Each site has some leeway in this but in general, the questions, as that quote states, should be largely the same question.
If you asked the same question about the US or India, they'd quite clearly not be duplicate questions because the cities in each case aren't the same, so you wouldn't even be having this discussion in the first place.
So, there are a couple of alternate examples here that might be useful to keep in mind:
- What is the most populous city in France?
- What are the top ten cities in France by population?
I think you'd argue that these are duplicates, no? They're about the same subject, the same metric, one is broader than the other but they are the same and number 2 is more useful to a bigger audience because it has more information in it.
So, what if we compare question one with question three:
- What are the capital and most populous cities for each country in Europe?
These, too, are similar enough to be useful to a wide audience in a way the individual questions for each wouldn't be - they'd even be arguably less useful if asked individually because you'd have to find each one (and ask a new one if one didn't already exist), so sometimes grouping information is helpful - Wikipedia even has such a page for all countries.
On Stack Exchange, this would be called a canonical question. They can be difficult to get right but they're useful to sites that get many very similar questions that are generally answered the same way or that make more sense as a collection because of how related they are. For example, on Cooking, instead of having one question for "What's the difference between biscuit in American English and British English", there's a helpful glossary of differences between English variants all in one post.
Admittedly, in some cases, canonical questions can get too big and may be overwhelming for users, so it's important to balance the breadth of the canonical question with ease of use. If it's going to take 20+ minutes to sift through a giant answer to find the exact thing you're looking for, then it may be better to divide things up.
When it comes to an answer on one question being broader than the question needs and answers another question - as in the example:
Q: What is the capital city of France?
A: Paris is the capital city and also the most populous city in France.
If this is going to be a dupe target for "What is the most populous city in France?" I'd much rather someone edited the question to ask both than just outright closing the new question as a dupe. Why? Because that leads to a lot of potential for overreach with duplicate closures.
With this example question, this is very simple because both the question and answer are very brief, so handling both questions simultaneously is going to be more useful in the end than having two separate questions. It may even eventually turn into question three above if this one question starts spawning identical versions for every country in Europe. Yes, we should consider whether editing the question would invalidate existing answers or not... keep reading...
If someone asks a specific question that's complicated and needs a longer answer and someone goes out of their way to write an exhaustive answer that covers tons more information than the question asks for they may be obscuring the answer to the question asked in their long answer, making it less useful to the OP and others coming after and they may also be over-answering - broadening the question beyond what would normally be OK on the site.
Stack Exchange thrives on having multiple answers to the same question - we even look for it in determining which sites are ready to graduate from Beta. If someone asks a question that is answered in another, but only in one of several answers because that wasn't the intent of the original question, there may be alternate or even better solutions or explanations that are never addressed because of it.
This is what I encourage duplicate close voters to be wary of - they should ask themselves:
- Is this information so related that failing to include it does the question a disservice and most if not all of the existing answers already address it?
- Is this information added to an answer because it's an interesting related piece of information and only one person thought to mention it?
If it's the former, then the older question can probably be edited to include the new question and all of the best answers (rather than a single one) should have already taken it into account. If it's the latter, we're forcing the person asking the new question to filter through several answers to find the one that actually addresses their question and it may not even do so adequately. In this case, the questions should not be considered duplicates.