Science desk
< September 25	<< Aug \| September \| Oct >>	September 27 >

Welcome to the Wikipedia Science Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

September 26

Coilgun

What would be the ideal specifications of a railgun with a 'muzzle velocity' of 30 km s^-1 for an 600 Mg projectile? Coilgun, not a railgun. Plasmic Physics (talk) 02:21, 26 September 2011 (UTC)[reply]

Are there two separate questions here? Are you prepared for the Recoil associated by Newton's Law with the time derivative of the projectile momentum you describe? Here are ideal specifications: Railgun: Fits in pocket, uses ordinary AA batteries and is clearly marked "Warning: bullet comes out this end". Coilgun: Available in a range of pastel colours, fits in pocket, uses ordinary AA batteries and is clearly marked "Warning: bullet comes out this end". A page in Wikipedia gives many sources of information on railguns, among them is Count Zero by William Gibson that states: "You can rig a railgun to blow itself to plasma when it discharges." This danger should be addressed by an obligatory warning label. Cuddlyable3 (talk) 13:17, 26 September 2011 (UTC)[reply]

I mean, how can it be achieved using existing technology? Assume that recoil is accounted for, that the weapon operates in a vacuum, and that it is no longer than one kilometer in length. Plasmic Physics (talk) 13:34, 26 September 2011 (UTC)[reply]

You would need to be very specific on what you mean by ideal. Engineering is all about compromises. If you want it stronger, it will either be more expensive, or weigh more is a common trade off, so without knowing exactly what you want, specifications are quite impossible to provide. Also, I would be surprised if many here are deeply involved in the finer points to coil gun engineering, so even with a great list of specs you need, we probably can't help all that much. Googlemeister (talk) 14:00, 26 September 2011 (UTC)[reply]

OK, I'll see how far I can get. It is a naval artillery piece, opperating in a vacuum and zerogravity, it needs to be energy efficient, easy to repair, prefferably not self-destruct on firing. Cost and weight is not not an issue, assume infinite construction supplies, so the design can be as exotic. Plasmic Physics (talk) 14:14, 26 September 2011 (UTC)[reply]

I'm still baffled by what kind of answer you are seeking. It seems the specifications of this weapon are that if fires 600 metric ton or possibly .6 gram (big difference there between mg and Mg) projectiles at 30 km/sec, operates in a zero-g vacuum, and is around 1 km long. The caveat about existing technology makes this simple, however. Such a device probably cannot be built at all with existing technology. For one thing, we don't have the technology to lift a single projectile into orbit (the closest source of zero gravity and vacuum), much less the firing hardware. We can't even begin to imagine the other engineering problems and technological solutions. We are talking about a device 1 billion times more powerful than any yet constructed. Your 1 kilometer storage space would have to contain a large scale nuclear powerplant just to charge whatever you used to store energy to fire the device.

If you have any followups, I can try to help more, but you should specify if you are talking about a railgun or a coilgun (and why do you care which?) and if you meant metric tons (Mg) or milligrams (mg). And I'm still struggling with what kind of information would constitute an answer.gnfnrf (talk) 17:43, 27 September 2011 (UTC)[reply]

Coilgun specifically, as the rails in a railgun firing at these energies, would turn to plasma on the 1^st firing. Eitherway, I've decided on a coilgun. And, megagrams nor miligrams. Assume the weapon is already in orbit, and it is powered by a nuclear fussion plant supplying all the energy it could need. Is it best to have multiple short coils, or few long coils? Would there be any advantage in using ferromagnetic liquids to better control the magnetic field evolution. How should the weapon be held together. This is the kind of information I need. Plasmic Physics (talk) 23:35, 27 September 2011 (UTC)[reply]

A gun that fires a 600-ton shell at a speed of 30 km/s in zero gravity will have a recoil that will certainly disturb the orbit of even the biggest battleship, very likely causing it to tumble out of control after the very first shot. I recommend you seriously consider reducing the caliber of your big cannon. 67.169.177.176 (talk) 23:47, 27 September 2011 (UTC)[reply]

