Ungooglable interview questions (was: Re: New perl features?)

[Kieren Diment]

On 17/03/2013, at 8:24 PM, Abigail wrote:

> On Sun, Mar 17, 2013 at 01:54:37AM +0000, Sam Kington wrote:
>> 
>> At $WORK we ask new hires, over IRC (because everyone potentially works remotely), a variant of chromatic's http://modernperlbooks.com/mt/2011/01/how-to-identify-a-good-perl-programmer.html
>> 
>> So far, when we've put people through the test (some people have been omitted because they obviously had a clue), nobody has managed to identify * as a sigil.
> 
> Have you hired any of the people any of those people? If so, does this 
> question have any value? Does it really matter when someone during an
> interview can remember that * can be used as a sigil?
> 
> But loooking at chromatics list, does it actually work? Does it identify
> good Perl propgrammers? I guess it can be used to weed out people claiming
> to know Perl very well, but don't, but someone who can answer all the
> questions may still not be able to code his/her way out of a wet paper bag.
> 

I''m not sure I'd want to hire someone for a routine job who was keen to remember that * is a sigil.  If someone I know and trust uses it, then it's usually clear what they're doing with it without having to talk to them about it.  OTOH I doubt I'd want to hire a random off the street who was keen to tell me about it unless we started off the discussion with an advisory about chicken's blood and mystical symbols, and that the rest of the discussion was about more conventional approaches.

Also, effective use of search is a desirable attribute in a technical person.  I'm not sure I'd be keen on an un-googleable trick question.  I think that the correct approach is to triangulate such that you have multiple independent lines of evidence that the person you have been talking to is right for what you're offering.