Filtering the noise is pretty typical of all crowdsourcing that goes beyond merely polling the crowd. But the fact remains that the crowd has intelligence collectively that the individual does not, and that the collective of wikipedians is more intelligent than its individuals.
It's not a very similar use of crowdsourcing if you are comparing it to voting systems (and it seems people claiming these applications don't exist have a pretty slippery definition of what they are talking about) and you are certainly right that the management of the community ends at a core of admins but that doesn't diminish the collaborative intelligence that the wikipedia accumulates.
In that case, the whole together provides more wisdom than any one individual and if the group were a lot smaller, they would collectively provide less information.
That's why they can overtake the coverage of an Encyclopedia with a smaller group of editors like Britannica. Because they take teamwork to webscale. There are plenty of flaws in their system, but it is
an example of wisdom extracted from a crowd. The crowd doesn't have to directly control the content at all to be an example of the crowdsourcing that ebrown is claiming doesn't exist.