Just to show what I mean: Suppose your battleship has a displacement of 45,000 tons (which is a pretty reasonable size), and you fire a 600-ton shell from one of the ship's main guns at a speed of 30 km/s. By Newton's third law, action = reaction, which (since your battleship is in orbit) translates to m₁v₁ = m₂v₂, or in this case m[shell]v[shell] = m[ship]v[ship]. So, when you fire that 600-ton shell with a speed of 30,000 m/s, your battleship will start going the other way at 600*30,000/45,000 = 400 m/s! And what's worse, this delta-v of 400 m/s will occur in a tiny fraction of a second, which will subject the ship (and its entire crew) to a sudden acceleration on the order of several hundred g's! Definitely a very bad idea. (Besides, there's no conceivable purpose for firing conventional munitions of this size -- any target requiring such massive firepower is better dealt with by using nuclear weapons.) 67.169.177.176 (talk) 05:56, 28 September 2011 (UTC)[reply]

Cruiser not battle ship. The cruiser is approximately 1.17 km long and has mass of over a 100 Gg. Is it possible for the cruiser to redirect the recoil? Have the gun recoil into a hydrolic chamber, combine this with explosive retrothrusters (a thruster that produces an instantaneous directed explosive force). Plasmic Physics (talk) 07:20, 28 September 2011 (UTC)[reply]

There is no way to disperse the momentum transfer of a shot like this. Shock absorbers disperse kinetic energy, but momentum is always conserved. Now, your ship is about twice as massive as the estimate above, but his point about the acceleration due to firing being dangerous is still valid. To reach 30 km/sec in 1 km, the projectile needs to accelerate at ~45,000 g's for 1/15th of a second. The ship would be accelerated to 180 m/sec in the same time, for a force of 275 gs. You could try to spread out the time by giving the weapon some recoil within the ship, but with only 170 extra meters to work with, you can't spread the recoil out even over a whole second. In the fist 1/15th of a second, the weapon recoils freely within the ship and moves 90 m. It is then slowed evenly over the next 80 m, which takes 8/9 of a second, and still causes over 20 gs of force on the ship. You can't just say "they'll use thrusters to counteract it", either. In the first case, the thrusters need to be exactly as powerful as the weapon. Even in the second case, they need to accelerate a 100000 metric ton object at 20 gs to counter the force of firing, which comes out to something like 20 billion newtons. This is the thrust of 588 Saturn V rockets.

The point of all this, and I do have one, is that this is science fiction. Nothing like the device you describe can possibly exist given present or plausibly extrapolated technology. So there is no point in asking for how the engineering details would be worked out, because they can't.

Most of the more extravagant problems of this device are in the mass of the projectile. You have a muzzle energy of 2.7e14 joules; this is about the same as a 60 kiloton nuclear bomb. So, if you used an actual nuclear bomb, you could delivery many many times the energy, even if it would be more widely dispersed. But lets say for some reason you are stuck on an inert kinetic impact weapon. Reduce the shell weight by a factor of 100, make it 6 metric tons. To keep the impact energy, it has to go only 10 times faster, at 300 km/sec. The shell is and the weapon are subject to more accelerating stress, but if we can imagine handling 45,000 gs, why not imagine handing 450,000? The good news is that the momentum transfer falls way off. You are down to a 10th of the gs, which still requires some recoil space to dissipate, but becomes survivable.

Now, here's the key. Make the engineering details up. Do ferromagnetic liquids sound cool to you? Then go for it. There is no right answer, because nobody knows how to do it. gnfnrf (talk) 04:12, 29 September 2011 (UTC)[reply]

I need a weapon can compensate for very large target distance. Nuclear weapons in this case, is only lethal on direct contact, so is too expensive compared. Plasmic Physics (talk) 07:48, 28 September 2011 (UTC)[reply]

Each "bullet" is 0.6% of the mass of the ship. Seems like you might have trouble if you get into a sustained firefight. Googlemeister (talk) 13:42, 28 September 2011 (UTC)[reply]

It is not really the kind of weapon to get into a firefight with, it is a rock to kill an ant, a final resolution. It is targeted by an A.I., requiring confirmation by the person on command. You tend to use it only if you're certain that it will find the target. Plasmic Physics (talk) 21:45, 28 September 2011 (UTC)[reply]

Even if the cruiser displaces 100,000 tons (which is an appropriate size for an aircraft carrier, not a cruiser -- most cruisers displace 10,000 to 20,000 tons), the recoil will still cause a delta-v of 180 m/s, which is just as unacceptable as 400 m/s. What makes you think that a nuclear weapon is only lethal on direct contact? The thermal pulse alone will set most targets on fire within a radius of at least several miles. 67.169.177.176 (talk) 23:56, 28 September 2011 (UTC)[reply]

