> In fact in most cases they barely read the work with any substantial sense of "read".
I would not say this applies to most cases.
I have got many papers reviewed, have done many reviews, and have got questions from people doing other reviews, and people usually put quite some effort into it. There are exceptions, but bad reviews are not the norm, and you can usually catch them when it happens and talk with the editor about it (either if you are the author or another reviewer).
Searching for similarities should indeed be part of the job, but it is not always easy, and I agree more work should be done in that aspect.
This may be different in other fields or specific journals but, please, don't generalize.
I would not say this applies to most cases.
I have got many papers reviewed, have done many reviews, and have got questions from people doing other reviews, and people usually put quite some effort into it. There are exceptions, but bad reviews are not the norm, and you can usually catch them when it happens and talk with the editor about it (either if you are the author or another reviewer).
Searching for similarities should indeed be part of the job, but it is not always easy, and I agree more work should be done in that aspect.
This may be different in other fields or specific journals but, please, don't generalize.