Given your qualifications for creating these projections, as to the qualifiers effecting the data as it unfolds in the real world, in your work with statistics are their not certain logarithmic formulae that can be applied, particularly with the data when the available pool becomes larger and perhaps more accurate, that might tighten your projections around probabilities?
Wouldn't it be more accurate to hash out columns rated against their range of probabilities, e.g., 10-20 % 10 to 20 millions; 20-30 % 10 to 5 millions, etc?
Just to clarify my statement about my background... I am neither an epidemiologist nor a statistician. Although both disciplines were part of the curriculum in Medical Informatics, they were presented in more of a "Statistics for Dummies" kind of way than as a major component of the degree. Medical Informatics centers more around the application of computers to the daily practice of medicine and to medical information.
That being said, you bring up an interesting point. At some point, the number of people in the susceptible population will be reduced to the point where the rate of increase in the transmission rate (what I call, somewhat inaccurately, the DTR), will start to fall, even to the point where it decelerates instead of accelerates. I do not believe that on September 10, 2015, there will have been 2,329,918,242 cases, as the model projects. Precisely for this reason, if for no other. This is why I was so emphatic that these are projections, not predictions. They simply say "If we take the current rate of increase in the rate of transmission and project out 1 year from September 10, 2015, and if nothing changes between now and then (including the size of the susceptible population), then there will be approximately 2,329,918,242 case by then." Of course, it won't work out that way. But it is helpful for short term projections and to indicate the seriousness of the epidemic.
Unfortunately, I don't know how to calculate the range of probabilities you mention, or how to incorporate the effect of the decrease in the susceptible population that will occur as the epidemic develops. I would welcome that kind of assistance from someone who can offer it.