I reckon it's the flight modeling that gives us more stability than the planes had in reality. It's my cumulative gut feeling from everything I've read and heard. There's a number of short-comings in the FM. For example, planes don't hammerhead as well as they ought to. This is from someone who flies acrobatic in reality. Level acceleration doesn't perfectly match reality either: e.g. the AH Tempest is one of the strongest accelerators, but historical trials show it was at the other end of the spectrum. Or like I said, we have perfectly still air. The Ta152 as modeled in the game wavers a lot unless you keep it in check with very steady inputs, and I reckon if we had real skies with varying winds, drafts, turbulence, etc, it would be much more like the real Ta152 that was borderline unsuitable for combat because of instabilities.... So that's how I see it - it's more due to the flight model, and adding an artificial dispersion to the Ta152 or Il2's guns if my hunch is correct, wouldn't be the right thing to do.
To answer your question, I read the battlefield.ru article ..
The worsening of flight characteristics of the new Shturmovik, as with the IL-2 with the ShFK-37, was due to a large mass/wing ratio and the design of the cannon pods which lowered the aerodynamic performance of the aircraft. From an overall point of view, the IL-2 had poor balance; this indefinitely lowered cannon accuracy. This major factor was amplified by the strong recoil of the cannons during firing. According to NII VA VVS KA data (a letter of NII AV General-Major M.V. Gurevich, from 19.11.1943, to S.V. Ilyushin), the maximum strength of the recoil , present for .03 seconds, on a ground-based platform(existing technology of that time didn't allow the measuring of "the strength of recoil" in aircraft and much less in the air during firing) consisted of a considerable amount - 5500 kg, with a recoil average of 2500 kg. All of this contributed to the wide dispersal of shells during aerial firing.
What exactly does that mean? e.g. "Wide dispersal"? They qualify it all negatively, but it's all too vague for us to try and replicate their trials. Not in any way that I can think of... Not unless someone can find a copy of the reference documents with more details of those trials.