People would argue over the fairness and accuracy of the metrics, resulting in changes and various kinds of work; the system itself would require development and maintenance. It is not a zero-cost solution.
Compared to the effort of compiling an all encompassing encyclopaedia it's negligible.
Especially compared to the price of scaring away a ton of contributors and contributions.
Why does it have to be a zero cost solution anyways?
It's probably that mindset that keeps wikipedia on its crappy centralised php codebase that requires it to beg for money ever other week.
We're at an impasse, as I disagree the effort is negligible. (I think that it could be an enormous time-sink arguing over the relative scores of articles, as well as coming up and changing the criteria used to come up with those scores.) But neither of us have any data, so we can't make any progress down that line of discussion.
To your last question, the argument I don't see much discussion around is that editor time and effort is a limited resource; deleting articles is a way to manage that resource. So any solution that addresses this argument must not incur a large amount of editor time and effort.
I also think a fundamental difference in our reasoning is that I assume a system where each article is edited only by those with an interest in it. Thus all effort and payoffs scale along the same variable. In which articles don't affect each other when it comes to resource expenditure, since the people working on them wouldn't work on anything else.
While it seems that you envision a system that is much closer to classical a encyclopaedia with dedicated editors, where each editor handles arbitrary articles unrelated to the personal investment that he or she has in it. Thus potentially swamping those professional editors with irrelevant articles.
No, I'm assuming the current model of Wikipedia, which is inbetween. In the current Wikipedia model it is not okay for some other pages to not meet standards, even if it's all localized in one subject. That concerns Wikipedia as a whole, and not just that zone. That necessarily means that those from outside of that area of expertise will be involved.
"in the current Wikipedia model it is not okay for some other pages to not meet standards"
Why not? I mean if you mark everything as not up to standard in the beginning and gradually make them as "proper" articles, what's wrong with having crap next to gold?
What is the problem then? There are certainly no technical limitations on storing an arbitrary amount of articles.