Slashdot Mirror
Big Data Knows When You Are About To Quit Your Job
HughPickens.com writes Quentin Hardy reports at the NYT that a leading maker of cloud-based software for running corporate human resources and financial operations has announced new products that provide the kind of data analysis that Netflix uses to recommend movies, LinkedIn has to suggest people you might know, or Facebook needs to put a likely ad in front of you. One version of the software, called Insight Applications, predicts which high-performing employees are likely to leave a company in the next year; it then offers possible actions (more money, new job) that might make them stay. In another instance, expense reporting software can predict which employee populations are most likely to exceed their budgets. "We've applied machine learning to affect consumer tastes," says Mohammad Sabah, director of data science at Workday. "Putting it to career choices, to pay and employment, have a huge upside if we do it right." Already, Sabah says, "we're surprised how accurately we can predict someone will leave a job." The goal is to predict future business outcomes to take advantage of opportunities and cut risk levels. One future product may be the ability to predict who will and won't make their sales quotas, and suggest who should be hired to improve the outcome. "Making an employee happy, improving the efficiency of a company these are hard problems that affect corporations."
When managers and supervisors know you're probably going to quit, it's a different story from "I hear he's gotten an offer from other companies". Data like this gets trusted implicitly, and if you weren't planning to leave this year, your new and improved toxic environment will make damn sure of it.
More likely though this software will be used to maximize everyone's ability to treat low level employees like machines.
Dont asks silly questions that you dont really want to know the answer to-- That's been one of the major reasons why HR drones have been looking up facebook accounts for many years now-- To see if you are naughty or nice.
So, YES. Big data knows that.
Big data knows who you voted for.
Big data knows what kind of hamburger you get from McDonalds.
Big data knows what fragrance your girlfriend/wife wears.
THAT IS THE POINT OF BIG DATA.
Big data takes shit loads of seemingly unrelated bits of information that people foolishly air in public, cross-references it, then uses it to make correlation based predictions.
Personally, I am opposed to the very idea behind big data.
(Then again, I harbor these "quaint" notions that things need to be allowed to be kept private.)
Well, if we're replacing all resources for machines, we will have machines hiring, firing, promoting and fluffing other machines.
On a thinly related topic, I still struggle with the concept of automation of work creating poverty instead of wealth. Imagine an alien being coming to Earth and saying
"We decided to make contact because you finally achieved the milestone of eliminating the need to work."
To which the humanity replies:
"Yep, we're all jobless, poor and hungry now."
It's as much about cookies as it is about ip address. Just so I understand your position(speaking as someone who buys prepaid cell phone/sim card/and minutes in cash):
It is voluntary because it doesn't effect you if you:
-don't subscribe to tv
-log in to any website with user generated content(almost all of them have tor exit nodes banned)
-use incognito mode + ghostery to consume all streaming media
-make all purchases in cash(forgoing all the efficiency benefits of online shopping)
In a similar vein:
DUI checkpoints are voluntary because you consent to them when you get in your car and drive somewhere.
Property taxes are voluntary because you don't live in a cave in the woods.
Taxes are voluntary because you have to have a job and use money for them to apply to you.
Their algorithm is pretty obvious just looking at their screenshot.
Analysis dimensions of significance:
1. Number of roles in the company: people with more roles are likely less assertive about what their job responsibilities should be and have less self confidence to refuse unwanted assignments. They are eager to please suggesting engagement in the company culture and optimism about future outcomes. Most importantly: they cannot simply go across the street to any old business because the niche they fill internal to the company seems to be a complex balance of complementary skills. While this makes them difficult to replace, it also limits the transferability of their skills to other employers. The longer they have allowed themselves to stagnate in this "devops" type jack of all trades role: the more difficult a departure will be. Their salary increases will not keep pace with their peers who do job-hop but by the time they notice they will already be a captive cog in the machine. Their replacement cost is likely manageable simply because the skills they posses are likely to be easily learned on the job, and their mastery of them is probably far down the tail end of diminishing returns, where the majority of the expertise "value-added" can be quickly gained through field promotions.
