For those who use the
Worst P-1 tool, the calculations have changed. Previously the P-1 probability was calculated as being done after the TF which has been done on the exponent. The problem with this is that if an exponent has had abnormally-high TF then the apparent probability for the P-1 drops very low, even though it was done to reasonable bounds. The new calculation is done as assuming TF was done to GIMPS-standard, which makes it easier to find P-1 that was actually badly-done and should no longer falsely report an exponent is a candidate for a better P-1 when the first P-1 run was fine but also had very-high TF.
This will mostly affect the lowest exponents that have had considerable factoring effort of all types, but it will also impact the "expected" values for what makes a good P-1 or not. As a rough approximation, where you say 2% before would be a rough analog to 5% in the new system.