Well, I'd check to see if the server's cooling fans might be spewing a stream of charged particles, van de Graffe-generator-style, with an electrometer whose input Z is up at *least* around 10^12 ohms---with improper grounding, those charged particles could be a nuisance! So, we ought to check for proper grounding via RF injection. If we find a stream of charged particles and either ground loops or lack of a decent ground in the area around the server, we'd be dealing with the possibility of large electrostatic fields occasionally building up and playing the occasional trick on *MOS silicon, or messing with SMD capacitors in the presence of passing charge carriers (radon daughter decay particles, passing cosmic rays or their post-upper-atmosphere-collision daughter decay particles, local sources of nominal charged particles)---typically this manifests as an occasional memory or transmission error, suggesting that if the server isn't using ECC RAM already, it ought to be (same goes for antiquated bus technologies with zero error-detection/retransmission ability), and that probably the next best step is check the output of ifconfig for checksum errors.
I mean, in an ideal world, software can be written perfectly, but hardware has physical limitations, and you have to look at the mechanical parts that fail most often and consider all possible paths of influence, right?

[note to pedants: yes, there are surely a few factual, conceptual, and deductive mistakes above---feel free to correct them, but don't forget, it's called "humor" *g*]

I believe it is generally understood that the presence of mass causes curvature of space-time around it, and the force of gravity that is exerted upon mass is due to space-time curvature. The universe can have a natural curvature (i.e., outside of curvature due to mass) that is not flat, as well, which predicts the final destination of the universe by changing the long-term outcome of expansion, iirc. but, ianap.

If someone falls asleep at the wheel and drives into the side of your house, can they be charged with assault?
How about brats with spray paint? Battery?
If they spray-painted 'idiot', that's like putting a sign on someone's back, and they're mocking a living being that is publicly known to have no cerebrum by making a public display on the defenseless body of that being. Now we have a potential hate-crime.

I think law school enrollment just shot through the roof.

Based on a few papers I found and a few quick, back-of-the-envelope calculations, it should help at least 10.3% of the infected population and at best will help 96%. The huge amount of variability comes from not knowing much about superinfection in HIV. I'd also like to know what strains VRC01 and VRC02 specifically /don't/ target; if the researchers are referring to HIV-1M, O, and N and HIV-2, then "over 90%" means 23/25 is covered, so I'm betting HIV-1N [see http://commons.wikimedia.org/wiki/File:HIV-SIV-phylogenetic-tree.svg ]---but, who knows.

Anyway---for every 100 person-years, there will be a few HIV reinfections in HIV-positive individuals, sometimes by viruses of the same exact subtype, sometimes by viruses of differing subtypes. Sometimes the viruses are more virulent than the original infecting strain of HIV. The time elapsed since the original HIV infection does not seem to make an impact on the distribution of times of second HIV infections. (Yes, I know that sentence could use re-wording, but exactly how is eluding me atm.)
http://jama.ama-assn.org/cgi/content/full/292/10/1177 suggests 5.0 reinfections per 100 person-years (population size 78) in SoCal, http://www.plospathogens.org/article/info%3Adoi%2F10.1371%2Fjournal.ppat.0030177 suggests 3.7% reinfections (population size 36) per 100 person-years in Kenya, and http://jvi.asm.org/cgi/content/full/79/16/10701?ijkey=30fccead91569e63031af4357a242da634620d52#SEC3 suggests 10.3 reinfections per 100 person-years (population size 20). A weighted average of these numbers, where the weights are the population sizes (not the best approach, but given the sparse amount of data found in the literature about this particular topic in general, it's better than no weighting, perhaps), comes out to 5.0 reinfections [reinfection or superinfection is very hard to define for HIV---see the methodology in the second article for more information about this] per 100 person-years. So if we look at http://en.wikipedia.org/wiki/File:HIV_time.png a timeline...

If we assume the number of reinfections a year is simply the reinfection rate (5.0/100 to get per-year) multiplied by the HIV-positive population that year, and guess that none of the population in 1980 (where the timeline starts) was superinfected, we get 22.7 million in 2008. If the population in 1980 had all been superinfected, we get 30.0 million in 2008.
On the other hand, if we assume that only the increment in HIV-positive population is eligible for reinfection (a lousy assumption, but with how little is known, it's as good as any---actually, I'm partly going off of the notes in the introduction of the second article about the known information about reinfection rates), and just multiple the difference from year-to-year in the HIV-positive population by the reinfection rate, we get 1.31 million. (It would not make sense to include the 1980 population here, based on the assumption made.)

We don't really know enough to guess, but we can probably assume there would be at least an easily noticeable impact. A 10% drop in HIV population would be very obvious---that's a few million people.