Also, the "hydrolic" (which BTW is not a word) chambers on ordinary (non-space) artillery pieces only work because the gun is set into the ground, or the ship on which it is mounted is floating in the water; i.e. they depend on the resistance of the ground or water to actually absorb and dissipate the recoil. Since your jumbo-cruiser is in space, the only resistance it offers to the recoil is that of its own weight, which is completely insufficient even with "hydrolic" chambers. As for explosive retrothrusters, they will have to fire AT THE EXACT SAME INSTANT as the main guns AND also create nearly the same amount of thrust as the gun. Even if such a massive explosive force could be assembled AND detonated instantaneously at precisely the right instant (which is practically impossible), such a tremendous explosion close aboard would most likely blow a big hole in the main armor, or worse. 67.169.177.176 (talk) 01:55, 29 September 2011 (UTC)[reply]

And that, is why I'm asking for solutions to problems such as these. This ship is classed as a light cruiser. According to the system used, a carrier is typically on the order of 3000 meters long. With 2 meter thick hull armour, a thermal pulse will only do superficial damage. I'd be more worried about the gamma pulse. Plasmic Physics (talk) 02:54, 29 September 2011 (UTC)[reply]

In this case, here's my answer to your question: A coilgun with the specified caliber and muzzle velocity is WAY too powerful for the ship's size, and will be a danger to the ship and crew. Nor do I see any need for an artillery piece of this size. You'd be much better off having the ship armed with much smaller artillery pieces (perhaps firing only 1-ton shells instead of 600-ton -- which would still be A LOT of firepower by anyone's standards), possibly with nuclear shells for the largest, most hardened targets as well as conventional ones for most other targets. 67.169.177.176 (talk) 05:02, 29 September 2011 (UTC)[reply]

Hmmm, well, thanks anyways. Plasmic Physics (talk) 19:05, 29 September 2011 (UTC)[reply]

Why not accelerate the projectile after launch? Nuclear pulse propulsion, Pulsed plasma thrusters, Variable Specific Impulse Magnetoplasma Rockets, Ion thrusters could be used, depending on the distance to the target; energy could come from nuclear reactors or via laser from the launch vehicle. Endless posibilities DS Belgium (talk) 01:03, 1 October 2011 (UTC)[reply]

Not the idea. Plasmic Physics (talk) 04:37, 1 October 2011 (UTC)[reply]

conciousness

I need to store my conciousness in the computer so my family can have acess to me after I die. Are there any software programs that can do this? — Preceding unsigned comment added by 78.148.142.217 (talk) 11:13, 26 September 2011 (UTC)[reply]

There is work being done in areas that would seem to relate to the question you pose, as reported here and here, but the results do not seem to be ready for prime time. Bus stop (talk) 11:34, 26 September 2011 (UTC)[reply]

Those studies are just about recording visual images you are currently thinking about. Next they need to be able to scan visual images stored in your memory, then non-visual images, then they need to read your personality, then they need to put it all together to form a consciousness. Each of those steps is a massive leap. So, there's a long way to go, yet. StuRat (talk) 12:06, 26 September 2011 (UTC)[reply]

They aren't even quite that, yet. What they are matching bloodflow patterns in your visual cortex to a library of bloodflow patterns from other visual cortexes. Which is pretty damn cool and just the tip of the iceberg, to be sure. But they aren't really about recording visual images — they're looking at patterns and finding ways to correlate those to other patterns, which through very clever work actually corresponds in a recognizable way to images actually seen. But they aren't really able to show what the visual cortex "sees" in its raw form, and it's not clear that it even works for things you are "currently thinking about," as opposed to "stuff your eyes actually are seeing right then." --Mr.98 (talk) 14:20, 26 September 2011 (UTC)[reply]

This might sound dumb, but is that kind of like how sound can be translated into little electromagnetic signals and then the magnets on a speaker can recreate the vibrations? Sir William Matthew Flinders Petrie | ^{Say Shalom!} 27 Elul 5771 15:45, 26 September 2011 (UTC)[reply]

Only if you consider that in both cases something complex is broken down into small, manageable units, but this is true of many tasks. StuRat (talk) 20:35, 27 September 2011 (UTC)[reply]

