The "article by Common Dreams" cites a a paragraph in a textbook called Politics in North and South Korea which says, in part:

There was no serious geographic separation according to political ideology until the imposition of the 38th parallel. In other words, communists and capitalists, and leftists and rightists, were not geographically separated in the Peninsula. The North had a number of rightists and Christians, and the South had large numbers of socialists and communists. For example, Pyongyang was once known as "Jerusalem of the East," with 25 percent to 30 percent of its adult population being Protestant Christians by the early 1940s (Lankov 2005). On the other hand, the survey conducted in the South by USAMGIK on August 13th, 1946, shows that a large number of South Koreans had positive views of socialism. On being asked their preferred ideology, 14 percent supported capitalism, while 70 percent chose socialism, and 7 percent supported communism (Donga Daily 1946). Therefor, the three-year period of military rule (1945-1948) on the two sides of the 38th parallel by the United States and the USSR was the process of artificial ideological purification according to the ideology coupled with political persecution. The ideological differences and political competition among the independence movements during the colonial period continued after the liberation, but the communists, headed by Kim Il-sung and backed by the USSR prevailed in the North, and the capitalists and pro-American leaders represented by Rhee Syngman, took over the south. Such process was marked by political violence and bloodshed on both sides and also involved massive movements of political refugees. Socialists and communists who were persecuted in the South went North, and rightists, capitalists, landlords, and Christians came into the South.

The cited source does not appear to be arguing that "South Korea is an illegitimate state" or that "USA was the aggressor in the war". Rather, it seems to be arguing that the construction of the geographic boundaries of both states were not based on a pre-existing cleavage of ideologies across geographies. This seems pretty anodyne, of course cultural-social cleavages don't respect arbitrary straight lines on the globe. The way in which the imposition of borders or districts can produce winners and losers out of existing cleavages is a a huge strain of work in social choice theory (the field which covers things like voting systems and preference aggregation). One example is Tom Schwartz - Your vote counts on account of the way it is counted (1987, Public Choice)

Next, it should be clear that the Common Dreams author combined the socialist and communist categories of the reported survey result. Is this appropriate? Would respondents of the survey have understood socialism and communism as like choices in opposition to capitalism at the time? Beats me. But the survey presumably chose its instrument in that manner for a reason and the source reported the survey in that manner for a reason, so to see a secondary author elide the distinction seems at minimum problematic.

In a contemporary setting, assuming that anyone who expresses support for "socialism" (which might be, say, Scandanavian social democracy, Sanders or FDR style left-liberalism, or actual socialism) must clearly support the DPRK Juche program, is facile. But I also think that's implied because your framing of the question is "I was arguing with some communist on Twitter, and" so it's not clear whether or not you actually think we should take the claim seriously.

I am not fluent in Korean and thus cannot pursue the original newspaper source. I am fluent in survey methodology. We would additionally be concerned that there's little reason to believe the USAMGIK survey was conducted according to reasonable norms; was this a representative population survey? How was the survey conducted? Were population targets known or knowable at the time? In what language was the survey conducted? Was the survey conducted only in Seoul, or in a geographically representative context? Many of the major issues in survey statistics were being worked out around this time-- Leslie Kish would have been at the U.S. Census Bureau at the time of this survey, years before he moved to Michigan and decades before the first major texts in survey methodology were commonly available -- but it is unclear to me whether the conduct of a survey on the ground in Korea would have been in keeping with best practices.

I'd also encourage you to consider whether or not any survey result could support the broader claim. I am skeptical. Suppose that 100% of South Koreans supported "communism", or else 100% of them supported "capitalism". Neither of these claims would legitimize the subsequent dictatorial, unelected, violent, and repressive regimes that mass-executed people in the streets. Trying to project a binary yes/no support for a one word summary of an ideology onto support for the specific policies enacted in the name of the ideology is reductive. The need to adjudicate one side of a historical conflict as legitimate and the other as illegitimate -- especially as a proxy for claiming today's states are legitimate/illegitimate is, I think, pretty unproductive, ahistorical, and non-academic.

The tl;dr: Your source article's source chapter quotes a similar though non-identical statistic to your source article, but endorses a different framing of what it means. It would be difficult to evaluate whether the statistic reported is accurate to the time. I don't follow how the statistic meaningfully connects to the question asked.