Quick reply...
GPS errors are affected by geometric dilution of precision and depend on signal arrival time errors, numerical errors, atmospherics effects (like, clouds), ephemeris errors, multipath errors (reflecting from buildings) and other effects. But these circumstances affect both getting GPS lock and maintaining a good reading of GPS signals, right? So why is getting satellite lock taking so long?
It's not simply matter of listening for all thirty two satellites instead of four or five of them. It's not even the fact that the receiver looks for each of the currently unavailable, far away, satellites, sifting through 1023 cases. It's the matter of long navigation message, containing the almanac - so that the receiver could decide which satellites can be contacted, and which cannot - and the ephemeris - satellite's position. The navigation message itself is constructed from a 1,500 bit frame, which is transmitted at 50 bit/s. Each frame contains only 1/25th of the total almanac. At this rate, 12.5 minutes are required to receive the entire almanac from a single satellite.
So, don't blame the hardware - blame lack of information. Fortunately, the almanac is valid for long time periods (up to 180 days). So GPS lock should be faster next time, once you have downloaded it the first time.
I have not ever used network-assisted GPS - I avoid cellular-network-based-Internet at all costs. I consider GPS on N900 adequate, even if it doesn't miraculously give me my location in the first five seconds after the request. Though, I have not actively used it lately.
Best wishes.