When do you need it by? We're not quite at the singularity yet, but give it a few decades and who knows. We'll even throw in a jet pack with it, for you. --Mr.98 (talk) 12:01, 26 September 2011 (UTC)[reply]

Actually, we do have jet packs now, it's just that they are rather dangerous and run out of fuel way too quickly to be useful. StuRat (talk) 12:08, 26 September 2011 (UTC)[reply]

It was meant to be a wry comment on the value of predictions about future technology. Just because something is technologically feasible does not mean it actually will be adopted. --Mr.98 (talk) 14:17, 26 September 2011 (UTC)[reply]

I'll take two please. And a death ray.-- Obsidi♠n Soul 12:09, 26 September 2011 (UTC)[reply]

For something less ambitious, how about recording video clips containing your thoughts on a variety of topics, and indexing them on your computer, so your family can access those clips whenever they wish ? StuRat (talk) 12:12, 26 September 2011 (UTC)[reply]

A single copy of your software for backup purpose is allowed but you may not distribute copies yourself of yourself. Before you go to immortal dataspace please install Skype so that one can still converse with you posthumously. Will Mr. Magill in Manchester grant an exclusive interview to this uplifting magazine ? Cuddlyable3 (talk) 12:43, 26 September 2011 (UTC)[reply]

Sadly, even if the software did exist, (and it doesn't), your desktop PC likely lacks sufficient memory to store your consciousness with full fidelity. Also, I suspect most humans last longer then most desktop PCs, and most hard drives contained therein. Googlemeister (talk) 13:56, 26 September 2011 (UTC)[reply]

I thought I read somewhere you would need 20 TB of HDD space to store a human conciousness, though I have no idea how they came up with it. Sir William Matthew Flinders Petrie | ^{Say Shalom!} 27 Elul 5771 15:45, 26 September 2011 (UTC)[reply]

The human brain has 10¹¹ neurons in it — which is 1000GB worth if each neuron were just a byte. But what's really important are the connections between neurons — there are about 10¹⁴ to 10¹⁵ of those (or, if every connection was 1 byte, between 100TB and 1PB). (In these byte estimates, I am using true GB, TB, and PB definitions, not GiB, TiB, or PiB). So I don't know where 20TB would come from, as a number. All of this makes a lot of arbitrary and no doubt incorrect assumptions though about how much data this technology would require. Assuming these were just stored as fairly simple numbers, you could probably achieve extremely high levels of compression... but anyway, these numbers are kind of meaningless except as an order of magnitude estimate of how complex the human brain is as an organ. (Numbers from Neuron#Connectivity.)--Mr.98 (talk) 20:49, 26 September 2011 (UTC)[reply]

I doubt that the question was serious, but in any case the most appropriate answer is a pointer to our article on mind uploading. Looie496 (talk) 15:04, 26 September 2011 (UTC)[reply]

In case it was, has the author thought about how unhappy such an existence would be? It would effectively be like living as a quadraplegic, able to see the world in front of you, but not actually go into it. Of course by the time the tech is available, advances in robotics might make my arguement null. Sir William Matthew Flinders Petrie | ^{Say Shalom!} 27 Elul 5771 15:45, 26 September 2011 (UTC)[reply]

Well, the robotics needed is far beyond the current state of the art, but the technology needed for mind uploading is quite a bit farther beyond the current state of the art. Looie496 (talk) 17:37, 26 September 2011 (UTC)[reply]

Believe it or not you can be simulated by a variety of methods of which neural networking is the current favorite used by IBM's Watson. Just restrict the input data to all of your past and present thoughts (rules) and Watson or its equivalent will use neural networking to simulate you. My personal favorite in terms of strictly logical thinking. --DeeperQA (talk) 19:04, 26 September 2011 (UTC)[reply]

Sigh. None of the previous paragraph is correct. That is a total misrepresentation of the Watson project; and the article you linked to is dumb. Comet Tuttle (talk) 19:24, 26 September 2011 (UTC)[reply]

