I'm really glad this has come about, not because the DSM itself is a useless book but because the attitudes towards it lead to some gross errors of judgement.
The DSM can be useful: if one clinician wants to communicate to another at a fairly high level the symptoms a patient is experiencing, then a DSM-defined disorder can be a reasonably efficient way of doing this. Also, the DSM does group together some symptoms which tend to occur as clusters under labels which can provide cues for looking for related symptoms which might otherwise be missed.
However... People make the mistake of thinking that because something is listed in the DSM it is somehow a 'real disease'. The Epstein–Barr virus is a real disease: it is caused by a specific virus. Type I Diabetes is a real disease: it is caused by the loss of insulin-producing cells in the pancreas (although there is the more distal cause of the cell loss). Depression is not a real disease, in this sense - at least, not at the moment. It is a cluster of symptoms which when the occur together are referred to as Depression. Nothing more. (That isn't to say a 'disease' will not actually be identified at some point, but I suspect that will be for a specific subtype of depression, not depression as it is currently classified).
On the radio yesterday, I heard an 'aspie' - who under DSM 5 will no longer be an 'aspie' since Aspergers will no longer exist in its current form - talking about how it was great when he was diagnosed because they finally knew what was wrong with him. The problem is this: they didn't and still don't know what's wrong - just that his symptoms fit a commonly observed pattern, and that there are particular interventions to try to address the associated deficits. Having a listing in the DSM doesn't make things any more or less 'real', but some/many people imagine that it does. Just because there isn't a diagnostic criteria for a very shy child (although I imagine one could be found if looking hard enough), that doesn't mean that there aren't programmes to help the child be more comfortable with social interaction. This becomes most manifestly a problem when conducting genetic, neurobiological, or even treatment research into the causes for 'a disorder'. Because these disorders are symptom clusters, and often have substantial variation in presentation, they are at times artificially grouped for research. This can hinder research into specific subgroups who show more common characteristics. Similarly, if there is a presentation which includes two DSM disorders (e.g. depression and anxiety, which is a very common comorbidity) then these people will tend to be systematically excluded from research because they are defined as 'having comorbidity'. Are both 'disorders' caused by the same underlying cause? Who knows, but being separate DSM disorders means that this group tends to be very underrepresented in research.
On top of this, there is the involvement of vested interests in the development of disorders, there is the interpretation of things as 'wrong' because they are a DSM disorder, etc.
In summary, the DSM can be useful for clinicians to communicate a summary to each other, when accompanied by further detail. It can provide gross groupings for treatment research, but lacks finesse of distinction which could help tailored treatments to individual characteristics rather than the broader presentation. People suddenly seem to think something is 'real' because it appears in the DSM, and so push to have ever more 'disorders' included. This all makes DSM as much of a hindrance as a help to good research and mental health practices.
I agree with most of what you say, but you make it sound like the purpose of the DSM is to allow clearer communications between doctors, and that isn't really the case in practice. That may originally have been one of the goals, but the plain fact is that nowadays the DSM exists primarily to facilitate billing and insurance claims. The big health insurance firms realized long ago that traditional psychiatric therapy (e.g. psychoanalysis) is both extremely costly and open-ended in duration. Requiring a diagnosis from the DSM allows them to cap reimbursement at some set amount and in the process facilitates moving treatment options away from open-ended talk therapy in favor of prescribing medication. Bottom line: when it comes to mental health in the USA, the DSM primarily serves health insurers, not doctors. It is essentially an artifact of our ridiculously inefficient healthcare system.