The first computer I played BG on only had a 2gb hdd. I didn’t even install the movies. Until weidu came along, mods were usually pure overwrites. If you wanted to revert to the pre-mod state, you had to make your own backup of the relevant files: bgmain.exe for any kind of hacks (no cd, or Taimon’s early stuff), dialog.tlk (most likely for the Baldurdash fixes, though other mods also used that as a base), and the override folder. Besides requiring a strict mod install order (similar to Elder Scrolls modding), the mods would necessarily be independent of each other: Later mods could only ignore anything already present.
Weidu was a remarkable advancement in BG modding because of two characteristics: it was non-destructive, and could account for other mods already installed.
Non-destruction meant that reverting to a prior state was as simple as running the installer again. Every file changed, including dialog.tlk, was automatically backed up, so reverting was simply a matter of taking that backup and throwing it back in the override folder to overwrite the mod’s now-unwanted file. With some of the larger mods, this can certainly produce space concerns. In some cases, maintaining the mod’s backup takes up more space than simply keeping a backup of the game’s pre-mod state.
Accounting for other mods seems to be the holy grail of the current scene, with such a variety of mods, all with different focuses. Now your item mod can play nicely with your NPC mod. The still painful exception is that AI mods, by necessity, cannot account for the innumerable variability of item and spell mods that might be installed. Besides that, most mods are compatible, and install order is rendered somewhat less important.
Sometimes, though, with weidu making these two things possible, mods utilize these features even when it doesn’t make sense to. Let’s use the BG2 Fixpack as an example.
For all intents and purposes, the FP is both the literal and spiritual successor to Baldurdash. These pages, from way back in 2001, show how the game was patched up fairly quickly by the community. Over the years, more things have been fixed — not without controversy in some cases — but the function remains the same. What also remains the same is the requirement for the FP to be the first mod installed post-official patch. This means that, other than for language translation, there is no need to use weidu to patch the files. The old method of blunt file replacement would work just as well, and faster.
The second issue is its backup: it is huge. This is where keeping a backup of the three targets mentioned earlier is better than keeping a backup of everything the mod changes. When you factor in how well-discussed and thoroughly tested the FP is, there’s certainly not the same likelihood of uninstalling as a kit or NPC mod. The final nail in the coffin is FP’s status as essentially an unofficial patch. Barring some real concerns over its more subjective changes, or duplication by multiple other mods, this mod is almost a required install. The ability to cleanly uninstall it is virtually superfluous.
The point is that the larger comprehensive mods, those that make wholesale changes to one class of assets like items or spells, do not need to use weidu. These mods are making changes to the fundamental resources of the game, not adding extraneous content that can exist outside of these changes. Further, these types of mods are the ones that still need to pay attention to install order; and even then, there will be cracks in the pavement. An item mod that applies a change to longswords will not account for a later-installed NPC mod that includes a new longsword. The only advantage weidu had in cases like these mods were the decreased size of the download, a minimal concern nowadays.
 Someone wondered why I was criticizing the FP. I think the text contains sufficient explanation, but I’ll reiterate: An early or first install mod doesn’t need to use weidu, and in some cases, is less efficient for both the player and the modder.