When looking at the specifications of smartphones that have been released over the past years, it’s remarkable to see how aspects like CPU cores, clockspeeds and GPU performance have improved during this time, with even new budget smartphones offering a lot of computing power, as well as a smattering of sensors. Perhaps even more remarkable is that of the approximately 1.5 billion smartphones sold each year, many will be discarded again after a mere two years of use. This seems rather wasteful, and a recent paper by Jennifer Switzer and colleagues proposes that a so-called Computational Carbon Intensity (CCI) metric should be used to determine when it makes more sense to recycle a device than to keep using it.

What complicates the decision of when it makes more sense to reuse than recycle is that there are many ways to define when a device is no longer ‘fit for purpose’. It could be argued that the average smartphone is still more than good enough after two years to be continued as a smartphone for another few years at least, or at least until the manufacturer stops supplying updates. Beyond the use as a smartphone, they’re still devices with a screen, WiFi connection and a capable processor, which should make it suitable for a myriad of roles.

Unfortunately, as we have seen with the disaster that was Samsung’s ‘upcycling’ concept a few years ago, or Google’s defunct Project Ara, as promising as the whole idea of ‘reuse, upcycle, recycle’ sounds, establishing an industry standard here is frustratingly complicated. Worse, over the years smartphones have become ever more sealed-up, glued-together devices that complicate the ‘reuse’ narrative.

Recycling Imperfect

One question that may come to mind when the idea of ‘recycling electronics’ is raised, is why this is such a terrible idea. After all, when you send a device in for recycling, it gets carefully stripped down and all materials from it are sorted before the metals molten down, plastics recycled and all the other bits and bobs handled in that industrial fashion that makes ‘How It’s Made’ episodes and kin such a delight to watch.

The reality is, unfortunately, less sunny and perfect. According to the UN, only 20% of an estimated 50 million tons of annual electronic waste (e-waste) is formally recycled, which is to say that those are recycled in properly equipped recycling centers. The remaining 80% of e-waste is dumped in landfills, or is ‘informally recycled’, generally by local people who burn the circuit boards and wiring to extract the metals, often without any kind of protective gear. These findings strongly highlight the need to reduce the amount of e-waste so long as we do not have the capacity to even recycle it.

Yet even within formal recycling facilities, only part of an old smartphone is truly recycled. For example, a massive problem is and remains plastics, many of which are highly resistant to recycling, especially when the economics of recycling plastics is taken into account. Worse, the economics of phone recycling are worsening over time, as fewer precious metals and other valuable elements are used in circuit boards and chips, as well as in smaller quantities. As a result, after shredding of the printed circuit boards and their components, recovery of these metals takes more effort for less material. Even with copper prices going up constantly, the economics of recycling are such that the concept of not recycling a working device, but rather reusing it can make sense from multiple perspectives.

Carbon And Economics

The aforementioned CCI metric proposed by Jennifer Switzer et al. is defined as: ‘the measurement of the lifetime carbon impact of a device versus the lifetime useful compute it performs’. In more basic terms, it tries to capture whether it makes more sense to use a computer (like a smartphone) for computing tasks rather than to send it off for recycling and buy a new device to replace it. Interestingly, it is also noted by their paper that between 60-70% of old smartphones are never thrown away, but rather kept lying around.

Would it make sense to use even a fraction of these devices as something like compute nodes instead? An interesting notion is that such smartphones are more than capable of running microservices and when powered using electricity from low-carbon sources (hydro, nuclear, solar, etc.), are essentially carbon neutral in their operation. The power efficiency of these smartphone SoCs are a great benefit here, as they are already optimized for performance per Watt, and they come with their own UPS in the form of the built-in battery.

In its simplest form, such a ‘compute farm’ using smartphones can be set up using nothing but a simple webpage, as demonstrated succinctly by the Edinburgh Parallel Computing Centre of the University of Edinburgh back in 2016. In this demonstration, volunteers would load the webpage that contained some JavaScript so that their device can then contribute to the impromptu parallel compute cluster. For a more custom solution, devices could be flashed with a custom ROM that optimizes it for a specific task.

Cycling Upwards

One aspect that really cemented the IBM PC as a computing concept that has endured to this day is the ability to upgrade, add and replace entire components through the use of storage, memory and processing unit modules. Attempts to accomplish something similar with smartphones have been attempted for more than a decade with so-called modular smartphones. Unfortunately, after 2015’s PuzzlePhone (died: 2017) and Google’s Project Ara (killed in 2016), there have been no significant attempts to make smartphones in general into a modular, easily repairable system. This – along with the traditional locked bootloader – significantly limits any reuse attempts.

In this regard, reusing smartphones in a compute cluster is probably the most straightforward option, which could e.g. for an Android smartphone involve using the native development kit (NDK) to run the same C-based code as would run on regular compute nodes. Less straightforward would be reusing especially an older smartphone as a dedicated media player, as eventually the device’s OS would be considered ‘too old’ for such media player applications. Here the lack of updated (binary blob) drivers for older mobile SoCs is a major reusability obstacle, as this locks these systems essentially into an older Linux kernel.

When we look at what Samsung was suggesting with its upcycling program before it got nerfed, concepts for reuse included everything from a smart home controller to a weather station and nanny camera. More importantly, it would unlock the bootloaders and remove the need to purchase a lot of new devices whose functionality could be easily covered by an older smartphone. Anything involving a display, WiFi, Bluetooth and a battery, essentially. Considering that for example a smart home controller is just an SoC-based device with WiFi, a display, etc. using an old smartphone here instead would seem sensible.

In light of this, the skeptic’s view could thus be that the problem lies with the phone manufacturers, who will just not let us have nice things.

Throw-away Society

That it is more efficient to keep using devices such as smartphones that otherwise end up collecting dust in drawers – or shredded to recover a fraction of the materials that went into their production – is something that should be clear at this point. The lack of such reuse being implemented is something that can generally be attributed to the general ‘throw-away society‘ attitude that has become more and more prevalent since the rise of industrial-scale production of goods in the 20th century.

Considering the related concept of planned obsolescence, which was coined as early as the 1930s, it seems now almost quaint to look at the IBM PC and the extreme extensibility and upgradability that it enabled. Not only did it offer a flexible upgrade bus that enabled whole new industries of expansion cards and more to spring up, the PC clone wars of the 1990s also essentially annihilated the fixed-design, limited upgrades of the home computers up till then, even if this was not what IBM had intended to happen. It’s possible that IBM’s experience with easy upgrades, maintenance and repairs with mainframes played a role in this design choice, but the effect was that the PC became the de facto standard, with all of these modularity benefits.

Due to the modular nature of PCs, a system can be configured and reconfigured to fit a particular role, all of which helps to extend its useful life. Even if as a modern-day version of the Ship of Theseus every single component in a system will have been replaced over the span of a few decades, it seems fair to argue that although it’s not the exact same device that it was at the beginning, from an e-waste perspective, each individual component will have served its maximum useful life.

Additionally, as a modular system, components from different PCs can be assembled into another system, which might go on to be useful for another few years. This is also unfortunately a property that laptops have lost over the years, and which smartphones and tablets have never embraced in any meaningful way. Maybe that with the Right to Repair movement making inroads at long last, some level of modularity will also make it into smartphones and other devices, which would make not just repair but also reuse significantly easier and attractive.

Who knows, maybe one day smartphones will feature the same DIY, white box and OEM systems as we see with PCs today, and people will use old smartphones for clusters and hobby projects that today require a Raspberry Pi board or kin.