We diagnose individuals through opinion rather than any scientific test, which is unreliable. The BPS Division of Clinical Psychology states that "due to their limited reliability and questionable validity, [psychiatric diagnoses] provide a flawed basis for evidence-based practice, research, intervention guidelines". What you've said makes sense if diagnoses were reliable and valid. However, you cannot guarantee that one person will always receive the same diagnosis by every doctor (despite reporting the same symptoms and limited changes in their presentation). So if we are deciding what treatment to give individuals based on the diagnosis, I feel it would be important that the diagnosis is consistent, and not open to bias.
I would argue that you can still provide meaningful input without a diagnosis. Peer Support Workers use the recovery model (alongside other professionals) and this states that a person should be seen as an individual other than their diagnosis, and many peer support workers prefer to abandon diagnosis and talk about their experiences (a shift from "what's wrong" to "what's happened"
. Peer support has been shown in research to reduce hospital admissions and increase service user satisfaction compared to a traditional approach.
With research, I feel there is still scope to strip it back to experiences. For example, if you are testing a new anti-psychotic drug, surely you could look to recruit participants who experience sensations/noises/sights etc. that no one else experiences, or individuals who believe strongly in things that others do not believe in, and perceive as odd and out of character. That would include everyone with those experiences, rather than individuals with a set diagnosis. It could even benefit research as it could allow individuals who share these experiences but have a different diagnosis to participate in research, which could enhance the findings. Not all research around interventions requires diagnoses, for example, social work research in mental health nearly always talks about experiences over diagnoses. And it is still possible from this to decide on appropriate intereventions.
In response to your "random number" bit - I have nowhere suggested that you just randomly select an intervention or treatment. You can still select a treatment without looking at diagnosis, instead, choosing to look at a person's experiences.
If there is a way to select treatment/intervention which increases the individuals empowerment, allows them to state what they feel their issues are, reduces the amount of stigma/discrimination they will face and is more reliable than diagnoses (which for some can change frequently) - then isn't it worth thinking about changing?