5--so, like I learned about how long a tank really lasts, the answer is it depends but generally blow off or shorten the deeper stops and keep to the shallower ones.
That may roughly be the concensus here, but in my opinion (for what THAT's worth), Steve from SDI/TDI has the best post in the thread.
The models and programs we use to calculate decompression schedules have gone through a certain amount of validation, and we know that the majority of divers can surface safely using those tools. We have NO testing of shortening deco schedules. We have anecdotal reports of people abbreviating deco (see the Truk lagoon thread) and either surviving it, or getting hurt.
When someone says, "I'd do this," there are no data behind the statement. (Obviously, the experiment of blowing off a certain amount of deco deep or shallow and seeing if people get hurt, and if so, how, isn't going to pass anyone's IRB.)
My buddy and I were discussing this yesterday, talking about why it's good to understand a number of different models and what the shape of the decompression is that's generated from each one. You may plan a dive perfectly and have plenty of deco gas, but your buddy rips his drysuit and has a full flood in 45 degree water. There's a thermocline and the shallow water is 75 degrees. Can you reshape your deco to get shallow faster? Understanding the difference between bubble models and pure Buhlmann may allow you to reshape your decompression to salvage the situation. But there, we are still remaining within the total decompression required by the models, so it's still using a validated system. There is no approach to shortening the total decompression that can be said to be superior to any other, as the data don't exist.