What are some common mistakes you notice when proofreading and editing, and how do you address them without over-editing the content?

I've found that I practically overhaul pieces nowadays because there are countless AI-generated pieces that come across my desk on a weekly basis. With that said, some of the common mistakes outside of relying completely on machine-based content are:

• Using the wrong tone for the audience in question.
• Improper use of punctuation (especially the Oxford comma, which I'm a stickler for.)
• Complex and compound sentence structures. 
• Really long-winded paragraphs that repeat the fluff without getting to the point. 

Ultimately, this is why I find myself overhauling portions of text and suggesting that some sections be completely scrapped. Sometimes, less truly is more and we have to help our clients get there. 

Based on what I see, the largest group of mistakes with human texts are simple typos or spelling errors. I am trying not to change the human style of the text unless the meaning is not clear.

On Fiverr the most of my proofreading jobs are machine translations, so sometimes it is about changing awkward sentence structure to the naturally sounding one.

Proofreading is all about picking up on errors and correcting them. However, if you're editing for flow or to correct awkward sentence structures, there does come a point where strict adherence to "what is right" is wrong for the piece in question. For example, the use of capitals for emphasis is frowned on, yet in some cases—such as technical manuals—it is useful.

Define over-editing. The proofreader is paid to correct all ‘hard’ or binary errors, while leaving the style intricacies of the text alone (non-binary errors).

Binary errors are those that represent clear mistakes in the translation, including issues like spelling, grammar, word choice, or syntax. 

In contrast, non-binary errors pertain to translations that are not entirely incorrect but could be enhanced. Examples of non-binary errors include problems with style, register, tone, or collocation.

Non-binary errors could potentially be corrected. But it’s not always the case. As a proofreader, you need to differentiate between what is wrong, and what you think would sound better.

The most common mistakes I found are too complex sentences, incorrect tenses, missing commas, and redundancy. Style is also a big part of it. There are some rules on how the text should look, but in most cases just choose one option and stick to it. Don’t change citing or list style every chapter.

If you want to know how to correct only what’s needed, then definitely focus on clarity and change what’s necessary for readability or comprehension. Try to preserve the style of the author.

That being said, the line between proofreading and copywriting is blurred now, especially with texts generated with AI, which need a lot of changes.

It's never "over-editing" to correct all of the mistakes in the work. However, in more general terms, common errors vary between types of work. For example, in theses and dissertations, many students use the wrong tense in the various sections or may not cite their work properly. In business plans, authors write too much and get too detailed. Writers who use AI or certain automatic grammar checkers often lose their "voice" and sound too robotic and bland. A great editor knows how to address these concerns without overstepping or taking over.