I'm a brand new member here, and after only a couple of days worth of looking around (and trying to understand all the various systems), the thing that bothers me the most is the lack of clarity on sourcing, what's considered well sourced, and what isn't, and what the heck is the difference between unsourced and sourced, anyway?
To start at the beginning, WikiTree has a HORRENDOUS reputation among Swedish genealogists. Because of poor referencing. As I'm currently in the process of tidying up my own rather dubious research, I thought that joining up here might be useful as it'll help tidy things up (and many other reasons, too, of course). That said, as a new member, I'm more confused now than ever about what is acceptable and what isn't.
I dabbled in academia for a few years, and as such I now have a very clear idea of what constitutes a source and what doesn't. I referenced a compiled volume of Greek mythology in a paper which I presented to my supervisor, who promptly told me to go back and find the original sources. That meant digging out the LOEB edition of Apollodorus (1st-2nd century CE) and citing that, with direct references to the lines in the poems.
To me, good genealogy = good research practices. Good research practices are to cite sources for everything you do. This doesn't always come natural to many people who don't have academic experience, but can certainly be taught and encouraged (heck, if I could learn it...!). Allowing sources such as Ancestry trees, references to wikipedia, and other continuously changeable databases is futile as the first and foremost consideration surely must be that the research must be reproducible by someone other than the author. That means that, regardless of the quality of the source cited, the next person that comes along should be able to find the referenced source. A link to another website is subject to change by the whims of the website owner or author and thus does not constitute an acceptable source.
On-line books are better. They would have been scanned from an original that could presumably also be found in a library. If the book was found online, then that should be cited, but the principal source should be considered the book itself. Books, when published, go into large public databases, get sent to libraries, and so on. This means that someone 300 years from now could look at your research and (preservation permitting) be able to find the same books used in the research.
Original, primary, sources will invariably be accessed on-line by most of us these days. A reference to the actual volume as recorded in the appropriate national archive should be the very minimum, with a link to where the material was accessed also necessary. Even better would be a screenshot of the appropriate part of the document. Unless the national or regional archives burn down, future researchers should be able to also find this material.
The next overarching consideration ought to be the quality of the source. As I learnt the hard way, if there's a primary source available, then use the primary source. In this context, a primary source is a document that was written at the same time the events it's used for occurred, or that are recorded by an eye-witness. In genealogy covering the modern period, this means that one should always cite the original document rather than the research of another person. Thus, if you're grabbing a fact from Ancestry or myHeritage, instead look up the source used in the tree you're looking at, and then cite the original document by sighting it yourself first. If the tree has no source cited, then look at the facts given and follow up from there in the primary documents.
Citing printed books as sources for this period is fine, but they should be treated as secondary. If that is the only source you have available, then it is better than nothing, and will allow other researchers to follow up and make corrections later on. It becomes common in the period between, say, 1500 and 1700 to use a WikiTree relevant example, that a lot of research has already been done and published. I see nothing wrong in citing this material, but I would still at least attempt to look up any sources that have been cited, and make a note of facts that appear unsourced or that I've been unable to find supporting evidence for. If the book is out of copyright, I would also frequently prefer to cite the whole thing.
Citing books and academic research papers (secondary) becomes inevitable when you move beyond the 1500:s. Charters and "diplomata" are only really available in their originals to academic researchers in national libraries. Many are printed which is a good thing, and these days more and more are appearing on-line so there may be fewer and fewer excuses to not cite these. Nevertheless, in the medieval period, many commonly accepted genealogical connections are essentially nothing but academic speculation that's garnered more or less acceptance. When establishing such a profile, the whole argument should be sourced, and the diverging opinions covered off in support of the connection made. I suppose this is why there's a Pre-1500 certification here (which I hope to obtain eventually), which is certainly a good thing.
So to come back to the original question: what is unsourced vs. sourced? Well, I would say it depends on the period we're talking about. What is certain, however, is that I believe that WikiTree needs:
1. A clear and strict definition on what qualifies as an acceptable source.
2. Appropriate categorisation of sources used (i.e. a distinction between primary and secondary sources, along the lines suggested in other answers above that would add a further "needs more" category or similar). This would help identify where improvements can be made.
3. Clear and strict definitions of when a profile can be considered sourced vs. unsourced.
It should be fairly obvious that I take a hard line in this sort of thing, but surely it must be better to have clarity - after all, the better referenced WikiTree is, the better its reputation will become. That, I believe, is a good thing.