Actually, it's acceptable to use the traditional, common Germanic (versus Romantic) prefixes and suffixes with any word if it does not already have an established form in the part of speech or with the altered meaning required. -hood, -wise, -ish/-esque, -ing, -est, -er, -ing, -y, etc. Some more recent prefixes and suffixes are generally considered acceptable, at least outside of formal writing, too. Everything assimilated into Middle English is, really; it's just that some assimilations in recent Modern English are still considered unacceptable in formal writing, which is understandable. A handful of Middle-English-era Romance assimilations are -al, -ment, -ic, -ion, -tion, -er, -ive, -able, -or, -ity, and one common Greek assimilation into early Modern English is -ism. Now, don't forget that typically the root word's language of origin has to match the suffix's most recent language of origin before assimilation into English (careful, one suffix listed had been in proto-Germanic, Old French, Middle French, Sanskrit, Old German, High Germanic, Latin, and Old Norse, but was assimilated into Old English from Old German), and that you have to follow the same patterns that existing, established words do, so as to avoid consonantal clusters that do not occur in English... (amongstment strengthtion, anyone?)

Oh, and my on-topic 2-cents: isn't this what freenet is for?
Just don't post who you are, or why you're posting on freenet. Don't create any timing-trail, either, to make anyone think it might be *you* who posted this interesting piece of code on freenet.

I don't think artificial gender lines for clubs are a good thing. The various scouting and campfire/outdoors groups should just re-brand themselves as various types of character-building and skill-building clubs; BSA could tend to target more activities and skills boys would be interested in, while Girl Scouts could target the activities and skills girls would be interested in. Neither would discriminate, so [just stereotyping here:] tomboys who would be happier with boy-scouting could go that direction, and the less-"tough"/more-effeminate boys who would be happier with the Girl Scouts could go that way. (Again, I emphasize that those are *stereotypes*... I don't really believe the division would ideally be just like that! In fact, having boys and girls learn skills and traits traditionally associated with the opposite gender and breaking social gender role molds in the process is really neat imho.)
That being said, the chaperoning issue is still a problem, with the way our society handles gender privacy and bathroom issues (not that our system is necessarily bad, but it could probably use some improvement - see #1 for something I disagree with, and #2 for an amusing comic on that subject - I think having more unisex and family bathrooms would be preferable, with several private, individual bathrooms for every multi-person unisex one). Chaperoning could be handled more by the parents voting rather than by gender-grouping [how many intersex parents are there for the intersex children who self-identity as intersex rather than male or female? *grin*]. There are plenty of adults of both genders who treat all children with respect and would never dream of harming them, and these adults would make fine chaperones for any children, to the extent that those children and their parents are comfortable.
{:prepares self for flames:}

I've often found that I cannot bring myself to believe many things in the Bible, but neopaganism holds my interest. I also equate "God + Goddess" with the whole of the Universe (or David Lewis' multiverse) instead of believing in a supernatural God.
It works for me, and doesn't require me to believe in any "voodoo."

The second link only needs two bjt's (1 npn, 1 pnp) and 5x 10kOhm.
Even from the local radio shack, $8.99 for a small solderless breadboard, $6.99 for a 100' spool of solid 22awg wire, $2.99 for a 15-pack of 2n2222 npn's, $2.59 for a 15-pack of 2n3904 pnp's, $0.99 for a 5-pack of 1/8W 10kOhm@5%, and $1.99 for a female DB-9 crimp connector is only $24.54 (before any sales tax). If you can solder, you can save $7 by going with the 417-hole pre-etched PCB, and you can probably de-solder any NPN, PNP, and five 10k's from old busted electronics (dead CRT is a good example: fewer SMD's and more lead-ed/through-hole components; and for that matter, you can probably break off part of a PCB from a CRT, de-solder all the components, and use the traces already etched to save the cost of a PCB). If you have a few feet of solid 18-24AWG wire around the house, you can save $6.99; solid wire may be harder to find in small consumer electronics, but is still not hard to find in random broken appliances (and if you're soldering it, you can just use stranded wire from any random old cable or consumer device). For the DB-9, you can always chop in half any cable that plugged into a serial port (you'll need to figure which pin goes with which wire---use an ohmmeter, or if you don't have that, a voltmeter (many battery checkers are basically voltmeters) and a battery, or if you don't have that, improvise by using a flashlight with the battery cover opened (turn flashlight switch on, insert wire/pin into DB-9 hole being traced and connect to the open side of the batteries, and then try each of the wires in the cable to complete the electrical connection normally completed by the battery cover until it lights up)); it'll almost definitely have stranded wire inside, so for a solderless breadboard you'll have to use a piece of solid wire and solder them or, without solder, twist the ends together well, wrap tightly in electrical tape (duct tape can be substituted if you have no electrical tape) until of sufficient thickness, and screw into a wire cap (the kind used to connect two household internal power wires together) to keep pressure on the connection. You can forego the screw caps if you wrap very, very tightly with good tape and aren't worried about the connection holding forever.

Getting solid wire, sewing needles, or the points of safety pins to stick in the holes (not the big ones, but the small rectangular ones next to each big hole) of the OBDII connector usually works. If you don't have luck with that, you can re-bend some paper clips into a U-shape of appropriate width or use the curled end of a safety pin of appropriate size.

