It is somewhat contentious what exactly separates major roguelikes from stable games. We should not be suprised about this. The major roguelikes list is seen as a type of canon formation, and thus everyone wants to ensure that their interests are not displaced.
One option may be to approach this historically, and thus list here the roguelikes that have been historically popular/influential. The problem with this approach is trying to find the cut off for "popular" and/or "influential". Nonetheless, we could build a list by adding any roguelike which anyone to believe meet that threshold, and see if the result become unmanageable. (Thus, please use extreme reluctance when adding to this list.)
Roguelikes with USENET hierarchies (rec.games.roguelike.*) are one candidate.
This categorization, however, is a frozen slice of time. The hierarchies are unlikely to change again, regardless of the influence or popularity of any new roguelike. Furthermore, they include roguelikes which, while still played, have been eclipsed by more recent incarnations. The Moria newsgroup is quiet compared to the Crawl related discussions in rec.games.roguelike.misc, or ToME specific discussins in rec.games.roguelike.angband.
The list of roguelikes with USENET hierarchies is:
- ADOM (rec.games.roguelike.adom)
- Angband (rec.games.roguelike.angband)
- Moria (rec.games.roguelike.moria)
- NetHack (rec.games.roguelike.nethack)
- Rogue (rec.games.roguelike.rogue)