Today James mentioned the decline in rival adventuring parties as foes in D&D. The earliest random tables (OD&D, Greyhawk, Monster & Treasure Assortments, Holmes Basic) included many NPCs in the random encounter tables for dungeons, though these were groups of magic-users, fighters, etc, all of the same level (e.g. Thaumaturgists). They may have been listed this way because it makes for a simpler table and an easier random encounter for the DM to stat, since they are all the same class/level. But my impression was always that these NPCs represented evil groups (cults?) operating in the dungeon rather than adventuring parties. However, either way it is interesting to note the decline in these encounters as D&D evolved in the 1970s. One theory I have with regard to magic-users is that this was due to Sleep spells being too difficult for low level parties to handle.
In the OD&D, Vol 3 (1974) Wandering Monster Tables, a 1st level
dungeon should have encounters with Conjurers (Lvl 3 MU), Theurgists
(Lvl 4), Thaumaturguists (Lvl 5), Magicians (Lvl 6) and Enchanters (Lvl
7) at frequencies of 3.2%, 3.2%, 1.6%, 1.6% and 1.6%, respectively.
That's more than 11% of wandering monster encounters on a 1st level dungeon. Note these are
encounters with a group of multiple spell casters. Given the usefulness
of Sleep against the other wandering monsters found on that level, one would
reasonably expect that at least one MU per group would have it
memorized. Sleep should affect 2-16 1st level characters. All it
takes is surprise or an initiative win by the MUs, plus a good roll on
2d8, to put most or all of a 1st level party to sleep instantly without
save.
In Greyhawk (1975) the chance of encountering wandering MUs on a level 1 dungeon drops slightly from almost 12% to just under 10%.
In Set 1 of the Monster & Treasure Assortment (1976) the number drops severely -
there's only a 3% chance of encountering a wandering MU on level 1.
In
the early printings of Holmes Basic (1977) the chance was higher, back up to about
6%. But by the 2nd edition of Holmes (1978), all NPCs were removed from the
Wandering Monster tables (presumably because many of the higher level
types listed were not described in the rulebook), bringing the chance to
0%.
In the AD&D DMG (1979) there is a small chance of encountering MUs
(max # is 3) on the 1st level of a dungeon - just over 2% using the
Tables in Appendix C.
In Moldvay Basic (1981) there's also a 0% chance of wandering MUs on level 1 (pg 53).
You can see the calculations for these numbers here.
So
there's a big drop in the chance of encountering wandering MUs in
the dungeon as the rules evolved during the '70s. This makes it much
less likely that low level PCs will get hit by a Sleep spell in the big games of the early 80s (AD&D, B/X). PCs still get their Sleep spell to use against the goblins, etc but no longer have to deal with the same spell being used against them. One could argue that the Wandering Monster tables were nerfed!
So if you really want to go old school, 1st level PCs should be encountering groups of rival magic-users on the 1st level dungeon more than 10% of the time. In OD&D, not all of these are immediately hostile given the Monster Reaction table, but on average 1/3 will be. Thus 3-4% of the random encounters will be with hostile higher level spellcasters, most of which should have at least one Sleep spell. Some may have Fireball as well...
(Portions of this post appeared previously in this OD&D Discussion thread).