I think it’s not unethical in general, even if there could exist cases when the decision could be poor judgement, hence diminishing the credibility of the authors and trustworthiness of their work. But this is not about ethics.
As said in a comment, I don’t expect authors to have read all the cited works (each wholly, instead of partly, or indirectly…). In the same time, when I read a paper and find a specific claim with a reference, say, The fact Y holds [X], I expect X to be the paper or book supporting the statement. I am not really interested to know whether the authors have read X at all: I have only to know if X supports the claim or not.
Nonetheless, what if I read X and find that it is misquoted? Or if I discover that its results are misunderstood and that they do not support Y?
Since I don’t know whether the authors haven’t read X and they are instead trusting (wrongly) another source I can’t trace back (since they missed to provide me with the proper clue), their credibility crumbles in front of my eyes. In the same time, the citing paper loses all its supposed value.
Thus I can see only two reasons why authors should specify that the claim is supported in X and that “the claim is supported in X” is stated in a paper Z (and that this is how they know, or believe to know that Y is true):
- to aknowledge authors of Z for their work (in fact, I suppose there’s a reason if the authors have read it);
- to keep every link of the “chain of trust”.
In my opinion, only the former has something to do with ethics, while the latter has something to do with how scientific (and non-scientific) knowledge proceeds. I won’t elaborate on this “chain of trust”, but think about few things: If Z (containing the claim supported by X) was peer-reviewed, would you feel suspicious when you read The fact Y holds [X]? Would you act like a reviewer or just like a user assuming Z already proved to meet the necessary criteria to be considered trustworthy, in general and so in particular about that specific claim and the references? Would you instead go and check and verify all references, and then all the references contained in the referenced works, and so on? Would you stop at level 2, level 3, level 4 or level N before you think something is trustworthy enough? And if the paper is not peer-reviewed at all, what does it make trustworthy to you in first place?
When you decide to write The fact Y holds [X] and skip the link with your primary source, you are making a precise bold judgement: The link you are skipping is 100 % right, errors-free and nobody needs to check it (or even to know you relied on it). This is problematic for you, as an author, but not for any reader: You are not cheating “against” the reader, since to judge your work he only needs to know if X really supports Y (in turn, if the reader trusts you, he might decide you are not lying, unintentionally or not – another link in the chain), and you give him the reference, so he knows where to check for the claim. But, skipping the link Z, you bring on your shoulder the burden of anything Z does, mistakes included.
Therefore I agree with anybody saying it’s good practice to be accurate and let people (and reviewers) know The fact Y holds [X] is written in Z, and that this is your primary source for the claim. But I can’t see grounds to say that if you don’t, it is unethical – rather, it is simply in your own interests to do so.
In your case, there’s no Z, if I’ve understood correctly. It seems there’s something like a common knowledge, i.e., a collection of sources agreeing on the fact Y and suggesting (or stating) that the claim is supported by X. Your paper adds one more to this “cloud”, eventually growing the loop and a possible bubble — bad approach (especially if actually the author hasn’t picked one of those sources at all), but isn’t unethical to me.
(Let me say that if Y is something like James Watson and Francis Crick discovered the double helix structure of DNA and the book is The Double Helix: A personal Account of the Discovery of the Structure of DNA, I would not say that it is a bad approach. So, it might strongly depends on what we are actually talking about.)
I hope my answers, while touching already cited themes, adds few sparks to elaborate over the reason why it should, or should not be considered unethical the behaviour the OP describes.