Really, any computer geek with patience can probably do this for zero cost ;) Even without a soldering iron, you can probably locate a section of PCB (again from CRT etc) with one NPN and one PNP BJT and without any surface-mounted components, remove other components by cutting the leads (keep in mind that longer leads might be snipped as close as possible to their component so as to create points for easy connection to wires) with wire cutters (or fingernail clippers, in a pinch), and cut unneeded traces carefully with an x-acto or razor blade (or by carefully sawing with a serrated kitchen knife, like a steak knife, in a pinch). The only part that you may have to dig around for is a cable with a female DB-9 connector at one end, and you may be able to get around that by squeezing the needed pins from the male DB-9 serial port between the insulating sheath and copper core of some wire by pushing the wire down onto the pin with small needle-nosed pliers (or tweezers in a pinch). Honestly, with a few hours and some old electronics that can be sacrificed, it's very jury-riggable, though like the parent poster states, you may need to pay a little more for decent housing and an actual OBDII plug.

http://www.elec-intro.com/obd-schematic schematic for ODBII to RS232. You can pick from a MAX232-based solution or an op-amp based approach. Use a solderless breadboard if you're not good at soldering or to test the circuit first to make sure it'll work for your car.
http://prj.perquin.com/obdii/ another ODBII to RS232 using just a few transistors.
For a slightly safer approach using optoisolators, try http://www.planetfall.com/cms/content/opendiag-obd-ii-schematics-pcb-layout

With freediag, you should be good to go.

What Dave said: "Open the pod bay doors, HAL."
What HAL heard: "Open the hot babe pornz, HAL."

HAL's speech recognition and morality programming* combined to give the famous reply, "I'm sorry, Dave. I'm afraid I can't do that." HAL knew certain things would have been too titillating to an all-ages film audience in 1968.

* Only for the film version. In the book version, it would have caused undue frustration to the reader, unable to see what Bowman was viewing. In that case, it was HAL's etiquette programming.

I'm leaving two free variables: what keep-the-beat task to use, and what syncopation/accentuation task to use. I'm expecting the drummer to be able to freely switch among many combinations on the fly, which is how I'm requiring it to be two tasks, and not just another learned merged-task. Of course, if you learn a particular combination of those two (e.g. for a particular song), then, yep, it becomes one task. Ask any drummer that sings and plays drums at the same time if he/she can sing the same part overtop of a different, arbitrary combination of keeping-the-beat and syncopation/accentuation, and you'll see him/her trying to focus on three tasks at once.

With practice, singing and drumming through a particular song will because (mostly) an isolated, single task---that's the point where nothing bizarre occurring onstage will throw the drummer off from singing or drumming, unless he/she is experimenting with a task combination that hasn't been yet learned as a single task (singing a different harmony part when another vocalist in the band is out sick, for example).

When you practice doing two tasks simultaneously, they become a new, single task. How do good drummers play syncopated beats? You learn a multitude of "keeping the beat" tasks involving many combinations of common patterns on the bass drum, hi-hat, and ride cymbal, then you learn a variety of syncopated beat tasks to play "overtop" of the other task. (You also have to learn strategies for performing these tasks at the same time, especially when you have to borrow a foot or hand from the keeping-the-beat task for an accentuated part and then un-borrow it; however, my point---namely, for a good drummer, many complex patterns involving multiple limbs, when practiced sufficiently, become simply "one task"---still stands.)

Oooh! I just had a *brilliant* idea!
Instead of protecting the driver by putting several feet of car engine between him and the front of the vehicle, why not put the driver out *right at the front*, in the position of vulnerability, and split up the engine into halves, sticking either half on the left and right side of the car? That way, the driver will surely be careful to look where he is going, but if someone runs the red light and T-bones him, he'll be okay.

Now, who wants to lend me some $ so I can file some patents?

I'm not sure about the wireless hacking from a laptop mentioned in TFS, but, as far as RF transmissions, these things can generate plenty of spread-spectrum modulation EMF when modulating the 240kHz signal carrier on wire.
There's a good discussion about eliminating ground loops so as to avoid broadcasting the signal as a source of interference at the Technical Library; I suppose one could always use an induction receiver to go the other direction, using a loop antenna. Obviously, modification of the above designs is needed for target frequency band. AM radio circuits might be a good place to start, too.
Actually, there are tons of good MW box loop designs that already go well below 240kHz; that page includes a calculator, and playing with some quick numbers suggests a 48cmX65cm frame [=56.5cm side length] for a 16-turn coil extending 21cm in length in parallel with four 470pF caps gives us resonance at 245kHz. Of course, with 20% tolerance ceramic discs, you may want to replace one of the 470's with a 4-40pF variable cap in parallel with anywhere from a 150pF to a 39pF paralleled with a 560pF, depending on how low or high the 470's are measuring.

[Disclaimer: I am an RF amateur.]

Computer memory, in an abstract sense, tends to be looked at in a hierarchical way:
* Registers
* Caches
* RAM
* Secondary storage (swap)

A filesystem is a datastructure, arguably just nominally imposed on a dedicated swap-space of sorts.
When you buy a gig of RAM, you expect 2^30 bytes, not 10^9 bytes. I've never understood why HD think that their "secondary storage" does not belong under the paradigm of "computer memory" when talking about sizing, despite the fact that all modern OS's use swap space, and filesystems are all data structures whose constituents tend to fall on word boundaries.