IMO, 40% is not something to preface with the word "only." 40% is almost half a webtoon's score in the contest. It's very substantial.
How do you think the judges will take into account external factors you mention (such as how long a comic has been posted / how large a following the creator had before entering / etc.)?
Let's say an existing creator has 500K subs on their old series. Their new contest entry gets 50K subs. Though they are now one of the most popular contestants, how will Webtoon 'measure' their success, taking into account the fact that they were already popular? The only option that's arguably fair is if Webtoon tries to estimate how much traffic the new comic would've gotten if the creator had actually started from zero for the contest (in other words, how popular the entry would've been if the creator had not already been well-known). But how can that be modeled? There are so many complex factors it's virtually impossible. This question further complicates itself in cases where existing series are reuploaded for the contest and then continued from where they left off. Those stories get more time to introduce their direction: how do you measure that type of advantage? Should subscribers who followed the old version (and are now following the reupload) be counted towards the stats total? The best that judges can do is roughly 'eyeball' things, which is highly subject to mistake or bias...
By using numbers such as 40%, Webtoon strongly suggests the usage of a point system. But as others, like addstract, have explained, if the popularity ratings are calculated linearly then it's almost 100% a popularity contest. I strongly believe that even though the rules may say that judges will take other circumstances into account, judges will not be able to fairly convert those circumstances to mathematical values. Thus, I think that either (1) the line that you read in their rules - that other circumstances will be taken into account - was only added to prevent uproar, and the judges don't actually intend on doing anything about individual circumstances. Or, (2) - audience engagement as a whole will only be roughly considered in that judges will sort entries by popularity and start judging from there (but other than that, they will not consider audience engagement at all - this includes the special cases you mentioned). (2) is the only possibility that allows the 'non-obvious' entries to win, like you predicted. (Also, if that is the case, then 40% is an arbitrary value Webtoon is just throwing around.)
Thus far, LINE Webtoon has been incredibly untransparent about its judging methods/criteria. But if the staff are keeping their word and following the procedures they said they would follow, then having no-name winners like you suggested is not statistically possible.