I think it's just a question of inaccurate statements and about how individual modders interpret compatibility.
QUOTE
Both Refinements and G3Tweaks (two random names) say that they must be installed last, but they are well compatible (except for the opposing armor tweaks, which are technically compatible but would have unexpected effects).
For example, this is how I do *not* interpret compatibility. If something might have "unexpected effects", they can't be "well compatible".
Additionally, as Sikret said, two mods that say they must be installed last can't be compatible (in certain modders' dictionary: "can't be 100% compatible") unless their statements are based on assumptions. I guess that the statement "it must be installed last" is attached to mods occasionally just to ensure that it overwrites/modifies other mods' unknown (or future) files or modifications (so inside a file, e.g. if they write to the same file position i.e. same attribute of the resource). Or at least, many things seem to be based on assumptions and practical experience by certain modders in these days. The "theory vs. assumptions" discussion between NiGHTMARE and us reflected this.
Sorry for being a bit general (slightly off-topic) again, but this is what I can see between the viewpoint of two groups of active modders:
1. "Players reported no problem so far. Only a very few players who use that spell might encounter problems, and just theoretically. Additionally, the probability that a player encounters it is very small. So they're practically compatible, even if that spell is surely incorrect and causes problems in theory."
2. "They're well compatible, except for [this] and [that], which are technically compatible but would have unexpected effects."
3. "They are partially compatible, so we can recommend players to install both. Only [that] issue might cause problems."
Each problem is now substituted by a version that is agreed by several BWL modders:
1."Players reported no problem so far. However, if our theoretical proof is correct, then the incorrect spell
will cause problems in practice, even if only for a few players. And Low probability * Many players = Some Players! Murphy: If it can happen, it will happen."
2. Something can't be "well compatible" if any unexpected (even if harmless) effects may arise. I want my mod to work "deterministically". There are enough random ShadowKeeper edits anyway, let's try to avoid the
known problems at least.
3. "Partial compatibility is partial incompatibility. 'Half empty. Or half full?' The issue isn't unimportant, it must be emphasized seriously when talking about how the two mods work together."
Nonetheless, it's nice that The_Bigg informed everyone about possible compatibility issues. I believe that compatibility should be discussed both in mod and in component granularity (when applicable). So if there are two mods, one of them has more components, and one component is incompatible with the other mod, the two mods are definitely incompatible. (This is what certain modders call "partial compatibility".) When we say that two mods are incompatible, we state something like: "For technical, conceptual or other reasons, it's not possible to enjoy both mods quality and features in 100% without the risk of unwanted interaction or problems". Component granularity is another matter. (IA's readme details incompatibility of many mods on this level.) Back to the aforementioned example of two theoretical mods, "all components except component 12 of Mod A are compatible with Mod B". This is okay. Correct.
It shouldn't be expected from authors of complex mods that cover many features of the game that they check other works in component* granularity. On the other hand, civilized discussions (such as this one) are okay to share information to each other -- as long as participants can afford the time.
*Or in even smaller granularity. Such as "If you don't do [this] and [that] in the game, they will work perfectly, they will be compatible".