IBM Watson does not use a neural network. IBM Watson uses a large, eventually consistent database and advanced natural language processing, as well as massively parallel computing. A neural network is a programming-paradigm for solving specific types of numerical math problems. It's actually a very poor algorithm, in my opinion, compared to formal numerical optimization. Neural networks have an amazing tendency toward system-instability and are not very resilient at escaping from local minima. The term "neural network" has been coopted by various science-fiction writers who think it "sounds cool;" in science fiction, it is used to describe "any type of smart computer software." In fact, a neural network is just a specific technique for ~~solving~~ estimating solutions for matrix algebra. User:DeeperQA, most of what you wrote above is incorrect. You might want to read the Watson FAQ. Watson was never intended to "simulate" a human. It is intended to "greatly improve information seeking tasks" and "help make computers more effective at communicating in human terms." Nimur (talk) 21:43, 26 September 2011 (UTC)[reply]

Where did you get the idea that IBM's Watson used neural networking? Or even that Neural Networks were the "Current favorite"? I thought they mostly passed out of fashion in the late eighties. I think nowadays they're mostly the "Current favorite" for High School level AI projects. (You know, science fair projects titled "Can a Computer be Taught to Recognize Faces?".)

Also, that article you linked is horrible. I wish I knew who wrote it so I could never read anything by him or her again. APL (talk) 21:53, 26 September 2011 (UTC)[reply]

Heh, I recognize the author's email addy; it's the handle of a years-absent Ref Desk "memorable eccentric". — Lomn 22:25, 26 September 2011 (UTC) [reply]

This only serves to further my theory that there are only about twelve actual people on the internet; four of them are reference-desk trolls. Nimur (talk) 22:32, 26 September 2011 (UTC)[reply]

...and I'm really a dog. HiLo48 (talk) 23:00, 26 September 2011 (UTC) [reply]

The question looks like an LC item. In any case, he's alluding to what Jor-El did in the first Superman film. Kryptonian technology was rather more advanced than ours (although their common sense apparently wasn't). Another approach could be the way Spock copied the essence of his being to Bones in Star Trek II. ←Baseball Bugs ^{What's up, Doc?} carrots→ 01:56, 27 September 2011 (UTC)[reply]

I wouldn't say 'years absent'. They've been coming back to the RD with accounts and IPs every so often usually getting blocked when they went too far. The most recent account should be obvious from their promotion of their fairly obscure (i.e. if you see someone promoting it you have to wonder who that person is) work both above and on their talk page (which I noticed a few weeks ago) as well as the similarity of intererests and views expressed on the RD. Nil Einne (talk) 02:55, 27 September 2011 (UTC)[reply]

We'll only have access to this technology for a very brief time, if at all, see here why. Count Iblis (talk) 23:35, 26 September 2011 (UTC)[reply]

Hang on, read the Consciousness article on this very site. Capturing a human conscience is not simply a matter of recording their memory - you would have to capture and replicate all of the operational aspects of their brain AND the input and output routes that helped to form and will be needed in order to continue to form the structure and experience of the concsiousness i.e. connecting the replicated consciousness to the outside world via an accurate simulation of a human with a body, eyes, ears etc.. otherwise the 'person' wouldn't respond or react to the world or experience it in a manner equal to the way they did in their original body - crucial to get right if you want an accurate simulation of a person and all their characteristics. I guess you could modify the programme to take account of the lack of a real body, but you'd be massively limiting the realism of the simulation. In short to accurately reproduce a human mind, I suspect that you would need to accurately reproduce an entire human being. — Preceding unsigned comment added by 213.120.209.210 (talk) 15:34, 27 September 2011 (UTC)[reply]

Cryonics could be used to preserve the information in your brain until it could be uploaded or repaired. — DanielLC 23:38, 27 September 2011 (UTC)[reply]

Maximum number of options from which a human can make a rational decision on optimal choice

What does research indicate is the largest number of applicants a single hiring manager could rationally consider and deliberately pick a single optimal final choice from without resorting to something arbitrary yet decisive such as pulling a resume out of the stack of equally acceptable applicants and being done with it, thereby reducing the winner's attribution of success to pure luck? 20.137.18.50 (talk) 14:31, 26 September 2011 (UTC)[reply]

One. (Sorry.) Looie496 (talk) 15:03, 26 September 2011 (UTC)[reply]

There's the well-known paper, "The Magical Number Seven, Plus or Minus Two", which claims 7±2 as the number of items the average person can usefully consider simultaneously. That said, I see no inherent reason that a hiring manager couldn't break an arbitrarily large applicant pool into some sort of bracket-style process. That might not get the "best" candidate at the end, for certain values of "best", but it's certainly not success by "pure luck". — Lomn 15:04, 26 September 2011 (UTC)[reply]

