Eventually, we'll need to restore normal permissions on the Korean Wikipedia, temporarily changed in T94388 following a local discussion which feared catastrophes. Catastrophes didn't happen anywhere, hence there is currently no rationale for the unconforming configuration.
Positive effect on other wikis
We should probably gather some data from a wiki which Korean users may feel as similar enough to their case and then present the data locally: perhaps Chinese Wikipedia? https://stats.wikimedia.org/EN/EditsRevertsZH.htm shows a slight increase in unregistered editing and a stable revert ratio. Norwegian Wikipedia shows a different revert culture. Vietnamese even has reverts going down proportionally, but the editing patterns are different. https://stats.wikimedia.org/EN/EditsRevertsVI.htm
Korean Wikipedia's unregistered users
The "Distribution of reverts" table shows us the workload: reverted edits (minus self-reverts) are produced in similar amounts by unregistered and registered users: 34 % vs. 42 % of the total respectively. As for who does the (likely) cleanup: unregistered users take care of half of the reverts of other unregistered users and 1/3 of the reverts of registered users' edits. So, in total, unregistered users seem to contribute to patrolling almost as much as they contribute to bad edits.
In total, 2015 and 2016 show a continued increase of the amount of edits by unregistered users, while edits by registered users are stable. It seems therefore that development and growth of the Korean Wikipedia depends more and more on unregistered users. Notably, the amount of edits increased while not affecting the ratio of bad edits: the ratio of reverted edits is still very low compared to Wikimedia averages, around 8 % (on many wikis this figure is near 20 %), while for registered users it's around 3 %. In other words, restricting ourselves to registered editors only gives us a potential marginal gain of 5 percentage points of "good" edits.
All this tells us that letting unregistered editors edit on Mobile is very likely to produce a net positive on the Korean Wikipedia, even more than other languages.