> you'll see components disconnect from one another and the thing crash, a lot

Isn't the whole point of project Ara that you can disconnect components and the phone doesn't crash?

> you'll lose performance as data is transitioned in and out of generic interfaces

I was thinking about this - I doubt they'll have a single interface that is generic enough to attach CPU and memory as well as other peripherals - I'm guessing each connector will bundle separate connectors for the CPU and RAM, so that whichever slot the CPU and RAM modules go into, they'll still be on "their" bus. Then you'll have the generic bus, which will be like a USB or PCI bus on a computer - you can attach anything else to it and it's all treated the same. The performance loss would come from the bus itself being inferior, not having to shoehorn disparate devices into using it.

> There is no empty space, a Project Ara phone needs room for connectors, housing, and similar. Where does this space come from?

No doubt, these will be larger phones.

> I genuinely hope they succeed and I'd buy a Project Ara phone if they can engineer something [awesome]

Me too :)

> Isn't the whole point of project Ara that you can disconnect components and the phone doesn't crash?

If the CPU is disconnected from the device it will not continue to function. Failing gracefully is the best you can do and even that is incredibly hard. Are you going to give the CPU its own battery so it can keep its cache when removed?

In the initial materials I read about Ara, they indicated there would be a very minimal, secondary CPU, memory and battery in the frame itself. So, when you want to hot-swap the CPU, it falls back on/suspends to that tiny system temporarily.

Why is it important to hot-swap the CPU? Who is this phone for? This is either a mobile high-availability server or engineer wanking.

Engineer wanking.

I believe this is like an eternal prototype, not meant for mainstream.

Make a fat, ugly device with low battery life and experiment with different concepts.

I assumption is they'll take learnings from Project Ara and apply it to a new, 'production' phone that has like a detachable camera or something.

In theory this is doable, I did it 6 years ago for equipment designed for Comcast. A few different programs running in embedded system - only one program was keeping the states. The rest of them are stateless - snmp, web, etc tasks. When 2nd CPU board was plug into the system standby mode, it synced all the states/config DB info from the master and become hot standby. After that you can stop, reboot, regrade the first CPU.

BUT, that was for one vertical application which the complete application stack is under my total control - all configurations are sync when in hot standby mode.

  Android system which anyone is able to install any type of Apps are completely different stories.  The design will be much more complex to handle all the cases - I am not sure if that's even possible. 

BTW, I watched the Ara demo for Io14 youtube - they can't even boot the system in the IO "demo". I worked on a few projects for Mobile SOC + android AOSP porting including JB, KK. Ara project has a long long long way to go.

It is a good science project - not sure about any commercial potential.

In the video where PhoneBloks visits NK Labs in Boston the phone boots.

http://blog.phonebloks.com/post/101254153023/video-project-a...

Interesting! Isn't this the kind of thing Erlang was made for? Telcos being able to swap out and upgrading switching boards on the fly without disruptions?

> Isn't the whole point of project Ara that you can disconnect components and the phone doesn't crash?

I'm not sure what you're asking here. I was talking about the design/construction challenges. If they are successful the product wouldn't be designed to crash, no.

I would certainly not mind carrying a bulkier phone in exchange for modularity and personalisation.

I may not be in the majority segment, but I'd wager there are others who feel the same.

As much as I hate it, I think you (we) are in the minority here.

Given how many geeks want better battery life over shaving a few mm off a device, and how that is exactly the opposite of what is happening, I think it's fair to say that the broader market want a slim phone.

I've not seen the dimensions of the prototypes yet, how bulky are they actually?

There is an interesting anti-slim trend on the outskirts of mainstream smartphone news; ruggedized phones like the Caterpillar S50 and Runbo X6 are three times thicker than fragile 'stylish' phones and would seem good candidates for Ara. Often it is internal components that fail.

In a traditional phone that should not break at all you can simply solder and glue everything together so that things will simply not break; then you put it in a casing that is water/dust/shockproof and be done with it. I guess ensuring the same thing for a modular frame might be harder. Drop it and get a slightly bent internal frame so that the screen module can't connect anymore, etc.

I think your fundamental assumption is based on "possible" performance. I would argue that required performance has flattened over the last few years. A nexus 4 is a perfectly serviceable phone still.

I think as technology improves, the advantages of modularity with trump any loss of performance.

More battery than generic performance. Performance may have improved but components haven't shrunk much and batteries definitely haven't shrunk.

You can likely reclaim some space by making a really tiny chipset, but batteries and screens combined are over 70% of your phone's total size. Project Ara makes components larger (due to connectors and housing), so you need to find space from somewhere, and there isn't a lot to be taken from most components (they aren't very large already).

Sure... bigger batteries will be bulkier, but they will at least be an option. My Nexus 5 battery sucks and there is nothing that can be done.

I think one of the more interesting ways they could regain that space is actually through spreading it out--because it's modular, it doesn't need to be one monolithic device. Obviously the system memory and processor need to be close together, but not all of the parts need to be in close proximity. For example, the battery could split into a small local battery paired with a larger battery or system of batteries sewn into a jacket to distribute the size and weight.

I wouldn't call it vaporware, Project Ara phones simply won't be top of the line, but that doesn't mean it can't happen; perhaps the compromises in performance and quality will be worth it for what modular phones enable.

If it ships at all it isn't vaporware. But my argument was that it would never ship, which is why I used the term.

Look into how far they are already, and the relationships they have already formed with manufacturers.

It's pretty much a sure bet that this is going to ship.

Now, whether there is enough interest to keep it alive is another thing altogether.

Phones are not plug and play, but this project lays the groundwork for them to be.

It's not as if anyone truly imagines every part needs to be modular, more that this allows parts to be modular which allows some degree of upgradability in the future or to have base devices more cheaply repurposed for specific usage scenarios.

>> It's not as if anyone truly imagines every part needs to be modular,

Are you saying that Project Ara will still allow the consolidation of separate functions into integrated SoCs, when it makes sense to do so?

Or are you saying that the consumer won't be obligated to piece their phone together? Nor swap out modules any more frequently then they currently upgrade phones?

Either way, you are right.

Maybe they can take what they've learned and scale it up to tablets (and maybe laptops)?

I would love a modular tablet. And I've love for them to eventually allow modules to swap with everything, like tvs and professional cameras and laptops.

Maybe in the future this(Project Ara) might be possible because I have got a hunch, Google won't give up especially the way things show up nowadays. Besides Google as billions of dollars ready to be spent on something like that.