I'm doing some sketches of a divided China as part of a TL project I'm working on, though largely as a consequence of a divided Britain which is forced to give up the Empire. (Which is contingent on numerous other things happening to get there because I'm assuming Britain and Germany, essentially, switch spots during the interwar period, so Germany ends up as one of the Allies instead of Britain, and a few other things.) As I see it, the only way I can answer the two questions you pose is that it is all contingent on several things:
- How willing Jiang was able to enact large-scale reform, even if it means pissing off numerous cliques within the GMD who were crucial to maintaining Jiang's hold on power. Jiang was able to get done in Taiwan what could not be possible in China itself because he did not need to do the balancing acts between the various cliques that he had to do business with (the 228 Incident also deprived the GMD of a potential clique, in the form of the old Taiwanese elites, that could have served as a bridge). On top of that, reform IOTL was impossible because he, his entire family, and many GMD cliques benefited from the corruption the system engendered. The wartime exile to Chongqing should have been used as an opportunity to engender reform to the system (even if it means justifying it with excuses about not attacking the Communists hard enough), at least enough to draw support away from the Communists and what little if any support the Wang Jingwei regime got from the people. The more this reform takes place, the likelier the chance the ROC is able to hold onto more territory, reducing the Communist-held territory possibly to Manchuria and the area around Yan'an.
- The level of control Jiang has over Taiwan. Ideally, the more distance between Taiwan and the Mainland - especially if Jiang does not do much reform - the better; likewise, the less of a chance of a war between the Nationalists and the Communists, the better off Taiwan would be. It is very tempting to state that Taiwan would be better off if Chen Yi was not sent over, but I don't think even that would be enough. What needs to happen is that Taiwan needs to be appreciated for its colonial legacy and build on top of and improve/reform it. Ideally, Jiang's claims on Taiwan would be ignored so that the Allies/UN could directly govern it as a Trust Territory for a period of time, allowing it to develop on post-War Japanese and/or West German lines; in order to be able for Jiang to save face, though, Taiwan would be designated as an SAR, with the SAR Government as a Liaison Office between the Mainland and the Trust Territory, and then eventually serve as the basis for a Provincial Government once the Trust Territory mandate ends. However, even if Taiwan is going to be under direct ROC jurisdiction, there are other ways within the existing system that would allow Taiwan to maximize its autonomy (including a fuller application of SAR status as a long-term project similar to the PRC's current use, at least in theory, of SARs), including retention of the Office of the Chief Executive and the Representative Council (省參議會). Before the ROC assumed control over Taiwan, there was actually debate within the Nationalist Government itself over what to do about Taiwan and whether a provincial government would actually be appropriate. The structure that ultimately took shape during the pre-228 period could thus be seen as a compromise, one which could be used to buttress Taiwan as an autonomous province or as an SAR. The main takeaway, thus, would be that the GMD should treat Taiwan on its own terms and accommodate the Taiwanese elites, and not what ultimately happened where the Taiwanese were treated as a potential fifth column under an army of occupation. The more there is a firewall between Taiwan and the Mainland, especially during the Cold War period, the better off Taiwan would be.
- The level of international support. IOTL the international community were so distrustful of Jiang that they relocated their international diplomatic missions in advance of a Communist victory. If, however, Jiang showed more of a commitment to reform, then there could be the chance ITTL of more international support and whether the Chinese Civil War could be TTL's substitute for the Korean War.
With all those factors working together, let's look at some possible scenarios.
A. The ROC as it stood circa mid-to-late 1949: Taiwan (Formosa and the Pescadores) itself and Fujian (essentially OTL's streamlined Fuchien Province, aka Quemoy/Kinmen and Matsu), plus Chekiang (Dachen Islands) and Hainan SAR, as well as nominal control over Szechwan, Hsikang, Guangdong, and Yunnan (as well as a few other provinces in a more titular capacity). If it maintains this disparate state, or even concentrates more on Fujian (either whole or in part) and Guangdong (leaving Szechwan, Yunnan, and Hsikang to the PRC), this maintains a more "Pusan Perimeter"-like scenario that could either serve as a larger rump ROC or the base for a Pusan Perimeter if we transpose the Korean War to the Chinese Civil War. This gets us to an OTL scenario until more international support changes the tide and pushes ROC authority further up. How far up, I certainly don't know.
B. Holding the line at the Winter 1948 cease-fire line: This gives us a map of ROC authority similar to this:
<
https://commons.wikimedia.org/wiki/File:Communist_Offensives_April_-_October_1949.PNG>
Assuming the furthest west for ROC authority is Hsikang (Xikang) province to the West, then the border follows from the northern border of Szechwan (Sichuan) province along the dotted blue line. This would still be considerable territory, though something would have to be done about Nanjing because it would be too close. Either the ROC could maintain its government there or relocate it to safety, treating Nanjing therefore as a Joint Security Area similar to Panmunjom IOTL. Bonus points here if the
Ma clique was brought on side, thus bringing at least Qinghai and Gansu under ROC control, eventually making it possible for the ROC to reassert control over Hsinchiang (Xinjiang, Sinkiang).
C. A variation of Scenario B would bring it closer to the traditional division of northern and southern China, the
Qinling Huaihe Line, which is close to the 33rd parallel north and the 0 °C January
isotherm and the 800 millimetres (31 in)
isohyet. Easy enough, were it not for a few problems because of historical changes to the course of the Huai River that means it's now a tributary of the Yangtze River. Either the ceasefire line in Jiangsu province in this case follows
the old course of the Huai River or
we take an irrigation canal out of commission. It does mean more area for ROC authority, with some potential for variation, such as including the remaining bits of Hubei and Anhui provinces, or even enough territory to bring Xi'an under Nationalist control.
Ultimately, there's no actual minimum defensible borders on the Mainland unless the GMD are willing to make an effort towards actual reform, and if the international community were willing enough to support Jiang or some form of GMD regime without Jiang at the head. Those two factors would make it possible to change the tide of the Nationalists and allow it to create some minimum of defensive borders. I am also not sure about strategically even, but I do know that Scenarios B and C bring it close to ancient regional divisions that would only be strategic if the Tibetan Plateau were brought under ROC control - which makes it easier to control South China's dense river transport network. Somewhere around those two scenarios would be what I would see as realistic for a Soviet-American cease-fire line - for example, a nominal adherence to the 33rd parallel (outside of Tibet proper, which is probably left out in this case unless the ROC makes moves on it), but in reality is all over the map.
