First... I think through a web client would be optimum.

Reasoning:

If an elgg site takes off enough, they can code their -own- apps for the iPhone or whatever. 

Thus, this is something that would take time away from the core development, which could be better spent on things like user management. 

Secondly, not everyone has an iPhone. Not everyone has a Blackberry. Ideally I'd want maximum usability. That means mobile web.

Especially since I myself have an N97 and am more likely to spontaneously become a woman than to buy an Apple product.

I do agree it should be generic as much as possible and web would be preferred.  A different stylesheet can then be applied to specific models.  I listed the iPhone and Blackberry for selfish reasons - they are the supported platform in my organization and I am thinking that to advance Elgg, we have to facilitate ease of use on the most popular platforms.  As far as priorities, I would leave it up to the coders who would be contributing to this effort.  I may take a crack at this but I don't think I'm up to par with the many of the community contributors on this platform.

Elgg's views system makes this extremely easy.  We could simply add a view called "mobile" and then check against a user agent string or something to detect mobile browsers.

Brett, I have to disagree to an extent there, but just with the word "extremely." It makes it -relatively- easy. But once you start having plugins, you have to create a view for each and every plugin, which feels very daunting if you weren't the one to create those plugins.

Of course, if everyone comments their code properly, it would be a piece of cake. :p

Thanks for all the feedback so far - its "extremely" encouraging ;-)

I've been looking into developing an iPhone App for Elgg but the word on the streets is native apps are out in favour of webapps. These run in Safari on the iPhone and Elgg would host all the functionality. You can even get templates/javascript that make it look like a native iPhone App. Needless to say, you could tailor the view to different mobile platforms. I took the view of splitting each part into its own app, starting with blogs. You can surf Elgg on Safari on the iPhone but it's not easy to see. I got as far as generating data (xml) that the native consumes, who's been blogging what, are they a friend etc. then I saw that webapps are in and you don't need the AppStore to distribute them.

just playing around with my iphone, looking at my install... it does a fine job at recognizing text input areas on a form, terrible with the tiny MCE. This does need to be addressed.

Ditto.  Only if we have a theme that will use a 1 panel which would a bit narrower than the panel where the login form is it will be sufficient.  If you have Wordpress, the WPtouch widget works nicely as it has an accordion type feature saving valuable real estate on a tiny screen.

i don't know, but i think in the longest run, considering that mobile devices are going become more the norm then bling material... it makes sense to just think about changing mce to be useful

More thinking about this, and I know some of this thought comes from my own current inability to do something that probably can be done.

but... if plugin index pages were held inside the views folder the same way the rest of the plugins are, it would be possible to easily add a different layout without much thought. Which would then allow these things to be quickly adjusted in a theme.