning into a per-core license model starting with their SQL Server Enterprise 2012.

So, by doubling the core count instead of just calling it hyperthreading, they can generate twice the license income for software producers.