jerm said:
How about with skills?
I see your reasoning, Dex check at -2 in 2nd ed would be a DC 12 Dex check in 3.5.
I wonder if skill checks translated from 2nd ed should be handled differently. For example, 2nd ed used Wisdom for spot and listen checks. Now there are skills for them. A Wisdom roll at -4 to hear something in 2nd ed.. sure we could do a Wisdom check vs DC 14, but wouldnt a Listen check work better? Assuming a character for whom Listen is a class skill, after just a few levels could easily have +7 or so in that skill. I guess what I'm getting at is that Ability checks and Skill checks should be expected to be different numerically. Even at high levels, one could expect an Ability score to reach only 20-22, which is a bonus of 5-6. The same level character will likely have many ranks in some skills (in addition to the bonus from the corresponding Ability).
[edit: Attribute, Ability....]
Well, proficiency checks in AD&D2 were based off of a modified ability, so, something like Wis -4. So, to simulate the same difficulty, you'd want to add whatever that number is to the DC. So a "Mining Check at -3" in AD&D would translate to an appropriate skill check at DC 10 + 3(base for skill) + 3 (modifier) = 16. In the absence of access to the AD&D2 books, or, if you want to simplify things, i'd just add 1-2 accross the board. At a quick visual scan of the proficiency tables, it looks like the average modifier, accross all proficiencies, is between -1 and -2.
Of course, that's assuming you want strict mathematical fidelity. While those checks weren't intended to be too difficult, they were intended to be something of a challenge. As others have observed, abilities didn't go up with level in AD&D2, and there was no practical way to increase the score for proficiency checks, either. So, for all practical purposes, such checks didn't scale with level. To compensate for this, you could either address it directly, by simply adding average character level (or half of level) to all such DCs, or you could make a more-general correction. Along the latter lines, it looks to me like 15 is closer to a baseline DC in D&D3E for a skill check, with 10 being reserved for particularly-easy tasks. Furthermore, the default method for generating abilities in AD&D2 was 3d6, not 4d6, so characters would have lower stats, by about 2 points (or +1 mod, in D&D3E), on average. So, figuring all this together, i'd suggest going with DC 15 + the modifier, whether an ability check or a proficiency/skill check. That should give the closest fidelity in terms of feel, if not strict mathematical conversion.