Maybe you're overestimating the effort because you think of something more complete than I do.

When I think of such a database, I think of the quality standards set by commercial sources now, like for example the POIs I bought with my navigation software. And they are low. Really low. When there's 50 restaurants and cafés near you, they'll show 10. They don't show shops at all, there's no category for them. They don't provide any means to update even the sparse data they have in real time, you get new POIs when new map data is available, usually every 3 months or so. In fact, the whole thing is a mess. Still it's useful for me. It's not complete, it's out of date, but it's good enough for every day use. (FCS, we used to have printed city guides, how up to date and complete were they?)

Community-driven projects on a wiki-style basis cannot get any worse. The aim is not to have an entry in the database for each and every shop in the world. The aim is to have at least some useful data mainly for the more populated areas, and this should be fairly simple.

The only challenge is to overcome the chicken-egg problem: People first need to use a service, and then some of them might want to improve it by contributing themselves. We're not even at a stage right now where people have the hardware to consume the data (=hardware that knows where you are and is always connected to the internet). Some cell phones, yes, but they're new and not wide-spread. The N810 is such a device, too. But it will take more time until there's the critical mass.