2. Tenure: They can treat all employees as an equal flight risk according to this dimension. They have a probability distribution for leaving the company after X years. The first 2-3 years they are a flight risk, but people who last longer than that probably have risk adverse personalities or have gotten comfortable.
3. Market Demand: This dimensions is based on job title. Data sources such as Craigslist job listings per month or BLS "growth statistics" are used to give a weighting factor based on the employees job title.
4. "How underpaid are they?": This dimension isn't explicitly shown on the website but it's obvious they would take consideration of the nice data on Salary.com or Glassdoor to figure out how appropriate an employees salary is. Below market rate salary translates to high risk of departure obviously.
5. "Time between promotions/pay-raises": This metric might as well be called "internal politics popularity contest metric" because it's simply a course granularity measurement for how appreciated or validated they feel. If they haven't gotten a promotion or pay raise recently: they are more likely to be thinking about moving on to greener pastures. Duh.
6. "Performance": I suspect this is actually a "feeder" data source for "Market Demand" as much as it is a measurement of which employees should be a priority to retain. That's the unwritten subtext to this entire thing though -> They talk about retaining the top performers but the side effect of this will be investing less money on retaining low performing employees. Their goal is going to be to keep the top performers out of the red, and keep the company at a nice uniform "yellow" with nobody dwelling in the "overcompensated-green-land".
Gaming the system then becomes obvious: find out what metrics are being used to drive this sterilized spin on "Rank and Yank" and then get even more catty and political to abuse your coworkers in to leaving your little office-"Survivor island". You make your metrics better(relatively speaking) by undermining the performance ranking of those around you via neglecting teamwork related activities that help your coworkers be successful at their activities in favor of self-aggrandizing individual promotion.
I'm guessing that their "algorithm" is more like:
X% change jobs after 1 year.
Y% change jobs after 2 years.
Z% change jobs after 3 years.
etc.
In my experience, people tend to change jobs because of something happening at their current job (or a personal/family situation change). And that's not something that can be predicted with any degree of accuracy.
LinkedIn has $5,312 revenue per employee. That's revenue, not profit. So giving a $10,000 raise to an employee means that employee puts the company in the negative.
If they had a revenue of $5k per employee they'd be bankrupt. They've got a profit of about that. From http://investors.linkedin.com/financials-statements.cfm the revenue in 2013 was about $1.5 billion, profit about $27 million. From http://press.linkedin.com/about they have about 6,000 employes. So profit of about $4,500 per head, revenue about $250,000
I'm always disappointed and not surprised by the points taken by debaters here on slashdot.
It's the same f*CK1ng thing every time...Black and White - either/or - up or down.
There is really no such thing as:
"I'm going to volunteer EVERYTHING and firehose my datasets everywhere..."
VS
"I live in a cave in the Yukon with a picture of the unibomber on the wall"
Give me a break people, WE ALL to varying degrees share things, whether with financial, government or online entities.
We've been doing it for years, just like here on slashdot.
I really get sick of the "Oh, you want privacy, what's wrong with you? Why aren't you posting to FB every ten minutes like "normal" people"
VS
"I change my route to work every other day, only pay for food with gold bullion and hack my neighbors wifi to connect to Tor" etc;
We can pretty much assume that groups like the NSA, with their vast resources and ability to pretty much data mine whatever they want, have the goods on most Americans, and have the nefarious ability to blackmail anyone they want. Why else would they datamine? Their mandate about terrorism is really just a facade for the concentration of power through knowledge.
With that being said, I sure as f*ck don't give out my info readily to Target, FB, Wal-Mart, or the vast array of businesses trying to datamine the f*ck out of every peon they sell a widget to.
It's bad enough knowing the NSA is doing that already.
We play the game with the bravery of being out of range
I'd like some software that will tell me how likely my company is to give me a raise vs tell me to get lost. This would be especially nice when it comes to companies I'm considering applying to.
As for the software under discussion, it is obvious that it will be used to deny raises to anyone except those who are both marked as willing to leave the company and whom the company sufficiently values (and by values I mean that it would cost the company more to replace them than to give them a raise).
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways