I think of the stream of questions posted on English Language & Usage as falling into four categories: uninteresting off-topic questions, uninteresting on-topic questions, interesting off-topic questions, and interesting on-topic questions.
The site benefits tremendously from removing the uninteresting off-topic questions as quickly as it can—and reducing the close votes required to do so from five to three certainly serves that purpose.
The second-largest category of questions posted on EL&U are uninteresting on-topic questions—duplicates and near-duplicates of the same 250 "grammar" issues that question posters have asked about here for the past decade, single-word requests of marginal utility, etc. Review-queue participants will undoubtedly disagree about whether a particular question is interesting or not, but I doubt that many question answerers feel much disappointment at the prospect that an arguably on-topic question of little interest is more likely to be closed today than it would have been in the past.
The number of interesting questions posted on EL&U on a given day is usually quite small, and some of those questions are disqualifyingly off-topic. For example, MrHen, an early stalwart at EL&U, asked the on-topic question, Where did "duck, duck, gray duck" come from? back in 2013, and its suitability for this site has (as far as I know) never been doubted. But the question When did “Duck, Duck, Goose“ migrate from Sweden to America?—although it might be reworked as an on-topic question about the English language (if, say, it asked when the phrase "duck, duck, goose" as the name for a children's game first appeared in North America)—is, as posed, properly a History Stack Exchange question, not an EL&U question.
Interesting on-topic questions may be the rarest category of all—and unfortunately their suitability for EL&U isn't always immediately evident. For example, the interesting on-topic question The phrase "do the lions", asked four days ago, quickly drew two close votes (out of the three needed to close it and prevent further answers from being posted) before the historically proverbial aspect of the expression—which the question poster had addressed in the original post—received any serious attention. This isn't to criticize the two close voters in this case: they had probably never encountered "doing the lions" as a proverbial expression (I never had either; it seems to have died out sometime in the early twentieth century) and so were inclined to see the question as a general-reference query about geography and the verb "do." But the very fact that these well-informed site participants had never heard of a once-commonplace proverbial phrase is evidence in favor of the posted question's worthiness. If Andrew Leach and I had not independently inquired into the history of usage of "doing the lions," the question would almost certainly have been closed as an off-topic irrelevancy.
A larger number of interesting, on-topic questions have this quality than you might suppose. And to the extent that close voters vote on the basis of what they are familiar with, without checking to see whether a particular question asks about a past or present meaning or usage that they are simply ignorant of, they put these questions in jeopardy. Shifting from five-vote closure to three-vote closure significantly narrows the widow of opportunity that an answerer has to research and post an answer to such a question without having to post a question/argument on Meta advocating that site participants reopen it.
I have argued elsewhere that interesting, on-topic questions already tended to be closed too often and too quickly under the five-vote closure system; moving to a three-vote closure system clearly doesn't improve that situation. But just as clearly, the change from five-vote closure to three-vote closure isn't aimed at interesting questions that may or may not be on-topic; instead, it is aimed at the flood of uninteresting, obviously off-topic questions that pour into the site each day. Since I have no quarrel with quickly closing questions that fall into that category, I accept the reality that the three-vote closure system is an efficient and popular way to dispose of such questions. Nevertheless, I reiterate my earlier recommendation that the site also consider ways to make it easier to protect interesting, arguably on-topic questions.
One way to do this (which I've mentioned before) might be to grant site participants who have demonstrated a site-defined threshold level of competence and interest in a particular EL&U topic special privileges to prevent closure of questions on that topic that are open but in danger of being closed or to reopen closed questions on that topic single-handedly. Another, less radical option would be to give such site participants the ability to post answers to closed questions, after which they could post a Meta question arguing that the answer provides evidence that the question is on-topic and should be reopened. The request might be voted down, and the question might remain closed, but at least the answer would be in place and visible to future site visitors.