doron :
Choice is fine, but the iPhone has proven that people will rather have a simple device that can do many things okay-ish.
In terms of smartphones, the point of modularity isn't really about pick-and-choose features. It's about upgrading at a lower cost. When you upgrade phones, most of the time the only thing that really needed upgrading was the CPU and RAM. Screen resolutions have pretty much maxed out, camera capability is "good enough" that anyone wanting better buys a DSLR, there are plenty of bluetooth speakers on the market, and GPU isn't exactly a make or break item.
Most people upgrade phones because the CPU has gotten too slow or the RAM is insufficient for modern software. Make it modular and you can upgrade just those components for under $100, instead of $500-$700 for a new phone. (Ok, another major reason to upgrade is because you dropped the phone and broke the screen. But if that's modular too, you can simply replace it.)
The main impediment to modular phones (at least in the U.S.) is the current phone purchase model, where the carrier subsidizes it and hides its cost in your monthly bill. This has led to a massive market distortion where you pay $200 up-front for the high-end phone, pay nothing up-front for the low-end phone, both have the same monthly service charge. So the high-end phone ends up only costing $200 more than the low-end phone, instead of the $500 or so true difference in price. Effectively, the high-end phone is subsidized by the low-end phone. That leaves very little cost savings for a modular approach. This is gradually changing with T-Mobile leading the push to make the phone a separate purchase, so who knows where it'll lead.
Modularity adds bulk, but as components keep shrinking that additional bulk becomes less of an issue. In the general market, modularity wins out all the time over one size fits all. You can put different components into your PC. You can mount different brand tires onto your car. You can plug different headphones into your MP3 player. etc.