Well, I don't think that the real discrepancy lays between IPs and users. Registration is purely a matter of formality : as you get acquainted with Wikipedia as an IP, you are likely to contemplate the advantages of having a unique identity. It is not that IPs never write FAs, but that, most of the time, they finally go through the registration process before finishing their work.
Yet, we can perhaps draw a reliable distinction between occasional and (relatively) permanent editors. The first one are rather passing by, writing only the stuff they are interested in, going away from wiki once they feel tired of it, possibly coming back whenever they feel like it. The second one go extensively beyond their initial scope of edition and get to assume the current general affairs of the community.
Therefore, does occasional editors matter ? I should say yes. For instance, on the French wikipedia some new editor did a brilliant job on [[Napoléon III]]. Once the article became an FA, he stopped being active : to him, its main, ponctual, work was over.
On Wed, Oct 31, 2012 at 5:04 AM, Piotr Konieczny <firstname.lastname@example.org>
Well, this is based on my experience
as GA author and reviewer. I have never seen an IP successfully
nominate an article (I did see nominations once or twice, they
failed quickly, as the articles were not up to GA level and IP
never came back). And of course, I have yet to see an IP GA
reviewer (that is not a troll or a useless if good faithed
newbie). If you are aware of any successful GANs were the primary
author was an IP, I'd like to look at them. I'd hypothesize that:
But that wasn't the point raised. The point raised was not: can IP addresses successfully navigate Wikipedia process? It was: can IP addresses successfully create content? If you're focused solely on process, then yes, correct. You will not see IP addresses engaged their because the rules generally prohibit it.
On the other hand, if the issue is can ip addresses create content, then it appears to me, yes, they can create content and do so successfully without getting their content rolled back. They are an important group. in the area I write in, between 10% and 35% of all edits to articles appear to be made by IP addresses. (Most of them based on the regional interest for the topic.) They often include information that has improved articles and can learn sourcing. This is not always the case, but happens often enough that their editor value should be considered.