Dr. Tim Maher of CONSULTIX, Damian Conway of Monash University,
W. Phillip Moore of Morgan Stanley, Karen Pauley of Kasei, and
Timothy Wilde of DynDNS.org participate in a panel discussion of Perl
certification. Before we get started, I’ll just note that I’m not a fan of
programming language certification. A general programming
certification might be useful, but looking at one language is too narrow.
Discussion opened with an audience member saying that he’d received a
resume from someone claiming to be Perl certified. The moderator said that
there were Perl certifications, but they weren’t necessarily credible.
Tim Maher’s opening statements: Perl is a great language that doesn’t fit
well with MIS expectations. It’s much too easily dismissed as a “novelty”
language. We’re losing market share to other languages. Perl needs a
Certification could be good for Perl hackers and Perl, but some current
tests are so bad as to gravely offend testees. The Perl language is
uniquely suited to testing (eval). Maybe we should create our
own certification process.
Damian Conway is introduced. He’s a Perl speaker, educator, hacker,
author, and evangelist, and an associate professor. He feels that
certification is a pragmatic response to managers’ need to assess potential
employees. It doesn’t even need to really measure competence so long as
reliers believe it does.
Phil Moore is Executive Director of the Unix Engineering team at Morgan
Stanley. He’s the original author of the MQSeries suite of Perl modules.
He found that many certified applicants didn’t know much other than what the
certification specifically tests. The certifications that people respect
are the Cisco certifications. The test included a hands-on troubleshooting
component where the testee had to exhibit real skills. For similar reasons,
people respect Red Hat certifications. For a high end shop like Morgan
Stanley, certification isn’t important, and most of their hires don’t have
them. The exception was hires from foreign universities, where they
couldn’t evaluate what the transcripts really implied.
Karen Pauley is the Managing Director of Kasei, and is a “Herder of Perl
Hackers”. As a manager, she finds it exceptionally difficult to evaluate
prospective Perl programmers, so she thinks certification might be a good
idea. It shouldn’t be run by Perl trainers due to conflict of interest, but
the big names in Perl should be involved.
Tim Wilde is the owner/founder of DynDNS.org. He’s a system
administrator, occasional Perl hacker, and a former professional Perl
programmer. He needs a mechanism to weed through resumes. When he posts a
job on Monster.com he might get 200 resumes, including one from a
construction worker with no computing education or work experience.
Existing certifications are no good.
What would be the purpose of Perl Certification? Not to
exhaustively classify everything someone knows about the language.
Testing Technology. Multiple choice is no good, just convenient
to grade. They’re susceptible to test taking skills. And there are lots of
ways to do something in Perl, and a testee might know only a few of them.
Instead, require on-the-spot coding, which is sensitive to actual coding and
problem solving ability.
General Discussion. There’s little consensus about this topic.
I’m surprised at how many of the managers say they want to hire “Perl
programmers”. I’ve never wanted to hire a “Perl programmer” (or a “PL/I
programmer, or “C++ programmer”, or “PowerBuilder programmer”). When I need
programmers, I’m always
looking to hire a good programmer. If I need Perl skills right now,
I’d prefer that the good programmer had good Perl skills. But next year, or
some day, I’m going to have a project that doesn’t use Perl that I’m going
to want that person to work on.
No clear direction came from this session, but some interesting points