News for hackers. Stuff that matters.
Re:Google Calculating Which Employees Are About to Quit by dxjones on Tuesday May 19, 2009 @04:38PM (#617578)
Google is probably doing retrospective data mining to identify statistical links between possible indicators and actual data on employees leaving ... (e.g., employee tells boss, "I feel my talents are not being used", or increased frequency of visits to "monster.com")

However, once Google starts to intervene and change things so that the most valued employees change their mind and stay, ... then that changes the equation. Furthermore, once Googlers know they are being "watched" in this way, some may change their behaviour.

A small version of the Heisenberg uncertainty principle? ("observing something changes it").

By the way, Googlers might change their behaviour in two ways. If you know you are leaving, you may try to mask that intention. On the other hand, if you just want a raise, or more challenging work, you might mimic "I am about to quit" signals, hoping to manipulate the system to enhance your situation.

Re:Google Calculating Which Employees Are About to Quit by kscaldef on Tuesday May 19, 2009 @05:16PM (#617652)
> e.g., employee tells boss, "I feel my talents are not being used"

If you need a computer program to tell you this employee might quit, you're doing something wrong.

> On the other hand, if you just want a raise, or more challenging work, you might mimic "I am about to quit" signals, hoping to manipulate the system to enhance your situation.

Are you under the impression that's not already a common tactic?

Re:Google Calculating Which Employees Are About to Quit by sethg on Tuesday May 19, 2009 @09:02PM (#617921)
If employee tells boss "I feel my talents are not being used", the boss may have reasons not to pass that news up the chain to HR. Middle managers who are good at playing office politics don't like to reduce their own headcount.
Re:Google Calculating Which Employees Are About to Quit by oldgregg on Tuesday May 19, 2009 @07:37PM (#617821)
Metric #1: Reads Hackernews
Re:Google Calculating Which Employees Are About to Quit by chris11 on Tuesday May 19, 2009 @07:01PM (#617782)
I see your point that this can just be used to be making up for bad management. But this method will get more information than one person normally knows.

A boss might not know that the employee has recently requested a transfer that has been denied. Or he might not be aware that the employee has filed several complaints with h.r. Sure there are major privacy implications with what data is being analyzed, but data mining might be more accurate.

Re:Google Calculating Which Employees Are About to Quit by Confusion on Wednesday May 20, 2009 @03:57AM (#618248)
IAAP. This has nothing to do with the uncertainty principle, because your intepretation of the popular rendering of the uncertainty principle ("observing something changes it") is wrong. I grant that nearly everyone, including many physicists, think your interpretation is correct, but it is in fact wrong.

The uncertainty principle even holds when the measurement procedure could not possible transfer enough energy to the system being measured to change the outcome of the measurement. It is a fundamental property of the mathematics describing quantum phenomena, that does not just come about when factoring 'the act of measuring and the consequent perturbation of the system' into the equation. Measuring a system disturbs it, but that also holds in classical systems. Any first year physics student should realize that after his first few experiments and will be able to take it into account when presenting his results (for instance: a thermometer takes heat away from a system, because it heats up itself). However, you can often take this into account exactly or at least so precisely that it does not influence the overall uncertainty of your measurement (due to the thermometer having only a 0.1 degree measurement accuracy). This isn't true for uncertainties due to Heisenberg's principle: they are always there, always have the same relative magnitude and are independent of the kind of measurement. Taking measurements causes you to notice the Heisenberg uncertainty relation, but it does not cause that relation.

This Google thing is more like the point made in Frank Herbert's Dune series over and over again: if your predictions are taken seriously, then you cannot predict the future without causing it to come about. This doesn't hold for the carny kind of predictions, because most people don't believe in them. In does hold for Warren Buffet's kind of predictions, which is why he doesn't want people reading the annual report of his company.

Re:Google Calculating Which Employees Are About to Quit by litewulf on Tuesday May 19, 2009 @05:30PM (#617674)
(Sidenote: You probably want the observer's paradox, http://en.wikipedia.org/wiki/Observers_Paradox which is basically identical, but specfically coined because it is something that comes up in social science realms.)
Re:Google Calculating Which Employees Are About to Quit by gort on Tuesday May 19, 2009 @05:42PM (#617689)
Re:Google Calculating Which Employees Are About to Quit by benhoyt on Tuesday May 19, 2009 @08:09PM (#617859)
Just added his link as a WP redirect to your link, so they both work now. :-)
Re:Google Calculating Which Employees Are About to Quit by litewulf on Wednesday May 20, 2009 @04:36AM (#618270)
(For my own curiosity+better editing-ness, how did you do that?)

(And I swear the link looked fine when I typed that comment out. Not sure what happened. Thanks parent and grandparent!)

Re:Google Calculating Which Employees Are About to Quit by gort on Wednesday May 20, 2009 @05:57AM (#618301)
On Wikipedia you can create a "redirect" by making an article whose sole content is, say:

#redirect [[Pelagic armorhead]]

Re:Google Calculating Which Employees Are About to Quit by ruby_roo on Tuesday May 19, 2009 @05:59PM (#617709)
Heisenberg's Uncertaintly principle incentivizes zero transparency when studying human behavior, does it not?

That is rather frightening.