The table below shows the chances of success with and without GoE:
Description | Without GoE | With GoE* | |
"nearly impossible" | 1% | 2% | |
"nearly impossible" | 14% | 26% | |
"very difficult" | 15% | 28% | |
"very difficult" | 34% | 56% | |
"difficult, but possible" | 35% | 58% | |
"difficult, but possible" | 54% | 79% |
As you can see, GoE makes the text descriptions less useful by increasing the range of the brackets, especially at the extremes when it nearly doubles the bracket ranges.
If I knew I had a 14% chance to succeed without GoE, I might use GoE and attempt the project, knowing that I would probably have a few failures but would eventually succeed. But if I knew that my chances were only 2% with GoE, then I would skip the project.
My preference would be to keep the existing text descriptions and in addition show the actual percentage chance. A helpful alternative would be to increase the number of brackets for which there are text descriptions, especially for the "nearly impossible" and "very difficult" brackets.
*Computations:
GoE grants a second d100 roll per attempt and takes the higher of the two rolls. The chance of success when using GoE is the complement of the square of the chance of failure when not using GoE. For example, if you would normally have an 80% chance of success, then using GoE would increase the chance of success to 96%, computed as follows:
1 - ((1 - 0.80)^2) = 0.96
The maximum benefit is at 50%, when the chance of success is 75% with GoE. The benefit of GoE decreases as the probability of success becomes higher or lower than 50%.