KickStarter - Most of the successful (funded) crowdsourcing campaigns are from here.
IndieGogo - The 2nd best platform, with less strict rules and guidelines.

I doubt something of this scale could be pulled off without the software already done, or without a major backer.

The smart way forward would be to use IndieGoGo, and develop an Android devices that caters to the needs of N900 users. However, to make the device "open" so that it can be dual-booted with Mer, PlasmActive, Ubuntu, openSUSEARM, etc etc.

Since expertise is the limiting factor, I'd advise against another "Ubuntu Edge" scheme. Instead focusing efforts on perfecting the design, feel, quality of the device. Putting emphasis on things like:
- high-quality microphones with noise cancellation technology unmatched by competitors
- Frontal Stereo Loudspeakers that beat the competition on clarity, bass and volume
- OtherHalf concept (detachable keyboard !!)
- etc etc

No need to go spec-whoring. Something like a 1.5GHz Dual-S4 PRO is heaps powerful and has LTE (just get driver source for 3rd party software). And a 2,000mAh battery is more than adequate (the iPhone 5 has a 1,500mAh and lasts as long as Android devices with 2,500mAh units). The key here is quality, not quantity.

At the end of the day, you want to make something unique but something that appeals to mass market. And it needs to be priced respectfully, like $500 to be close to impulse buy range (unlike Edge's $700+ demand). And nothing too out of reach of accomplishment; don't ask for $15Million / 30,000 orders. Something more sane like 5,000 orders / $2.5M is more likely to succeed. And don't promise to deliver in 3 months, something like 9 months should be minimum.

And remember, a crowdsourcing project is not a pre-order.... but today's consumers treat it as such.
If a project is successful, and the backers are provided with their order... listen carefully to criticism. The finality of the project shouldn't be a finality at all; it should be the beginning. The start for your chance to up the ante and deliver the same idea without the shortcomings and with the ideas you couldn't implement in v1.0
One example would be to polish up quality and QC for second batches, stuffing a better SoC/battery/components and delivering true Linux out of the box (and dualboot Android for those who want it).

All in all, its a difficult thing to do.
You need a dedicated team of smart entrepreneurs, lots of brainstorming, lots of time, lots of money, and a dedicated fan base to pull something of this scale off.