The traditional method has been to keep deflections to L/500. I've also seen L/400.
Though this isn't a hard limit, it's not codified anywhere. And if you actually dig into some of the articles promoting it, this limit is based on a 10-20 year return on the wind (I've seen articles say 10, commentary to ASCE 7-05 Appendix C says 20; commentary isn't included in most printings, you'll need to search for errata). Most use the limit in conjunction with service-level wind (i.e. without the 1.6 factor), which is wildly conservative. ASCE 7-05 Appendix C commentary recommends using a 0.7 factor on wind to get to 20 year return.
I would only use this limit for schematic stages or projects without a wind tunnel consultant. Just something to keep your designs in check during the early stages before wind tunnel consultants get involved (which should be much earlier than it typically is, but that's a separate discussion). Humans don't 'feel' displacement, only acceleration. So this limit has zero bearing on human comfort. You'd need to look at maximum accelerations to address the human comfort issue. You'll likely need to get a wind tunnel consultant involved for this aspect. Once a wind tunnel consultant gets involved, I'd throw the 1/500 limit out the wind and let them do the heavy lifting on this part. They know more than you. Just make sure that you let them know of any major changes in building geometry or stiffness. Major changes in building geometry may require a new analysis if they're major enough. Major changes in stiffness they should be able to address without as much rework, it's mostly just plugging numbers in but shouldn't need any retesting for the typical level of analysis.
