Re: RFC: a new try for an official community approved certification

On Thu, Jun 1, 2023 at 6:05 PM Jaime Casanova
<jcasanov@systemguards.com.ec> wrote:
> And this is an example of this happening, recently I got a
> certification from a less-known company, and I found at least one
> question in which there weren't any correct answers so I had to choose
> the answer that was at least partially correct. And of course this
> could happen to anyone.

I don't think a community approved certification is a good idea. The
problem that you're complaining about here is that the certification
was bad, and the way to solve that problem is to have something
better. I'm in favor of certifications being better. If they are going
to exist, they should be good, just like anything else. But having
something be community-approved doesn't make it automatically better,
because community members can make mistakes and do shoddy work just
like anybody else. Making something a community effort does the
following things:

- It makes it the official version of a thing, which means more people
will use that thing even if it's worse than some other version of the
thing.

- It means there's a group of people who are in charge of that thing,
and it's usually very hard to replace that group of people if they
stop doing the thing well.

- It means that decisions need to be made by consensus, which is good
for things where the cost of mistakes is high (like core development)
and things where by their nature only one can exist (like the
postgresql.org web site).

A great example of competition-is-good-for-the-project is psycopg2. If
there were just one Python connector for PostgreSQL, it wouldn't be
called psycopg2. Because we never picked an official one, it helped a
bunch of projects thrive, and the one most people use now is the one
that won the competition. If there had only ever been one, it probably
wouldn't be as good.

A good example of the difficulty of managing a large overhaul through
the community process is the documentation. There's a lot of great
content in our documentation, but there's also a lot of old stuff that
doesn't really get updated much and maybe isn't even really that
relevant. Much of the valuable content is buried multiple layers down
in the documentation in places where it's not necessarily that easy to
find, while things of more peripheral importance are quite prominent.
But getting agreement on how to address these problems, or even what
the problems are, is pretty hard. However, since the official
documentation is a thing of which there can by nature only be one,
we're kind of locked in to accepting only those improvements that can
make their way through that process.

But we're not locked into such a ponderous process for certifications.
The way we're going to get a great certification if someone goes and
writes a bunch of great questions and then updates them regularly and
vigorously based on feedback and changes in each new release -- and
that requires either a fanatically dedicated volunteer, or for that
person to be getting paid to do that work.

--
Robert Haas
EDB: http://www.enterprisedb.com