Once upon a time about 30 years ago when I was in grad
school (engineering) a good friend of mine (from the
philosophy dept.) and I went on a camping trip for a
week.  Our original intention was to get isolated from
all civilization for a week and to communicate without
using any language we already knew.  If we needed a
word for something we would make it up and point at an
object or pantomime an action to get the meaning
across.  Our hope was that at the end of the week in
the wilderness we would have invented some kind of
minimal conlang.

What we ended up with was far too primative to be
called a language.  It involved more point-and-grunt
than anything else, and we certainly couldn't discuss
any lofty philosophical notions in our
invented "language".

But ever since then I have been intrigued by the idea
of creating a conlang in exactly the opposite way
conlangs are usually created.  After all, natlangs
were all created, as needed, by people who knew
nothing whatsoever about formal linguistics.  Whatever
worked to communicate an idea was kept and refined by
usage, and hatever was difficult, confusing, or just
didn't work was discarded.

So the idea that I've had in the back of my mind is a
collaborative conlang created by a group of people,
but without any "planning" whatsoever.  Gramatical
ideas are NOT discussed by the collaborators.
Vocabulary is NOT discussed by the collaborators.
Instead the collaboration happens by way of discussing
other topics of interest EXCLUSIVELY in the newly
developing language itself.

A minimal seed a a few hundred words would be
generated entirely at random, but no grammatical rules
whatsoever would be laid down.  There would be NO
central authority to adjudicate disputes or pass
judgement on what is or is not acceptable usage.

Collaborators would use the minimum vocabulary in any
manner they saw fit.  As time went on a consensus
would emerge, not by arguing the merits of the
different grammatical approaches, but by watching to
see which grammatical structuires emerged as the most
popular ones in articles written about home gardening,
photography, or stamp collecting.

Not until the language had evolved a reasonably
consistent consensus grammar and vocabulary would
actual discussion of grammatical points be permitted.
Even then, such discussions would be descriptive, not
proscriptive.  The grammar and vocabulary of
this "organic" conlang would never be changed by
decree.  They would change only when popular writers
of the language itself set the examples that were
adopted by the community at large.

To avoid biasing the language toward any one
grammatical system the collaborators would have to
come from diverse liguistic backgrounds. Otherwise
the new language would end up being a relexification
of English, or whatever the majority language was
among the collaborators.

In fact, it would very interesting to observe a small
group of five or six people with no single language in
common work out a system of communication based on
a "seed" vocabulary of a few hundred random invented
words not taken from any of their native languages.
Kind of like growing a pidgin under greenhouse
conditions.  Have the group meet a couple evenings
a week and "chat" in their own unique evolving
language.

Or for that matter, have them get together in an
Internet forum were the rule is that NO natlang or
pre-existing conlang would be permitted in any post
on that forum.

That would be an intersting way to grow a conlang.