The more complicated a plan is, the more likely it is to fail.
Project Ara is dead.
Just four months after Google announced they were going to work on a developer model for the end of this year and a consumer model sometime in 2017, the program was laid to rest. And really, is anyone surprised? The deck was stacked against them since the idea was born.
The idea itself is great. Forget the folks who say they will carry around different cameras or different speakers or different anything because those will end up just being one more thing left on a dresser after a few weeks. But Google's example of an insulin monitor that can be plugged in when you need to test your blood is awesome, and I'm sure smart people had ideas of other ways it could actually be useful. That's how some great ideas are born — think big and sort the details out later. But back to reality for a bit.
Great ideas and practicality seldom mix.
Phones and other small electronic things that use embedded hardware are pretty specific when it comes to the software that powers it. Writing the software and optimizing the code for a very specific set of components is mandatory. You have limited storage and limited power so you're forced to use both very efficiently. Writing the software so that it can support more than one part for a single function would make that very difficult. It could be done, and Google may very well have a solid plan to make it work. Google isn't the only company involved, though.
Android might be open source software, but the open portions can't be used to write a working operating system. The people who make things like the processor or the camera or the memory controller will have some very model-specific software need to make them work. Without these files, your phone just isn't going to work. A modular phone would not only need software support from Google but from the companies who make each individual part, too. And then those companies will need to be convinced to allow Google to redistribute that software as a downloadable package. And we already have seen how a single company can bring things to a screeching halt when it doesn't like that idea, as Qualcomm held back needed software for the Nexus 7 2013 Lollipop update. Things can get a little ugly. Good people can quit their job in frustration.
Software support comes from many different companies.
Now multiply that situation by the number of companies who build the parts that fit into one of the six slots on the Ara chassis. Suddenly, unoptimized software that takes a lot of storage space isn't such a big issue.
We aren't sure why the project was scrapped, and I'm not saying this was the reason. But this would be a pretty big stumbling block and I don't think even Google could work around it. Other folks in the "industry" that I've talked to feel similar, and we even heard some doubt about the project's future at Google I/O 2016 as the news of Project Ara's future was being announced.
Other companies are still going to work on a modular phone. One day someone will succeed and make one we can buy.