iwatt, that is a fantastic idea! To answer your question, I think the way Stalker came up with his numbers, is that he determined which DC would produce the desired success rate for an individual check, and what success/partial-success thresholds would produce the desired success/partial-success rate for the overall challenge. So as long as your new table of DCs has the same success rate for individual checks as the D&D table of DCs, the table of success/partial-success thresholds should be the same for both systems.
You should start a thread about this on the True20 boards (if they ever come back up...
). The main issue I see is that in True20, the skill modifiers can vary a lot more widely than in D&D. For example, by level 7 (T20) you can have a +20 skill mod (10 ranks + 5 ability + focus + talented), while your teammate only has a +0 in the same skill. So against a DC 21 check, you succeed automatically and your ally fails automatically (it gets even worse if the Narrator allows you to Take 5). In D&D, getting a 20-point difference is difficult before level 15, and even then it requires a level of commitment that is a bit more expensive that in True20. So it's easier to have a one-size-fits-all DC in D&D than it will be in True20. You can probably account for this by having degrees-of-success or options like Going for Broke or some way for the +20 guy to help the +0 guy out. I am interested to see what you come up with.
-- 77IM
And yet another word from our sponsors
