What our opinions have in common is that they are not falsifiable. Unless we validate our conjectures with empirical evidence, we're really just sharing our taste.
"Oh no," you may say, "I have tested it, and when I write a longer essay, my retweets and upvotes decline." Very well, but the plural of anecdote is not 'data.'
Somewhere else on the Internet, there is someone with poetic, dense writing. When they simplify it, their audience dries up.
Writers and audiences find each other. If you try to measure the effect of your writing using only an audience that has selected your writing voluntarily, you are not measuring your effectiveness, you are measuring the degree to which your writing pleases people who already like your writing.
Absent a properly constructed empirical approach, we might as well be discussing hemlines and necktie widths.