Hi guys, just wanted to put in my two cents.
I am against restarting Omni over again, especially if we're going to go back to the method that was used at the very very beginning and ultimately failed. The original
methodology for going about creating Omni was exactly what Neqitan has suggested: hashing it out with the phonotactics and orthography first and building grammar separately afterward in a "logical" fashion.
Unfortunately, for a project of this kind that method doesn't work--we tried it already the first time. All that happens is that people become enmired in creating the "flavor" of the conlang and can't progress further.
No, the best way to build Omni, IMO, is exactly how mayatec put it on the second restart: to let the language grow just like a natural language would
. A natural language doesn't progress logically from building sounds and having people agree that they sound good--it simply starts in whatever form it's in and the speakers (or writers!) alter it from there.
That is the best chance for any progress to be made with Omni--to simply keep the structure open-source for anyone to suggest modifications. That way if you don't like the "flavor" of the language you can simply suggest alterations that bring it closer to how you want it. Although everyone's opinion may not be accepted by the majority at first, eventually things will shift to the optimum form for everyone to be satisfied.
And that's way over my two cents. Sorry for the epistle. I'm just really into this.
I hope I've been clear enough.