With enough skill you can replace broken parts in "non-modular" mobile phones as well. That's not the problem.

Try to upgrade CPU on your laptop, let's say from Sandy Bridge to Haswell. Then we can talk.

Exactly. Replacement can be even done in current mobile phones, that's not a big issue.

But it's still (almost) true. Of course such phones *can* be built, it's just a matter of money and time tossed on it. Problem is somewhere else - how do you plan to make such phones competitive at all. How do you want to provide high performance? Battery time? Low latencies? Low price? Size that will still fit in the pocket?

Mobile devices are evolving to be more integrated, not less. Modularity requires you to give up any integration - which makes things like high performance or long battery time hard to achieve. What's important - you cannot even simply sacrifice one of it, as odds are that it won't help much with others anyway - unless you sacrifice modularity of course.

Ara is just a R&D project. Maybe it will bring some useful knowledge that will be later integrated into real devices - but I don't think it will bring the oh-my-god-so-modular phone to the market. I'm more excited about projects like Neo900 - this is how "modularity" in mobile devices should be achieved. Plus the solution with two PCBs gives hope for even more "modularity" in future, with potential partial updates.

Why not? I'm doing that on my Openmoko Neo Freerunner for past 6 years. Same on Nokia N900, OpenPhoenux GTA04 and soon Neo900.

If you choose to buy crippled, locked-down phones, then you're the only one to blame.

Looks like in USA it was the same like in Europe. We here also had simlocks, making the phone working only with carrier you bought it at. But so what? You always could just buy the phone outside the carriers. Most of people didn't - but that doesn't mean that they couldn't. Of course that would mean paying something like $399 instead of $1 plus two-year contract.

Now there are laws that require carriers to remove simlock after contract expires and most of operators don't even install the simlocks anymore. But that doesn't matter for me. What matters is the mere ability, and knowing that some CDMA carriers in US aren't even using SIMs and just sell preactivated phones instead, I was afraid that some of this lunacy is also present with GSM carriers when I was reading comments like "maybe in Europe, but in US it's not possible".

But it seems like it's just people not aware of their own market, but behaving like they actually are - and that's the thing I don't care about. If you want to be conscious customer, you do your research before buying anything, so it's your fault if you don't know that you don't have to buy your new phone locked by the carrier :)

Android is free software as well, so what? Its application in real life generally is not, and judging from how well Firefox OS is able to overcome it (hint: it isn't), Ubuntu for phones will probably just follow the same path.

"Enjoy our awesome free system, but if device manufacturer locked down the bootloader and made access to root account troublesome, then it's not our fault, but manufacturers!"

Fortunately projects like OpenPhoenux exist, Neo900 will be available in few months. Both Firefox OS and Ubuntu for phones should work on it, as well as many other distros, including standard PC ones (making Debian usable as a phone is a matter of few apt-get installs and some basic configuration); without any proprietary drivers sans GPU acceleration (but looking at Replicant 4.2 running smoothly on GTA04 without 3D acceleration, I guess you can live without it if you really care about "software purity").

While Mozilla and Canonical might be doing some good job in software development, I simply don't care about their efforts in building hardware ecosystem, as they're even not trying to hide that they're doing it exactly the same like Android one, and that's unacceptable for me.

I don't know about Verizon, but people on Neo900's "compatible carriers" forum thread came into conclusion that there shouldn't be any problem with Neo900 compatibility on AT&T network, citing AT&T itself: http://www.att.com/shop/wirele...

So is AT&T lying on this page? You know, I'm one of the people working on Neo900, so I'd be grateful for some proof in case it's true, so we can properly warn our users :P

Based on comments above, looks like it is. I have yet to see any reason for why people in USA can't do that, other than their own laziness.

I'm pretty sure I've seen some people from USA in Openmoko community, using their US-banded version of Neo Freerunner. Given that the only way to buy Freerunner was described by me in parent comment, as no carrier even sold those phones, I conclude that it must be possible. What's more - nowadays phones are quad- or five-banded, so you don't even need separate US and Europe versions, like tri-banded Freerunner had.

Of course I can be wrong, but then I'd like to be proven wrong instead of just hearing "everything is not like where you are".

Then just buy a phone from Europe? Or straight from manufacturer? I can easily buy US-banded version of Neo Freerunner or GTA04. Probably won't be problematic to buy any US-compatible Android phone from eBay as well. AFAIK ZTE Open is even sold there straight from the manufacturer. So you're telling me that all those phones won't work in US? What's the problem then?

Uhmm, are they? I never bought any smartphone on contract with operator. The last phone I bought like that was a standard dumb phone many years ago.

Of course there are offers like that, but saying that "smartphones are tied to contracts" is bullshit. You just buy a phone, put SIM in and you're ready to go. If you're doing it in different way, that's your problem.

I think that if Nokia was in better state few years ago, Elop wouldn't be chosen as a CEO at all, so he wouldn't have a possibility to destroy it. But maybe that's just me being naive :)

>earlier management said to retire Symbian and migrate to the newer system

They did, but not before years of internal competition making both teams reinvent the wheel and limiting collaboration between them. Maemo 6/Meego as seen on Nokia N9 was entirely reimplemented three times. When it all started, Symbian team was also doing transition to Qt - just like Maemo; but internal conflicts, on both management and "engineers morale" levels led them to implement completely separate, incompatible libraries, wasting time of both teams.

Management eventually made a right decision, but it was already too late to stop Elop from destroying it all completely.

That's what I've been referring to. Putting some parts together to get modem working on Raspberry Pi maybe isn't hard, but it's definitely a great experience for a hobbyist doing something like that for the first time. In contrast, adding WiFi module and installing some SIP software on it is actually brain-dead simple (apt-get install linphone? :P) - it won't teach you anything more, and it won't work in any reasonable way as well thanks to Raspberry Pi limitations.

Doing stuff like that is great when they're fun, you can learn something from it or you can brag about it. Implementing suggestions above will provide none of these things, while it might provide a disappointment when you'll keep your hopes too high.

Because extending the scope of cool project from cool to pointless is just... pointless :)

By the way, this is not the first DIY mobile phone: here's a phone made out of Nokia N810 Internet Tablet: http://talk.maemo.org/showthre...