Recently I corresponded with a friend, an established scientist, who had served as a special-issue editor for a prominent, for-profit, online journal. He and an editing partner were responsible for setting the special issue's scope, soliciting manuscripts, substantive editing, and so on—the usual editorial tasks. They didn't get to see the final versions of the articles until they had been assembled into an electronic book. By then the articles had been published online already, so it was too late to make changes. They had become a part of the scientific literature.
When he saw what had been published, my friend wasn't happy. He and his coeditor found more than 100 small errors: misspellings, incorrect grammar, and so on. His coeditor's name was misspelled on the title page. (They were able to fix that mistake.) Apparently, copyeditors introduced some of the errors.
My friend took the publisher to task for the lousy copyediting; the authors deserve better, he figured, for the $1000 or so per article they paid for publication, and copyediting was supposed to be covered by the publication charge.
But the fault doesn’t lie entirely with the journal. The editors may not have seen the proofs, but the authors did: Why did some of them sign off on error-riddled manuscripts? Maybe they assumed those errors would be caught during a subsequent proofreading stage, or maybe they were too busy to take a final, careful look. Some scientists may even assume that because the article will be published online, it will be easy to fix those mistakes later. But the official scientific literature—mistakes and all—is sacrosanct; errors cannot be fixed so easily.
Preserved in perpetuity, such errors cast doubt on the reliability of a paper's scientific content. They make a bad impression on hiring, promotion, and grant-reviewing committees. Literature riddled with trivial errors could even undermine public confidence in science.
As science continues to move away from print and toward online publishing, with faster turnaround, fewer resources invested in each article, and less and less editorial intervention, errors like these are becoming more common. (This is not entirely a new phenomenon in scientific publishing. For a 758-page book I coedited in the mid-90s, I didn't just do my own copyediting; I laid the whole book out in LaTeX.)
One could argue, as my friend did, that publishers have a responsibility to their authors and to science to uphold a certain standard. But, though it depends on where they publish, scientists cannot assume these days that an editor will catch their misspellings and casual grammar. Scientists themselves must take responsibility for ensuring that the papers they publish are clean. That's yet another specialized skill that today's scientists need to master.