2. But a tissue start to off gass, i.e. the partial pressure in the tissue decrease, when the nitrogen partial pressure in the tissue is less than the nitrogen partial pressure in the breathing gas.
I think you miswrote this -- You offgas when the the pressure of nitrogen in what you are breathing is LESS than the pressure present in the blood.
Redrover, yours is a very good question, and really, nobody has an answer for it. We all need to remember that the models we are using for decompression are precisely that -- mathematical MODELS of gas transport. They have been figured out using very good physics, but not by putting probes into people and measuring nitrogen tensions in various tissues at various depths. (Thalassamania tells me the Navy has done some of that, actually, but those results aren't available to the general public.) There is a lot about decompression that is not well understood. A really interesting educational resource is GUE's "Mysterious Malady" DVD, which consists of interviews with people doing active research on decompression, but I can boil it down to one sentence: There's a lot we don't know.
The whole idea of deep stops is still a little controversial. The original decompression profiles came out of a pure dissolved gas model. Using that, you want to push the gradient, or difference between inspired and tissue gas tensions, as high as is tolerable, and then sit there and offgas. These "Haldanean" profiles use a relatively rapid ascent rate to shallow depths, and then do long decompression periods there. The depth of the first stops were designed to reach the maximal gradient where significant bubbling wouldn't occur.
But then we found out that everybody bubbles, even using those profiles. At that point, people began to look hard at bubble formation and bubble behavior. The nice thing about bubbles is that they can be detected and measured in live people without sticking anything in them (by the use of Doppler imaging). Because of this, a fair amount of work has been done to look at various ascent strategies and their affect on Doppler-detectable bubbles. The Marroni papers on the DAN site are a couple of those studies. (Unfortunately, the number and size of intravascular bubbles detected by Doppler hasn't correlated nearly as well with symptomatic DCS as one would hope.)
Using bubble mechanics introduces the idea of deeper stops and slower ascent rates. You're basically keeping the bubbles small by the use of higher ambient pressures. (Small bubbles are more likely to collapse than to grow.) The problem with this approach is exactly what's bothering you: Staying too deep, too long results in ONGASSING, at least in some compartments, rather than offgassing. Complex algorithms have been derived to try to create ascent profiles that combine good control of bubbles with efficient offgassing, and it is still quite controversial and open to extensive argument as to which is the best, if any. (Good discussions about this on TDS, with the authors of some of the algorithms weighing in.)
For recreational profiles, the DCS rate is so low that it's really hard to see the effect of deep stops (or safety stops). As Dr. Sawatzky observed in a piece in Diving Magazine, the biggest reduction in DCS came with Haldane's original observations, and much of what we have done since is trying to fine tune the last 5%.
I think it's fairly safe to say that, within recreational gas supply limits, time spent above 30 feet is all decompression time (unless, of course, that's your max depth for the dive!) In other words, at the end of a deeper dive, you can spend as much time as you like between 30 feet and the surface with incurring any greater gas load that you need to worry about, and in fact, depending on the profile, you may be offgassing most all compartments during this time.
For dives where significant time is spent deep, a stop at half maximal depth is probably beneficial. I believe this is what NAUI is now teaching for their recreational divers. Significant offgassing usually begins around 2ATA above the bottom depth, and for recreational dives, this turns out to be very close to half the max depth. Precisely how long this stop should be, I don't think anybody really knows. From there, you can proceed to the shallows at a controlled rate, and spend the rest of your time there.
This, of course, does not jibe with a terrain-based ascent, which frequently more closely resembles the 10fpm continuous ascent that was found NOT to be ideal in Marroni's studies. Terrain-based dives (and especially the ones I've done in the Islands) often spend more time deep (defined as between max depth and 2 ATA shallower) than one would want to do if one used max depth and ran a decompression software program to design one's ascent. But the difference there is that you haven't dived a square profile. At least in my experience, most of the deep Hawaiian dives spend very little time at max depth -- maybe one or two minutes -- before beginning to work their way upslope. At this point, you have to try to figure out what your true "max depth" for gas loading purposes is. Depth averaging almost has to come into play here, but it can be difficult and may not be very accurate when the depth of the dive is continuously changing. No program designs an ascent for a dive where the whole dive is an ascent!
Anyway, don't know if this has been helpful (I know it turned out to be long!)
My approach is to remember that the practical application of any decompression algorithm is akin to "measuring with a micrometer, marking with chalk and then cutting with an ax." My rule of thumb is to add a Pyle stop if the algorithm calls for an initial shallower stop, and to lengthen stops shallower than 30 feet from a little to a lot, depending on the day and the site (how I feel & what there is to see)... works for me 
