I've been interested and involved in semantic representation language design, implementation, and use since the mid-1970's; a few of you "old-timers" will recall RLL, the Representation Language Language that I did at Stanford during that era, with my then-PhD-student Russ Greiner. From the early 1980's through to, well, the present day, I've led a team creating the CycL language, extending it -- kicking and screaming all the way -- from an RLL-like frame-'n-slot formalism through FOPC to HOL. Along the way, our group (specifically R. V. Guha) carved off a restricted subset of CycL that became RDF, which I quite sagely told him, at the time, would never catch on and lead to anything.
My main interest is in seeing an entire space/ontology of OWL versions, of which OWL 1.1 is of course one instance, where "useful" here means sufficient (in expressive power) for some large collection of important applications. And by "sufficient" I don't mean theoretically, but rather pragmatically. E.g., since there are many applications in which critical pieces of knowledge are of the form "Person x believes that there must be some organization y which is aiming to block person z from finding out that...", one of the features of at least SOME of the languages in this space is that they support a natural, terse construct for expressing nested modals and quantifiers.
Two applications which speak dialects of OWL that support more common features should be able to talk to each other more faithfully than the situation in which one of them (or both of them) need to "dumb down" some of what they would otherwise express, down to the common denominator (most specific common ancestor in the "specLanguageOf" hierarchy), which in turn is, however, still a vastly better situation than obtains today.
I'd like to help with OWL 1.1, but my "secret plan" is to use this opportunity to engage all of us from time to time to keep in mind that broader space of OWL languages.
I do not yet know if I will be able to make the Manchester F2F, but other travel constraints are making that unlikely at the moment.