We can reuse previous attempts to document a given topic by putting them in the "Draft" namespace, instead of losing it to a delete. If we archive previous drafts as we do with talk pages, the draft archive can be a good starting point to find information (prose, lists, sources...) that other editors added to the project in a way that was not ready for the main space.

Holding a large amount of low quality content in Draft space is not a problem - that space has been created precisely for that, as all its content is sub-standard! Its content doesn't hold more nor less weight in the database being hidden than visible, and there's no benefit in having it accessible by only a few selected elite.

We are expected to keep the process of creating the encyclopedia in the open, to expose our bias and allow anyone to assess by themselves how content is written; that won't happen if we keep hiding under the rug everything that ashames us. There's no backlog for cleaning up drafts and it would be impossible to fix all of them, with every one being a stand-alone repository of content; nor it's reasonable to expect that one can be improved to the point of being viable within any limited period of time. But by keeping it around in a place where it's stored out of the way, it can eventually arrive to the point where it can be used within an article, if enough improvements are accumulated during an arbitrary period. But if you hide verifiable information that is not problematic, it's impossible to accumulate knowledge through a slow and steady process.

Wikipedia needs in order to advance, like any other complex information project, to move a copy of its "stable" version status into a "broken" unstable version where things are allowed to change. Software projects work that way, for a very good reason: you can't make major changes without breaking something along the way, but that doesn't mean that you shouldn't make major changes ever - just reserve a dedicated space where you can experiment, whether you call it "development branch" or "draft space". For the first time in years, we have a space that is not strangled by bureaucratic rules or veteran editors camping to own articles; exactly the same conditions that allowed Wikipedia to grow and become popular in the first place. Let's use them to counter our bias, allow new editors to settle and overcome the intertia that has stalled the project.

Deleting drafts that hold viable content under a time limit makes no sense: the content is not really lost anyway, and it's still occupying content in the Wikipedia servers; the only practical effect of deleting a page is cutting access to the content from the people who could use it and improve it. If the purpose is to build an encyclopedia, the mechanism is iterative improvement; any deletion of valid content hurts that purpose. This artificial barrier of asking for a restoration of the content, by introducing a human in the loop, makes it more difficult to recover the content, impedes the readers from assesing by themselves whether the content is relevant on the spot, and introduces a delay which can go from hours to days; therefore, assessment of a large collection of drafts is incredibly hard.

There should be a very good motivation to support hindering editors in that way, and doing it is usually not justified. Draft pages are not indexed nor appear in search results.

Editors usually underestimate how much it hurts the project the simple fact of removing content from editor's view. Even if a page couldn't ever describe a notable topic and be improved into a stand-alone article, its content could still be reused at other viable articles, or used as a starting point for editors wishing to investigate - allowing them to find reliable sources for a related subject. The threshold imposed by the Foundation for hiding content is quite high - only content that is harmful to living persons, or that would put the project in legal risk is required to be hidden. Let's use the new Draft space to avoid losing everything else for no reason at all.

Keeping informative content that could appear in an article, even if in bad shape, has traditionally been the way in which Wikipedia has been written, as reflected by the WP:IMPERFECT and WP:NOTPAPER policies. Although there are rules allowing some kinds of drafts to be deleted, such as those in user space and the Articles for Creation process, those don't necessarily apply to the Draft space. The old processes assumed that drafts were mainly a one-man effort, with a main contributor pushing the content into main space and other editors making tweaks. But the Draft space can be used as in 2001 Wikipedia, as a blank slate where anyone can contribute until a fully developed start-class article emerges.

Current enforcement of policies and guidelines is great for keeping the main article space clean and stable, but they're also holding the project growth back and driving new and existing editors away. The project to write an encyclopedia could use a place where the mentality of Article space where "only good, 100%-policy-compliant content may remain" is not applied. In this "beta quality", "unstable" version of the encyclopedia can allow under-developed areas to grow (Asian famous people, African culture and infrastructures...) and new users to experiment with new ways to develop articles, without being bitten by the bureaucratic set of rules that is currently in place.

Removing bad drafts based on the quality and suitability of the content does make sense, though; if the page is a personal recollection of stuff that could never be adopted in articles, it can be safely proposed for deletion and removed from view. But if the content could reasonably belong in the encyclopedia, just because no editor expressed an interest in improving it now doesn't mean that no one will find it interesting later, some months from now or two hundred years in the future. Wikipedia's nature as free content makes it intended to last forever, and an abandoned draft could very well provide whatever hints a scholar from a future civilization needs to kickstart a research project about our knowledge on an obscure topic. So why make iterative improvement harder by removing viable content from sight, when that content is not featured as a Wikipedia article?

Some editors have concerns that a permissive policy to keep drafts would pollute the namespace and make it unreadable. But the same policy (to keep almost everything) is applied at talk pages, and the problem has been kept in check. The same recipy could be applied to drafts: remove only content that couldn't possibly apply to building an encyclopedia, keep everything else, and archive old content on the basis of volume, not antiquity.

"Bad drafts" could still be deleted subject to the common deletion processes like speedy deletion of duplicated content, procedural clean-up, deletion discussions with a consensus, and breaking the rules of copyright and biographies.

Drafts can be usually found from the mainspace page of a deleted article. I.e. an interested editor still can found the desired information, and even work to fix the incomplete content.

It may be too soon to write an article about a new product, movie, or book, but it can become notable when it gets popular and reliable sources start writing about it some months later. As the draft space is not subject to notability requirements, it can be a good place to park those articles that might or might not become famous in the mid term.

If an article fails on notability, that is not a reason on itself to hide the content from view. If an article makes a credible claim of importance of the kind required to exclude it from speedy deletion, there's no benefit in deleting it; WP:PRESERVE is as much a core policy as WP:NOT, and Draft is the perfect place to keep around "facts or ideas that would belong in a finished article". This already exclude all garage bands and vanity pages (it wouldn't harm if we remove those from view, but those do fall under CSD); the administrator evaluating the PROD request has leeway to decide whether the claim is not credible and therefore the article can be speedily deleted. When the claim is credible, or if there are doubts, the article should be subject to a discussion in order to hide it from view. There's nothing to gain, and much to lose, by hiding the viable content as a deleted article; the only reason to delete such content would be blind adherence to rules, and we know that mere following rules because they are rules is against the rules. You always need a good reason to make rules, and you need to check that this reason applies each time you follow the rule.