Re^3: What is your opinion in Perl Certified Professional?

[Tanktalus]

I wouldn't mind if he blessed a test. Because, if he's been "clueful enough to not bless any particular test" so far, he's likely to only bless tests that meet a high level of reliability.

It's sort of like when I'm playing Euchre (or any other partner/trick based card game) - I may have a bad hand, but I want my partner to bid something. Not because I'll support them, but because that means they have a good hand. I have no idea what a good test would be, but if Larry said it was a good test, maybe that means it'll be a reliable test of someone's applicability for perl-related work.

That said, given what is attributed to Larry in perlstyle, I can't imagine him saying that any test was definitive enough to be acceptable. ;-)

[tilly]

I would mind if Larry Wall blessed a test, because the economic dynamics are against any particular test or testing organization being able to do a meaningful job.

In general there are two kinds of certifications.

The first are ones which might look good on a resume, but are not that hard to get and don't have a good reason to exist from the perspective of public policy. Think MSCE. These pretty much universally suck. They tend to be run for maximum profit on the part of the certifying authority. Frequently said certifying authority is associated with a vendor, who uses the test as an advertising channel for the products that they would like people to adopt. Since passing them is fairly easy, the people who are most motivated to collect on them are people whose other qualifications are relatively weak. They act as a filter reducing the number of abysmally ignorant people, but are worse than useless for figuring out who is competent.

There are good economic reasons that certifications tend to do that. Often they start out being fairly good with the best of intentions, but they don't last.

This includes all certifications that I'm aware of in computers.

The other category is certifications that serve an important public policy purpose. Well-known examples include board certification for doctors, engineering certifications, CFA, and so on. These certifications tend to be very difficult to get and are very meaningful. Typically they require that you both pass a series of exams and that you have specific amounts of work experience doing specific jobs under supervision. They are also recognizeable by the fact that there tend to be laws mandating that people possess these certifications to do certain jobs.

I simply don't see the necessary dynamics for the latter kind of certification to come into existence in programming Perl. And, as I said, I am highly skeptical of the former kind of certification on general principles. Therefore if I saw Larry Wall bless a particular certification, I'd strongly suspect that he got paid to do that, and would be disappointed by the action.

[Steve_p]

The only computer-ish certification that comes close to the later category that I know of is the Oracle Master certification. Along with the typical tests and week long expesive classes, it requires a two-day hands on practical exam. No multiple choice questions. Not many people go as far as to get an Oracle Masters certification, since they've usually conned they bosses by then. Those that do have it really know their stuff.

[adrianh]

They act as a filter reducing the number of abysmally ignorant people, but are worse than useless for figuring out who is competent

If you're the poor SOB with 150 CVs on your desk and only a couple of hours to get an interview shortlist then a "filter reducing the number of abysmally ignorant people" is a tool that you will often reach for.

If you want to be hired by that poor SOB then you have the problem of CertificationAsProgrammerTax .

[tilly]

You will only reach for the tool if the tool is there to reach for.

I personally do not wish to be filtered for/against on that kind of basis, nor do I wish to work with people who were filtered for in that way. This means that I have no desire to see that tool come into existence.

I am therefore happy that Larry Wall hasn't given his endorsement to anyone who is trying to do that.

[apotheon]

I've never found the "A filter is needed!" argument to be very useful or valid. If all you care about is a filter, use a random number generator (like maybe dice). If what you want is a GOOD filter, you should probably avoid certifications.