I probably should write a few lines about this, since the question was obviously a reaction to my edits. (But it will be certainly nice if some more experienced users share their view.)
I consider improving some posts on MO as a useful form of procrastination. So I hope I will procrastinate less in the future and, as a consequence, my edits will annoy less users around here.
More often than not, when I edit a post MO I tend to leave a brief edit summary. I take it that in these cases the edit summaries were not clear enough to see why I considered the edits useful. (And I hope that at least in some cases, if I hadn't edited those posts, somebody else would.) I will also add that quite often I visit MO editors' lounge, so if there are some issues with my edits, you can also contact me there. (This room was created for discussions related to editing and tagging, it was announced here by quid.)
Let me respond here to some questions asked above and also write a few lines about my views on editing questions on MO (or Stack Exchange sites in general).
In my opinion, it is better to edit questions while there are new or if they were recently active (i.e., they have already been bumped for some other reason) than bumping old questions - the latter should not be done too often. (Meaning of "too often" probably depends on the site's culture, on the traffic the site gets, and many other factors.)
You have linked to an older post here on meta which contains a recommendation not to edit more than three old questions per day.
I am aware of the linked post and I try to limit my edits even more than that (typically not more than one edit to an old unbumped post per day).
The linked examples are mostly questions which were posted recently or were already active for another reason. It is better if a question is edited in such situations than if somebody notices only much later some reason why a post should be edited and bumps it again in several months or in a year or two.
(The first example in the question was an edit to an old question, but it was a post which was recently bumped by adding a new answer. Probably I should have indicated in the edit summary that this post has been bumped so that this fact would be better visible. Here is another example of my edit where I explicitly mentioned in the edit summary that the question had been bumped - in that case it was by adding a new answer, too.)
For me it is little annoying to see the same name on most of the questions/answers as "edited ** ago".
Probably the best solution to this that I can think of is that more MO users help with editing and retagging. (I see that you have already started and I certainly appreciate your efforts.)
There are various types of edits which can be considered useful.
Personally I would consider even grammar corrections and correcting typos beneficial. The sentence "there are various reasons why I edit posts" is easier to read than "there are various reason why i edit pots". Although such a minor edit would not be a sufficient reason to edit an old post.
There was a separate discussion about adding links to the material referenced in MO posts, and judging by the response to that question, such edits are also considered useful: Is it worth editing old posts to add links for references?
I consider also retagging questions which do not have correct tags rather useful.
Correct tags are useful for several reasons. They help when searching the site. (I think that at least some of the questions and answers might be useful also to users other than the asker.) And they help also the poster of the question - since people following the added tags are more likely to notice the question. I don't think I will summarize it better than what is already written here: What are the advantages of proper tagging?
One of the specific features of the tagging system on this site, is that every question is supposed to have one of top-level tags, see also: What are top level tags, and why should I try to use them? and Why are MO tags formatted as they are?. (Again, one of the edits mentioned in the question is of this type, I have explicitly mentioned this in the edit summary.) Although at least in some cases finding top-level tags might be problematic (see, for example: On the top-level tag ra.rings-and-algebras and Which “non-arxiv” tags are de facto top-level tags?). Regarding top-level tags, I will quote one of the MathOverflow moderators:
Yes, the tag management on MO is very different than on Mathematics. There are several reasons for that. For example, the systematic use of broad area tags borrowed from the arxiv on day one brought a lot of stability to the general tagging system here.
The top-level tags have been also mentioned to you recently in the Homotopy Theory chatroom:
You don't need to put only one tag, you can put up to five. In general it is recommended to put a very general tag (usually an "arxiv" tag) to indicate broadly which sector of math your question is in, and then more specific tags.
With respect to tags, MathOverflow has one more unresolved issue, which is deprecated tags. There are several tags which should be eliminated, which cannot be done without some amount of manual editing. It seems that several edits mentioned here were of this type - and also the edit mentioned in this recent post on meta: Restriction on editing closed questions. When I do such edits, I usually mention this in the edit summary (as I did in those cases - 107989, 242173). Considering the fact that the number of the questions with deprecated tags decreases at the rate about 10 questions per month, it does not seem very likely that they are going to be removed anytime soon. But perhaps at least something could be done to prevent this number from increasing so that these tags cannot be added to new questions as I suggested elsewhere.
Personally, I consider use of tags especially important on meta - since many issues mentioned there tend to come up repeatedly and tags help a lot when searching and also when populating the related question lists. To include an example, I will use something from another site that I am more familiar with. One of the issues that repeatedly come up on meta is that comments containing long MathJax expressions tend to be jumbled sometimes. On Mathematics Meta, I can simply use the search using combination of the two relevant tags, and I can easily find several past discussions about this issue. (For comparison, on this meta I was able to find two posts about this issue, at the moment neither of them has an answer: The math! It's overflowing! and Long MathJax in comment breaks linewrapping throughout the comment thread.)
Even though some of them are mentioned above, let me go through examples mentioned in the question one by one.
- "Book about fluid dynamics asked Apr 13 '10 edited 1 hour ago" - since this question has been bumped anyway by adding a new answer, it seemed rather natural to look whether there are other things that can be improved by editing the question or some of the answers. The tag books was used there incorrectly, at least if we follow the guidance given in the current revision of the tag-info. I have included link to the tag-info in the edit summary. (Let me also say that I have asked a few times in chat about the (books) tag, since it seems that the tag is often used differently from this guidance.) I guess that when somebody searches for a question like this, they might typically use tags fluid-dynamics+textbook-recommendation or fluid-dynamics+reference-request, so the change of tags can be useful for searching.
- "Are there examples of conjectures supported by heuristic arguments that have been finally disproved? asked 21 hours ago edited 2 hours ago". Here I have added (conjectures) tag, to me it seems as a question that one would hope to find when searching using this tag. It was a recent question and it was really active (it even got in HNQ). I would not consider the fact that I bumped this question much of an issue - some 10 minutes after my edit it was bumped again by an edit to an answer. Moreover, the question is a community wiki.
- "Division ring on a field asked Jun 14 '16 edited 2 hours ago" - among the questions listed above, this is the only old question which I bumped. In this case it was to remove the deprecated abstract-algebra tag, which I have also mentioned in the edit summary.
- "Halin Graphs with Highest Number of Hamilton Cycles asked 8 hours ago edited 5 hours ago." As mentioned before, it is useful when question has also some of the bigger tags, not only very specialized tags - which is why I added the graph-theory tag. You may notice that after my edit, the question was further edited by other user and then by OP. (Maybe I should have also added a link to the paper mentioned there to make this edit more complete. On the other hand, the post has a link to a Wikipedia article and this paper is there among references with a link.)
- "Applying a piecewise linear function to vertices of a polytope while remaining in facet representation asked yesterday edited 6 hours ago" - in this case I was adding a top-level tag, as I explicitly mentioned in the edit summary. Again, this was quite a recent post and since it was bumped a few minutes before my edit, I do not think my edit caused any harm.
- "Is every metric continuum almost path-connected? asked Aug 14 at 6:49 edited 18 hours ago" - in this case the text of the question indicated that the OP wanted to link to another question and they linked to an answer, most likely by mistake. So I have edited link and again mentioned this in the edit summary. The subsequent edit by the OP suggests that this is probably what they wanted.