[lpi-discuss] General comments on LPI levels

Mark Miller mark.miller at lpi.org
Sat Sep 17 16:35:52 EDT 2005

On Sat, 2005-09-17 at 10:07 -0700, Jack Coates wrote:

> 1) In my opinion, LPIC-3 is a better place to put priority than LPIC-0.

I agree

> 2) LPIC-3 should be a very difficult test with hands-on access to remote 
> machines.

This won't happen at LPI. The costs and infrastructure requirements are
too big, particularly in the Third World.  There is no research that
psychometrically supports the notion that a hands on test is superior to
a well crafted written exam in this domain.  The desire to do it on a
computer is a feel good thing.  Believe me, doing something once is NOT
proof that you really understand something.  In every performance based
training situation I have been in (with more than 20 years of military
experience and experience in Intel's development fab for 3 years) you
never rely on a single performance of a complex task. Since (as I have
said so much lately) what we are testing is purely cognitive there is no
advantage to a performance test.
> 3) Real world scenarios should be part of the test suite.  I think that 
> real-world means integrated with non-Linux solutions, but it's just an 
> opinion.

Getting real world scenarios has been a challenge.  You think it is hard
to get geeks to write documantation, try getting them to write tests (or
even scenarios to ask questions about)! Ideally we will move closer to
this as much as we can. We must however stay focused on Linux and not
the literally hundreds of other technologies that we might interface
with.  So many in our discussions assume a Microsoft mixed network when
in my experience you might face MS, Unix, Netware, VMS, and OS/2!  I'm
not making that up, I saw exactly that at Intel on their networks. The
more you get away from an office setup the more you see such things.

Mark Miller
Program Manager
Exam Development Level 1
Linux Professional Institute

More information about the lpi-discuss mailing list