You've got 375 resumes on the table and your superior is demanding that you fill the position by COB this Friday. 20.137.18.50 (talk) 15:14, 26 September 2011 (UTC)[reply]

You didn't say anything about time pressure in your original question. —Bkell (talk) 15:20, 26 September 2011 (UTC)[reply]

I would suspect that a hiring manager not subject to any Workplace stress in his/her position in the real would would be like unto a Spherical cow. There are many things about life that inhibit ideal behavior. I wanted to know the usual result of how successful the optimization effort is after taking the real world into account (which has such pressures and more pretty consistently). 20.137.18.50 (talk) 15:36, 26 September 2011 (UTC)[reply]

(ec) With 375 resumes, I would first separate them into 7 +/- 2 piles, eliminate some of those, subdivide, and repeat. -GTBacchus^(talk) 15:38, 26 September 2011 (UTC)[reply]

375 resumes and a week? No problem, at least for your statement. Divide and conquer. Interviews of finalists the last couple of days. I'll repeat: "might not get the 'best' candidate at the end, for certain values of 'best', but it's certainly not success by 'pure luck'". — Lomn 15:40, 26 September 2011 (UTC)[reply]

Would your grouping not be simply mechanical "pulling out of the stack," making it a matter of luck whether someone was in the group that got picked all the way to the finals, making the finalists at least lucky that they got there? To elaborate, imagine the question: Why did you initially put applicant #253 in group number two? Did you have a rational reason for doing so? What did he/she have in common with the other 53 people in group number two and not have in common with everyone not put in group number two, and what made you eliminate group two? Was it the aspect that all of them had in common (if indeed there was one)? In other words, what's the difference between your grouping game and just picking seven resumes off the top or bottom or middle? If one of the applicants that was eliminated for being in the wrong 53-person group upon your first elimination was clearly better upon hearing him/her in an actual interview than any of the other finalists, would not that be very lucky for the finalists that he/she wasn't there? This could be the case for any of the 368 humans you didn't see that got eliminated along the way. 20.137.18.50 (talk) 15:48, 26 September 2011 (UTC)[reply]

You seem to have confused my proposal with GTBacchus'. His appears to arbitrarily eliminate groupings. I'm saying "grab 5 resumes, set aside the best, discard the others. Proceed thusly through all 358 resumes. Now return to your set-aside stack of 70 and repeat. Interview the appropriately small subset of finalists." This is necessarily a rough grading (thus "for certain values of 'best'"), but it is not random selection nor significantly luck-driven. — Lomn 16:32, 26 September 2011 (UTC)[reply]

What kind of terrible HR person would do anything arbitrary? Allow me to specify more detail: I would separate them into groups according to some criterion (e.g., type of experience), and then make eliminations based on my professional judgment. Then I would choose a finer criterion, etc. Sheesh.

The disadvantage of your method is that the first 5 you grab might be the 5 best. The next 5 you grab might be the 5 worst. Why eliminate one of the best and keep one of the worst in the very first step? -GTBacchus^(talk) 17:09, 26 September 2011 (UTC)[reply]

The criterion of being a member of the ingroup of the hiring manager or having one of his or her trusted employees as a reference is a very fortunate "qualification." 20.137.18.50 (talk) 19:13, 26 September 2011 (UTC)[reply]

Secretary problem is about a different scenario, but may be vaguely relevant. AndrewWTaylor (talk) 16:24, 26 September 2011 (UTC)[reply]

There is a huge qualitative difference that is being overlooked here. One issue is the problem of selecting the optimal member from a large number of items in a group. A second issue is the problem of defining "optimal" by combining a large number of variables - some that are weakly defined - and trying to distill this into a single value suitable for strict "better-or-worse" comparison. The former is a problem of scale, and is fairly trivial to deal with. The latter is a problem of dimensionality. Even with a powerful computer, a high-dimensional optimization problem is difficult to solve; a human may sometimes outperform a digital computer on such problems by using heuristics to sparsify the search-space.

