There is a similarly-titled question that asks whether it is possible to split an invention into multiple patents. The question here is what are potential risks of doing so.
Patents may potentially be split up when two processes/systems have overlapping and non-overlapping steps/components. Patents may also be split up when one further expounds on or makes an "improvement" to a subset of another (e.g. more in-depth step/component).
It may be desirable to have a quantitatively large number of patents and/or non-terminal disclaimered (non-TDed) patents for reasons not discussed here. In order to achieve this, one may file according to two different scenarios A and B, shown below.
Scenario A involves a provisional containing all related subject matter, which will be split up via a non-provisional and continuation. The benefit here is that one has all the original information together if needed for claiming purposes. Also, if one is lucky, an examiner might even issue a restriction/election requirement, giving safe harbor to a 'consonant' divisional, thus avoiding TD issues. However, because the disclosures of what are believed to be two distinct inventions are combined, the examiner may be more likely to issue a TD.
In scenario B, the subject matter has been split up from the beginning (i.e. Prov 0 split into Prov 3,4). Splitting up the material as shown in scenario B may reduce the likelihood of an examiner requiring a terminal disclaimer, among other things. However, the risk is that if non-provisional 1,2 are not 'inventive' enough to stand on their own, it may be too late to salvage the overall invention (e.g. attempting to file a continuation-in-part from the two non-provisionals may pose prior art issues).
Questions: Should scenario A always be used? Is scenario B too risky? Are there other ways to salvage scenario B if the two different disclosures need to be combined in order to get an allowance? Is the assumption that scenario B is less likely to invoke a TD requirement incorrect? Is there a better way to safely acquire multiple non-TDed patents on related but distinct inventions?