One of the most ambitious ideas in blockchain is that the technology could enable not just people and businesses to transact with each other seamlessly, but also machines.

If this scenario, first articulated a few years ago, comes to pass, devices ranging from refrigerators to automobiles would not only collect and share data as part of the emerging internet of things (IoT), they would also be endowed with cryptocurrency wallets and unique, blockchain-based identities. 

The business possibilities from this physical internet of value are vast, potentially expanding the transactional economy in once-unimaginable ways. Self-driving cars might pay each other to cut ahead, for instance; a fridge equipped with sensors would know when it's running low on milk and zap the grocer some crypto to deliver a fresh carton. 

But cut through the awe-inspiring visions and hype, and you'll find debates now starting to take shape over nitty-gritty details.

Recent conversations with specialists at the intersection of blockchain and IoT reveal disagreements about the best ways to build various parts of the so-called machine economy, suggesting that at a minimum, the next phase will be full of trial-and-error.

As Carsten Stocker, CEO of Spherity, a startup specializing in blockchain and IoT integration, told CoinDesk:

Stepping back, one problem with IoT is that it's a hopelessly broad category.

On one end of the spectrum are high-value devices such as cars, with their abundance of computing power and battery life. On the other end, we enter the low-power world of hundreds of millions of simple devices.

At each extreme, there is a unique set of difficult design questions that are now being confronted by those seeking to tie internet-connected devices to distributed ledgers.

But while the business and utility potential around cars in the IoT-blockchain world is tantalizing, the auto industry faces its own set of interesting business and design challenges.

For example, there is an emerging debate about where the "birth certificate" of a car, or its identity, should be located. Should it be the gearbox, where a lot of BMW's and Mercedes' intellectual property is concentrated? Or the engine? How about the hard drive that collates sensor data?

Alexander Koppel, CEO of Riddle & Code, a startup specializing in giving physical objects unique blockchain identities, said discussions have also been around the battery as a transactional component within a car. It charges and stores energy, and can start to hedge power and even sell it.

"The battery companies think they are the winners," Koppel said, adding rather poetically:

Meanwhile, Tobias Brenner, senior consultant at the Deloitte Blockchain Institute, thinks multiple wallets in cars is the intuitive solution, with the expectation of communication between the engine and the battery, and possibly also solar panels on the roof. Then there's the in-car entertainment system which could have a wallet, and so on.  

"Multiple wallets within those systems may be a little bit more complicated but in the end would be very transparent, like which engine used what kind of power, for example," Brenner said. "So, you have to balance it between the transparency and complexity of the whole system."

Riddle & Code has been working with electric car battery makers to help secure this key piece of hardware (apparently such a battery has the explosive potential of a hand grenade so has to be secured against anybody hacking into it).

One thing is certain, Koppel said: "The innovation and business potential from making an object a wallet is phenomenal."

At the far-off edges of the IoT, where "things" tend to be very low-power and computationally simple, the argument is over how much complexity is helpful when integrating the network with a blockchain.

Some IoT experts, taking a practical view, think the only requirements at the end-points should be to deliver secure identity and no other complexity.  

Amir Haleem, CEO of Helium, which is building a decentralized network of wide-range wireless protocol gateways and a token to connect edge IoT devices, said adding complexity to end devices "is like a gigantic hurdle to people actually building things."

Apart from anything else, there's the cost.

"People get very sensitive about the bill of materials (BoM) when you start talking at a scale of millions or tens of millions," said Haleem. "You start proposing like a 60 cent addition to a BoM and all of a sudden that's a meaningful number."

Haleem said it makes no sense for end devices, like sensors that track and monitor medicine or food supply chains, to actively participate in a blockchain because these have to be power-efficient and cheap in an IoT setting. But delivering strong identity in the form of hardware-secured keys is essential, particularly in the face of recurring widespread vulnerabilities, botnets etc.  

"We move all the complexity up to the gateway, which is the device that creates the network coverage and acts as a node on the blockchain," he said. "We try and keep the devices dumb and stupid effectively – and cheap."

But not everyone sees it this way. Hewlett Packard Enterprise (HPE), for instance, recognizes that the majority of data is generated at the edge of the IoT world and says the goal should be to enrich those end devices in terms of computing power.

Christian Reichenbach, a transformation consultant at HPE, argues that a proliferation of IoT devices capture so much data that some computation and analysis has to take place there to sort through it all.

At last year's HPE Discover event, Reichenbach was part of a team demonstrating how an iRobot vacuum cleaner with a Raspberry Pi could use crypto tokens it earned from cleaning rooms to pay a smart plug for energy when it needed to recharge – what HPE calls "the servitization of products."

"We need computing at the edge to get intelligence out of all this data and then that can be transferred back to data centers or a blockchain," said Reichenbach, concluding:

Internet of things image via Shutterstock