Formally, if the set is comparable, and follows a strict total ordering rule, or at least partial ordering, it is trivial for a human to optimize the selection. The most appropriate technique would be the manual application of a sorting algorithm; the best algorithm to choose depends on the size of the set and the expected distribution. There will be a time-vs.-space tradeoff for each algorithm, and a tradeoff between "best-case," "average-case" and "worst-case" performance. The use of a sorting machine or a digital computer will invariably speed up this process, but is not required. A human can run quicksort on a pile of paper resumes, if they know how to do it. (Use selection sort if you have a small desk; use quicksort if you only have ten minutes until your next meeting).

If the set is highly dimensional and is not strictly orderable, the human will be trying to solve a high-dimensional nonlinear optimization problem; human brains and modern computers are poorly equipped to solve these problems in the general case. Resume-sorting is a good example: there's not a "quality number" printed on top of each resume; the human must estimate whether Resume #3,041 is "better" than #3,042 by applying heuristic interpretations of the resume content. The assumption that there is an optimal choice is formally an assumption that some total-ordering metric can distill a one-dimensional objective function out of the problem, and then we can try to find its "best value." Nimur (talk) 17:31, 26 September 2011 (UTC)[reply]

The Paradox of Choice is about consumer decisions, but this is what I was thinking of as I asked the question. The hiring manager is trying to "buy" the best employee for his (company's) dollar. 20.137.18.50 (talk) 17:49, 26 September 2011 (UTC)[reply]

Hopefully he plans to pay for an employee and isn't planning on forcing free labor! -- kainaw ™ 17:59, 26 September 2011 (UTC)[reply]

That is pretty much how I do a lot of things. For this, I'd make two piles: Above average and below average. Then, I'd toss the below average in the garbage. I'd repeat again with Just Average, and Better than Average on the previous Above Average stack. I'd toss the lower group in the garbage and repeat. Eventually, my good stack will have a few resumes that I can look at in detail - even though I've already scanned them multiple times and know them pretty well. It probably make me comfortable to sort this way because it is just a binary tree sort, which is something I've worked with for about 30 years. -- kainaw ™ 17:45, 26 September 2011 (UTC)[reply]

Honestly what is going to happen is that of the 375 applications, probably 200 will be at least qualified, and the top 50 will have only a negligible difference in quality. Thus mix the pile of applications and take the first 50 from you new mixed pile. Toss the rest. This will have the added advantage of separating the lucky from the unlucky applicants, since you don't need unlucky people in your company. Divide that group into qualified vs unqualified. Toss the unqualified. You should be down to around 30 qualified applicants. Now go through in detail in groups of 10 and keep very highly qualified, and toss those who are barely qualified. Interview these remaining applicants and have a nice week. Googlemeister (talk) 18:22, 26 September 2011 (UTC)[reply]

..."the top 50 will have only a negligible difference in quality" - this is only applicable to some scenarios. Nimur (talk) 19:15, 26 September 2011 (UTC)[reply]

Simply use a search engine to find all of the applicants that included all of the attributes you desire. If the results are greater than one add another quality until you find "The One". --DeeperQA (talk) 19:21, 26 September 2011 (UTC)[reply]

There is, of course, no guarantee that any particular combination of search criteria will result in precisely one hit. There's also the practical matter of what kind of search engine to use, and the question of applying judgment to individual cases. Can search engines screen for everything that we expect HR professionals to understand? Why not replace them all with robots, then? -GTBacchus^(talk) 00:59, 27 September 2011 (UTC)[reply]

Some writers start by trying to memorize every word in the dictionary and to that end playing crossword puzzles and scrabble. The psychologist's goal as a scientist is to form words and phrases that describe the most esoteric of conditions. All that is required to automate the process of identification by search engine is to use these words as keywords and to use them in the personnel database(s) to be searched. The key is to use all of them by either stating that they define or do not define the entry. The creating of new words, phrases and phrase combinations may one day be the privy of computers but for now it is still pretty much limited to the privy of human beings. See Optimal Classification--DeeperQA (talk) 03:34, 28 September 2011 (UTC)[reply]

And what guarantees that some combination of criteria gives you exactly one hit? What if your first 7 criteria narrow the field to 5 candidates, and then every other criterion you try next cuts it to zero. Then what?

Also, what do you say to the HR person who doesn't possess this database and search engine technology? -GTBacchus^(talk) 01:55, 29 September 2011 (UTC)[reply]