On Sep 7, 2009, at 9:22 PM, Kiran K Karthikeyan wrote:
I would be interested to see what questions silklisters use to guage
analytical, problem solving, and other skills when you are looking
for both
a creative and analytical profile. This is of course apart from
questions
based on their domain/work experience to guage the same skills, but
I use
standard questions so that I can compare one against the other if
required,
especially when the backgrounds and experience vary considerably.
The couple questions posed here were problematic in that a person
could be familiar enough with those types of interview questions to
give an easy response even if those individuals were not particularly
good candidates. I have seen those before as classic questions, have
built-in answers, and I imagine plenty of other people will as well.
At the same time, I don't have really good answers for this.
My business is exceedingly technical. Ultimately, I am looking for
someone with breadth and depth -- a polymath -- but also the ability
to apply that knowledge to problems they have never seen before or for
which there is no answer you can lookup in a book. This makes it a
little more difficult to analyze, but helps ensure you get a good
reading. One of the ways I test that is to ask broad questions with no
answers but which have many avenues of attack and for which there is
something resembling objective metrics for the answers given. This is
usually sufficient to gauge a technical person very quickly. For
software engineers, this would be open algorithm problems; not
theoretically hard problems (e.g. provably NP-complete or similar) but
just well-known holes in the literature of which there are hundreds.
There are also well-understood and familiar problems that generalizing
well requires considerable design nuance and experience such that most
people will use at least some naive methods even if a known solution
exists for various design parameters; if you come across a true
expert, monkey wrench them with real hardware that operates under
different parameters such that it violates intuition. Good multi-
aspect reasoning is much harder to fake than encyclopedic knowledge.
I am looking for two things: the kinds of tools they have in their
tool box and their adeptness with the tools they have. The difficulty
is in coming up with a good problem for them to use their tools on
that does not require specialist knowledge in order to perform well.
Not all cases are the same, there are no experts in the wild for some
topics as a practical matter and in those cases it is all about their
ability to learn arbitrary theoretical domains i.e. the ability to add
and incorporate new tools.
For more real-world situational experience with products and
customers, I draw on the numerous "wtf?" problems that occurred at
various points in my past career. At a minimum you get a sense of how
they deal with edge cases, though that may be misleading with respect
to their ability in normal cases.