When you fire any of the vanilla artillery pieces in manual mode (ie. not using the artillery computer), they consistently fire at a different elevation angle than the one that's indicated in the scope, which leads to systemic inaccuracy. This mostly seems constant per artillery piece, ranging up to around +/- 1.5 degrees error, but in my testing I've also seen them seemingly randomly become accurate or extremely inaccurate (8+ degrees of error) for seemingly no reason. 1.5 degrees may not seem like much, but it can easily represent hundreds of meters of inaccuracy at medium-to-long range.
I discovered a lot of these issues while trying to script my own artillery computer. I would get solutions that look correct and match up to what the in-game artillery computer seems to report, but would miss the target in a way that's not explained by dispersion.
There are a few ways to see this in action. First, you can try targeting things with the built-in line-of-sight computer. This is where I've seen the most variability; sometimes the calculated solution will be completely accurate, sometimes it will be laughably inaccurate. In fact, as far as I can tell the *solution* is always accurate; it's just that the actual fired projectile will not fire at the angle that the in-scope elevation indicator indicates. For whatever reason, this is where you see a lot of those 8+ degree errors.
Second, you can try to fire rounds from the artillery computer, then fire in manual mode without adjusting your aim. This will consistently result in two different grouping despite the fact that the projectile *should* be fired at the same place. Again, as far as I can tell the *solution* set by the artillery computer is correct (as in the displayed elevation should result in you hitting the target), but the projectile actually fires at a different angle, resulting in a different hit.
Finally, I've written a script to calculate the actual angle of the projectile when fired, which frequently ends up being a different angle than the one displayed in the scope. This number always seems to be consistent with how accurate the actual aimed shot is (ex. when it's incorrect, I over or undershoot my target).
On top of this, the SPG artillery pieces (all but the mortar) exhibit different error depending on which direction you fire relative to the vehicle. In other words, using the same fire mode and elevation angle, your shell will go a different distance depending on whether you shoot over the front or the rear of your vehicle. This appears to be because these vehicles have a small (~1 degree) "tilt" to them in rest, meaning that it's like you're on slightly slanted ground even when you're on flat ground (tested on the flat pavement at the Altis airport).
Here's some data I recorded firing at different elevations and in different directions. This is the consistent error I've seen, and does not reflect the occasional erratic errors.
Maybe related to https://feedback.bistudio.com/T83261
Testing done on 1.82 stable without mods