Wikipedia:Shoot it early

Aside from edit wars and conflicts between contributors (such as between those with differing points of view or tastes), there are many edits that serve to degrade articles. Though vandalism is not uncommon, there are many other types of edits that do not belong, from simple mistakes (including those made by malfunctioning scripts), to refactoring the text in promotional and non-neutral manner, injecting unsourced controversial material somewhere amid well-sourced encyclopedic text, attempts to impose a censorship, and simply incompetent changes which may look like a copy editing, fixing mistakes, or even implementing Wikipedia guidelines. If the article is watched by one or more competent editors who consistently act to repair the damage then there is no problem. However, unfortunately, this is not always the case.

Paths of degradation
There are infinite possible versions of an article, but there are far fewer paths to a good version than to a bad one. Possible bad versions are immensely more numerous than good ones – an arithmetic progression in quality corresponds to geometric progression in number of states. This relation is corroborated by the notion of entropy in thermodynamics and information theory. It is easy to make a degrading edit. It is not difficult to make a heavily degrading edit. Assuming we do not undo something or use the edit history in some other way, it is almost impossible to improve the article by a random edit. And it is even less possible that an edit on a heavily degraded revision, even a careful one, will compensate for the loss of quality.

So, if we did not use reverts but always modified a recent version, Wikipedia would lose the quality rapidly.

Mass edits and "secondary silliness"
Users making quick edits, such as drive-by ambox insertion, disambiguation resolving, and reformatting, present a special problem when they alter and save degraded revisions of articles. If an ambox appears in a damaged article or section, this suggests that the adjacent content is legitimate. Where degraded content is accompanied by improved formatting and less obvious errors, the degradation is less obvious. In both instances, a subsequent editor is less likely to examine the history for the cause of a problem.

So, users making thousands of edits monthly can actually help to conceal various degrading edits if they do not examine edit histories. Were all or most degrading edits reverted in time, the problem would be negligible. Currently, however, when an article becomes degraded and is not quickly restored, the probability that the next edit will be made by such a prolific editor is considerable.

Conclusions
Obviously, reasonable users should check the edit history in case of any doubts about the actual content—especially before changing anything—to avoid wasting their efforts.

Also, some task force should search (and destroy) recent degradation, but patrolling over such a wide front as English Wikipedia is a challenge. It requires a large number of users whose competence would cover (most of) Wikipedia.

Probably, some tools to facilitate searching for recent degradations would be useful. Russian Wikipedia demonstrated that flagged revisions (for all articles) do not help to solve the problem. The majority of degrading edits are not examined thoroughly and any edit (if not an outright vandalism) promptly becomes "flagged" (i.e., approved and thus inserted into an article). But raising the requirements for an edit to be flagged would result in a huge backlog of doubtful edits.

P.S. I searched Category:Wikipedia administration for an appropriate subcategory, but did not find the word quality. Could somebody create a quality-related category for Wikipedia: space pages?