IBM's ADEPT Project: Rebooting the Internet of Things


The Internet of Things, only a few years old, needs a reboot. So says Paul Brody of IBM. The IBM Adept project is a proof of concept for an open, decentralized Internet of Things.

IBM Think D100 Test

I recently spent some time learning about IBM's ADEPT project. ADEPT is a proof of concept for a completely decentralized Internet of Things. ADEPT is based on Telehash for peer-to-peer messaging, BitTorrent for decentralized file sharing, and the blockchain (via Ethereum) for smart contracts (this video from Primavera De Filippi on Ethereum is a good discussion of that concept).

The ideas and motivations behind the project as presented at IBM's Device Democracy align nicely many of the concerns I have raised about the Internet of Things. To get a feel for that, watch this video from Paul Brody, vice president and global electronics industry leader for IBM Global Business Services. Brody, speaking at the Smart Home session of the IFA+ Summit, says “I come not to praise the smart home, but to bury it.” Its worth watching the whole thing:

Note: the video doesn't show Brody's slides. I couldn't find these exact slides, but this presentation to Facebook looks like it's close if you want to see some of the visuals.

The project has a couple of white papers:

  • Device democracy- Saving the future of the Internet of Things (PDF) is a business-level discussion of why the Internet of Things is already broken and needs a reboot.
  • ADEPT: An IoT Practitioner Perspective (PDF) is a more technical look at the protocols they chose and how they come together to create a completely decentralized Internet of Things. The paper describes their proof of concept based on Telehash, Ethereum, and BitTorrent. It’s worth reading to understand the way they’re thinking about trust, privacy, and device-to-device (D2D) and device-to-vendor (D2V) interactions.

Brody says current IoT is broken, and won't scale because of

  • Broken business models
  • High cost
  • Lack of privacy
  • Not future-proof
  • Lack of functional value

One of the key ideas they discuss is autonomous coordination. This is critical in a world where any given person might have thousands of connected devices they interact with. we simply won't be able to coordinate it all ourselves (part of the reason the current IoT needs a reboot). For example, they use an example I've used of electrical devices coordinating their use of the home's power to avoid a surcharge from the electric company. That's a hard problem that doesn't easily admit centralized solutions.

The ADEPT concept imagines each device being connected directly to the Internet and consequently they spend some time dealing questions like "what if my device is too slow or doesn't have enough memory to use the blockchain?" One of the reasons I'm a fan of creating virtual proxies of physical devices via persistent compute objects (picos) is that they can provide processing and storage that a simple device might not be able to provide because it's too slow, too small, intermitently online and so on.

The more important reason for using virtual proxies on the Internet of Things is to provide representation for things that aren't physical things. People, places, organizations, concepts, and so on all need to interact with things. Picos provide an architecture for accomplishing that. Picos provide a foundation for the primary activities we need in a decentralized IoT:

  1. Distributed transaction processing and applications
  2. Peer-to-peer messaging and sharing
  3. Autonomous coordination and contracts between peers

And they do this for everything whether it has a processor or not.

The conclusion of the Digital Democracy white paper says of winners and losers in the IoT economy:

Winners will:

  • Enable decentralized peer-to-peer systems that allow for very low cost, privacy and long term sustainability in exchange for less direct control of data
  • Prepare for highly efficient, real-time digital marketplaces built on physical assets and services with new measures of credit and risk
  • Design for meaningful user experiences, rather than try to build large ecosystems or complex network solutions.

Losers will:

  • Continue to invest in and support high-cost infrastructure, and be unmindful of security and privacy that can lead to decades of balance sheet overhead
  • Fight for control of ecosystems and data, even when they have no measure of what its value will be
  • Attempt to build ecosystems but lose sight of the value created, probably slowing adoption and limiting the usage of their solutions.

One of the things I really like about the IBM vision is that they do a good job of tying all of this to business value. Speaking of the effect the Internet has had on the market for digital content they say "The IoT will enable a similar set of transformations, making the physical world as liquid, personalized and efficient as the digital one." They use the idea of "liquifying the physical world" to bring this home and discuss why this enables things like the following:

  • Finding, using, and paying for physical assets the same way we do digital content today
  • Matching supply and demand for physical good in real-time
  • Digitally manage risk and assess credit
  • Allow unsupervised use of systems and devices, reducing transaction and marketing costs
  • Digitally integrate value chains in real-time to instantly crowdsource and collaborate

This is a bold vision that aligns well with Doc Searls' thoughts expressed in The Intention Economy: When Customers Take Charge. This kind of business value is what will drive the IoT, not things like "turn on the lights when I get home." I think that's what Paul Brody meant when he said "I come not to praise the smart home, but to bury it." The smart home isn't where the business value will be and a centralized, proprietary, and closed vision for creating it is bound to fail.

I'm working on a white paper that lays out a similar reference architecture for the Internet of Things, so I find this project fascinating. More to come...