Not sure about this one, seems like the opposite to me. Twenty years ago, PC's were very modular, and it was very common to add/upgrade a sound card, memory, cpu, video card etc. Fast forward to now, and for the most common computing devices (think tablets and smartphones), there is almost zero modularity or upgradability. Even with current PC's (i.e. mostly laptops), it's increasingly limited.
The same trend is apparent in other similar technology. For example with analog TV, people could add a PAL block if their TV was NTSC or SECAM. Upgrading your car used to be significantly easier etc.
"Hugh Trevor-Roper is supposed to have said that 'history teaches us nothing except that something will happen', and this applies pretty well to tech: this is so young and so fast-changing an industry that drawing parallels is more misleading than helpful."
Feels about right to me.
Until now, smart phones/tablets have had little in the way of peripherals. Some headsets, keyboards, memory cards, etc., but people seem to be clamoring for certain peripherals above all others. They seem less concerned about CPU types, just like the in Mac and PC world. Key peripherals seem to be camera and some kind of "geekport"/internet of things.
I don't see a need to modularize every part of the phone a la PCs, but to expand the peripherals would help. It might be as simple as adding extra USB ports.
Integration is crucial for meeting consumer needs when the tech was not as good: the first iPhone had a tiny CPU and GPU, but they managed to squeeze adequate performance by connecting everything with extreme efficiency.
But once it's good enough, people start to worry about other factors - like I want a big battery, but I don't need 3G (or vice versa). Customization, not just replacing broken bits. Here, a modular approach shines. But it's less efficient, so there needs to be spare power. The key driver is that users want customization more than they want extra compute power.
Android is modular, compared to iOS, and it's worked out well - once phone got powerful enough. It couldn't have in the very earliest days, because the spare power wasn't there. Hardware modularity is the next step. It may be a little early, but I think it's clear that high-end phones have wildly overshot mainstream needs (a quad-core HD phone!? what a boondoggle). Actually, today mid- and low-end phones are perfectly fine for most people.
PS: this pressure will also drive the next category - whether it's iWatches or iGlasses or something else, I don't know, but it will be smaller, more convenient, cheaper, easier to use - and there won't be power to spare in such a small form-factor. Therefore, it's likely that Apple will do it (hence the i's) since they are the company most integrated across all the levels of hardware, silicon, OS, apps, appstore. They are best placed to squeeze the needed performance (doesn't guarantee they will though).
I am curious about what YC partners think of Christensen's modularity vs integrated (and his other theories), since they are VCs, in the business of predicting technology. But I've never heard them even discuss it...
Similar for Moore's Crossing the Chasm
So in the early days of a new market/category of product, the products are highly integrated, for several reasons. One is that the market is still new, so there isn't much of an "ecosystem" to begin with. Another is that the company with the "first mover's advantage" wants to keep stuff proprietary as much as possible, and another is that the product still kind of "sucks" in some areas (camera, battery life, in early days of the iPhone for example - compared to the traditional competition). So they need to make everything work as tightly as possible, to squeeze all the possible optimization out of it.
But eventually, the market becomes mature, the ecosystem grows, and the products become "good enough" for most people. So much of that extreme optimization or need to keep everything proprietary and in-house isn't needed anymore, and you actually start getting some advantages from the modularization of the market, such as buying a better modem than you can make from a "modem company", and so on.
For a while I wasn't sure this was going to happen to the smartphone market (ignoring the fact that there has been an increasing trend towards customization through colors and whanot), because for one the smartphone is a very tighly put together product, and it's hard to imagine how it could've been separated into a dozen different pieces without being junk, and two, for a while the trend was towards increasing "closeness" of devices, rather than openness.
But it seems it's going to happen, and ARA looks just about right (I wasn't a big fan of the Phonebloks pin-model). Still, even if the strategy here is "correct", and most likely on the right side of history, Google will still need to excel at execution, and make sure using such a phone gives very few disadvantages compared to a regular phone, but many more advantages (being able to use whatever camera you want, without buying a $700 phone every year, and so on). Otherwise, people could be turned off by the initial version, and then it will be a lot harder to convince them what a good idea this is. But for now I'm optimistic.
[1] - http://www.amazon.com/Innovators-Solution-Creating-Sustainin...