Should we avoid changing canonical question titles so as to preserve their recognizability?

Question

I just reverted the title of the Accessing ArcObjects from Python? question from a recent revision which substantially changed the question title (to the point where I honestly did not recognize it). This is one of the top questions on the site and is by any measure a canonical question.

In my opinion, the previous title worked well, was recognizable, and did not need clarification (I don't believe that anyone would have confused arcpy with ArcObjects as they are clearly different APIs, with the former being a small subset and extension of the latter, but are otherwise incompatible).

Can and should we restrain ourselves from the temptation of revising titles for canonical or "historically significant" questions?

Answer 1

I'm no more sentimental about titles than I am about preserving the original body and tags of a question.

I think the most important thing about the current state of a title is whether it presents the best chance of a visitor being attracted and becoming engaged to read its all important body and answers, and finding what it advertises there.

I have no problem with the title of any question being changed, as long as that change is an improvement when judged against the above criteria. In this instance the change I made may not have met that criteria so it being rolled back is no problem, but I do not think that it should stop us from continuing to think about how it might be worded better.

As some background as to why I thought that question would benefit from some editing, my recollection is that it came out of my looking at:

• FAQ for python where it sits, I think, appropriately at #1
• FAQ for arcobjects where it sits, I think, appropriately at #1
• FAQ for arcpy where it was at that time sitting at or near #1 and potentially keeping another arcpy question that might be of more value to more ArcPy developers off the front page

I was grappling for a way to reduce the chance of it being retagged to arcpy and explicitly excluding it from the title (to reflect what was written in the body) seemed like the way to go.

Perhaps a title like "How do I access ArcObjects direct from Python?" (which my preference would be to write more concisely as "Accessing ArcObjects direct from Python?" :-) would have been better.

Answer 2

I suppose that depends on the question. If a canonical question title is very specific, yet the answers apply to a broader range of questions, changing it might make it more accessible to new users or searchers even though if you remember the question it'll be harder to find.

I recently changed the title of a question that specifically asked how to find the area summary of each class in a kernel density analysis. However the answers, and in fact the question itself really applied to any raster, not just a kernel density. In this case, the change was in my mind to make it a better candidate for a canonical question.

But if the question is already canonical by any such 'extreme' measure, I would expect that it had already been edited to a point where further title changes were unnecessary. Not to say anything couldn't be improved, but sometimes it feels like we're really making trivial edits that don't substantially add to or correct anything.

So no, I don't think any question, including canonical or historical, should be immune to edits just because. But at the same time, some consideration should be given to how warranted the edit really is (perhaps more in such a case - the more historical/canonical a question, the more significant an edit should be).