Yeah - it's tough man. There is definitely a growing disparity between what is known (or theoretically understood) about how certain index phases statistically connect to enhancing probability for certain weather phenomenon, vs. what is often observed. Take your '95-'96 winter: I was living in Massachusetts earlier that summer feeling cheated out of what was at that time a record breaking hurricane season; mainly, because New England was spared (just like this year - argh). Anyway, as you know, that winter was exceptional. At one point we had 40" of snow on the level and if you cut through the pack it was stratified into different layers of aggregates related to temperature at deposition times from successive events. Kind of like what avalanch researchers look for in predicting what layers are more prone to bifurcate and cause slides to take place... But, there we were at sea-level with our own micro-study.. It got so inundating with huge snow banks narrowing streets that it kind of changed the land scape unrecognizably in the intermediate sense of it.. But then, in latter January, huge thaws caused flooding...February, one last shot and we again getting belted by deep cold and heavy snows... That was it until 2000! In 1998 we were in the upper 70's almost everyday for a week in early December and much of that season was warm enough to play tennis outdoors.

Point being, at first glance one would be inclined to think that the huge hurricane season earlier that summer and autumn may somehow have been connected. The fact of the matter is they likely were not... Because, as you intimate, the phase of SOI and its connection to the ENSO state was in flux... Now, we do know that there is a time dependency for rapid transitioning ENSO modalities and the atmospheric response. That really confuses matters and makes your research look dubious if you don't acknowledge that as being true. If you are trying to write a thesis paper of some kind, pertaining to the connectability of index phase states to tropical phenomenon (or extra-tropical for that matter) you'd do your self a world a good by considering/planning for a time lags.

I didn't know that the SOI had changed or assumed any states as emphatically as you allude. Shows what I know.. I had just read NCEP autum climate eval and it didn't really lay mention to anything like that. They were pretty clear that neutral states of the ENSO were afoot and they believed also that it would likely persist into spring of 2006. That does not sit will with a SOI that wants to drive easterly trades does it? Kind of sounds like SOI (which we all know is highly correlated to ENSO) is indicating the neutral could be more transient than perhaps NCEP is currently aware or thinking.

I have been doing my own independent research in the matter regarding indices.. just for s*(&! and giggles I've been doing correlation cof on different states to see how well they move together... So far, I've nailed down some interesting findings.. One is that the AO (Arctic Oscillation - of which the NAO is a nested component) does not always concurrently operate when the NAO is in a given phase. Also, there is some weak correlation between the 2 year periodicity of the QBO and NAO, which is really weird because I'm not totally certain how the heck they could be shaking hands...it is a "weak" correlation at best however.. Anyway, if you are trying to connect SOI to hurricane frequencies, I have the about 50 years of SOI, QBO, NAO and AO all in spreadsheet (EXCEL)... I could email them to you, or provide you with NWS websites that provide the indexes going back to 1958. But just remember, the earlier phase indices are a little bit dubious (at least to me) because in their write ups the admit that the earlier measurements are based on study rather than actual measurements.. That is because many of these indices weren't readily known or explored before 1970-ish...