I keep reading that the big tech companies, most notably Google and Facebook, are finding that job performance isn't highly correlated to an employee's college grades or even the reputation of the school attended. And I also understand that tech companies are less inclined to ask interview questions such as "Why are manhole covers round?" Apparently the answers to those questions don't predict future employee success.

So now Google and Facebook and perhaps others are using secret new methods of collecting information on the Internet to identify great candidates they can poach from other companies.


Sorry, my bullshit detector just went off.

I think what is really going on is that employee success in the tech industry is most correlated with luck. But if you work in Human Resources, and your job involves identifying good employees before they do something great, you need some sort of flavorful bullshit to make it seem as if there is science to what you do. Whenever I hear that someone has a secret algorithm, or they discovered something while data mining, I get highly suspicious.

In my experience, people who managed to get good grades from prestigious schools are indeed far more effective than people who didn't. I expect a Stanford grad to do be smarter and more effective than a Chico State grad at least 80% of the time.

But there have also been studies showing that the worst kind of work group is one that has too many smart people. Ideally, you want one smart person and several competent followers on a team, or so the studies suggest. So it doesn't surprise me that Google or Facebook could be hiring geniuses and experiencing project gridlock as the brainiacs stand around arguing. So that might be the problem.

I wonder how anyone can identify a great employee working for another company when that employee has only worked on teams. Often it is the team dynamic, the timing of the project, the chemistry of the group, the effectiveness of management, and a hundred other factors that create success. Most of it looks like luck.

I can see how a "Moneyball" approach works in the limited case of baseball. A batter is a member of a team, but the team has little influence on how he hits. A player's batting average is all about his own skill. But how do you evaluate, for example, an employee whose every move is part of a larger collaborative effort? You can't moneyball that.

I think the secret sauce that makes some groups successful is the chemistry of the team, along with luck, of course. And, as I mentioned, good team chemistry might mean having one smart person and several followers. The problem is implementing that system. Could a manager really get away with organizing teams by brightness level? "Okay, team. Susan is the smart one and the rest of you are . . . the other ones. Go do something awesome."



Rank Up Rank Down Votes:  +97
  • Print
  • Share


Sort By:
+18 Rank Up Rank Down
Jul 11, 2013
I can still remember fighting with HR to hire a software developer from Starbucks. He didn't have a college degree, they wailed. We can't hire people like that!

People in HR use things like grade point average, and school, because it's all they have. If you were them, how would you filter people? I get it, but it doesn't make it right :-)

Well - Mr Starbucks was part of a team that was winning Hack-a-thon competitions. Pretty good indicator? Yeah.

Did I mention that is my favorite way to find talented people to add to my software teams? Forget the resume, go for the winners who can prove they can make something happen.

Successful? Very. That kid kicked the a** of people who got a 'proper' degree.

This notion that really smart people need to be surrounded by average people is also crap.

Two to three smart people in a room bounce ideas off of each other creating sparks and energy. I'd take 3 super men over 20 average men any day. Again, past experience. if you have a dumb guy, adding more dumb guys isn't going to make it better. It'll just make the dumb guy feel smarter.

Oh - one last thing: Predictive analytics just works. I did some work with some models with an insurance company, and was shocked by some of the indicators that turned out to be strongly predictive.

I'm not sure what indicators would work. But I know I can rank the developers in my organization from best to worst. Given the indicators, and that ranking, there's no reason someone can't create a predictive analytics engine that would do the trick.

Frankly, I can't believe Google hasn't done this yet. They might not be sharing - but I'm sure they've done it.
Jul 11, 2013
Batting average is not completely individual. The situation of the game affects what pitches you see and therefore can completely affect your average. If runners are on base, pitchers pitch from the stretch with less power, and you'll see better pitches. If you have a good hitter behind you, you'll see better pitches because they can't pitch around you, etc. There is a slew of cases where hitters' averages change by a lot year over year strictly because of lineup changes.
Jul 10, 2013
I think skill tests, like those mentioned in prior comments, can be very helpful for the narrow subset of jobs that are completely or almost completely skill driven. However, for "softer" occupations they alone won't give you a full answer, because they likely can't uncover personality traits, persuasive abilities, teamwork, etc.

That being said, I think the verbal interview is 100% useless for evaluating job candidates. I worked for a financial firm that relied entirely on resumes and interviews, and we'd often wind up with junior people who couldn't find their way around an income statement. The whole process drove me nuts.

However, I do believe that the recent increased use of sophisticated (and admittedly tedious) survey techniques for psychological/personality profiling can be very effective. It's the best way to try to figure out who the actual person is, as opposed to the performer sitting in front of you for a 30 minute interview. God forbid I ever had to take one of those things. They'd have figured out that I am the kind of employee who spends all day reading and commenting on Dilbert blogs.
+7 Rank Up Rank Down
Jul 10, 2013
Smart is a very broad term. Showboat smart? People smart? Organizational savvy? An effective manager of projects? An effective manager of people? Egotistical narcissist? All are smart, but who gets things done?

Imo it is possible to be absolutely shockingly brilliant ... but not be capable of getting oneself to work much less leading a project or a team. Raw brainpower is over-rated and does not necessarily make one a good team member or leader. Emotional stability is a good trait. Good work habits, consideration for one's teammates are plusses too. Attributes negatively correlated with very high IQ?

Is luck the magic sauce? Or is it a variety of mixes and approaches that are difficult to quantify and unlikely to be formulized. Good leadership is a constantly changing art seldom encountered. Witness all the fodder for Scott's regular ridicule of management. But an abundance of ridicule does not equal management expertise any more than any of the management guru books offer anything more than buzzwords, cultishness and fads.
+16 Rank Up Rank Down
Jul 10, 2013
<ksaliga>Our company gives perspective developer candidates an actual test</ksaliga>

I implemented that at my company. We have candidates fix a couple bugs and add a page to a simple example web app based on our production code. We found it amazing how some people sound great in the phone interview then can't debug simple code written in what they've professed to have 5 years experience in.
+15 Rank Up Rank Down
Jul 10, 2013
Our company gives perspective developer candidates an actual test, which could take upwards of two hours. The test was created by other proven developers on our staff, who have an interest in getting team members that can actually contribute. The test involves the candidate building a minor application based on a written spec sheet, and using specific methods as outlined by the project description. The talent we've hired using this method has all been very good.
Jul 10, 2013
Could anyone tell me why the RSS feed for strips does not deliver athe strip to my RSS reader?
From the Dilbert website:
They're here!
For all you RSS fans, tons of Dilbert goodies to fill your reader:
strips, Mashups more. Click below to feed!
Jul 10, 2013
Org charts are still used in an rigid, hierarchical manner. It seems that managers are experienced employees who moved up the ranks. There should really be big distinction between "project managers", "administrative managers", and "technical leads". Project and administrative managers need not have a technical background, though a 2-5 years is probably appropriate. The technical leads need to be very experienced employees, but they need to be okay with a 25 year old boss who is acting as a "servant leader", not a boss, but a resource to help others succeed.
Jul 10, 2013
...Am I the only one who thinks maybe the solution is for the smart one to be promoted to the manager position? Or is that mid-20th century thinking?
Get the new Dilbert app!
Old Dilbert Blog