For liquid yeast, the viability of 96% seems to match to Jamil's Mr. Malty app, and the aging rate of 21% seems reasonable (Jamil seems to use 21.5%, FWIW). However, as you move the production date further back in time, things diverge greatly between the two calculations. The calculations in BeerSmith seems to be nonlinear, while Jamil's is linear.

For example:

Production Date Viability-Jamil Viability-BeerSmith

Today 96% 96%

1 month ago 75% 75%

2 months ago 53% 59%

3 months ago 32% 46.5%

4 months ago 10% 36.5%

The Mr. Malty calculation bottoms out at 10% and seems to stay at that number for anything older than 4 months (which seems suspect to me). But in general, he seems to be subtracting 21.5 "percentage points" per month until reaching the 10% floor.

On the other hand, the BeerSmith calculation reduces the number by 21 "percent of the previous number" per month.

With fresh yeast, the differences aren't significant. But with older yeast, the differences are significant. Jamil's formula takes only 4 months to reach 10% viability; with BeerSmith it takes over 9 months to get down to that number.

Is this an intentional difference? Which is "more correct?"