We Propose a New Stack for the Net


Back in the ancient tech times, when phone signals traveled over wires and TV signals traveled through the air, the limitations of then-existing hardware and software compelled a layered, flexible approach toward getting any information from Point A to Point B.

The most elegant and powerful solution to this situation was the TCP/IP stack, which underpins the Internet and continues to connect the myriad of devices around the world today.

TCP/IP, which stands for Transmission Control Protocol/Internet Protocol, cooked up a four-level layer cake for tying systems together:

1. Application layer: What the user actually interacts with.

2. Transport layer: Grabs information packets from the app layer and sends it to another server somewhere on the network.

3. Network layer: Controls the connections among all the servers and other known devices on the network; this is what “routes” information so it gets where it’s going.

4. Data link layer: The translation from the network to the requirements of any specific device, e.g., a PC, a smartphone, etc.

TCP/IP has been one of the sturdiest and most valuable concepts in technology. First released in 1974, a time when the dinosaurs of mainframe and mini-computers dominated the tech landscape, this stack has successfully integrated local area networking, wireless communication, cell phone networks, Bluetooth and a host of other technologies not even a glint in a techie’s eye back at its origin.

For four decades, TCP/IP has been the root structure underlying the systems the world and its people depend upon. But, in tech as in life, nothing lasts forever. And I think the era of TCP/IP may be nearing its apogee.

TCP/IP was conceived around fixed computers and a relative scarcity of tech resources on networks. Today, though, the world of tech is no longer computer-centric. Smartphones dominate and operate alongside IoT devices, Bluetooth earbuds, Alexas and Google Homes. Networks, too, aren’t what they once were. Cell phone systems now rival the traditional Internet as the carrier of valuable human and business information. Nor are keyboards, and the letters and symbols they contain, any longer the only form of inputting information into the worldwide network. Video dominates now, and voice is coming on strong as the natural interface for the on-the-go, device-in-your-pocket or on-your-body future.

So, here is a modest proposal for a new four-level layer cake for the emerging network future:

Voice/motion layer. The dominance of applications will wane as voice dominates as the interface for most network activities. The top layer becomes one where spoken words, possibly augmented by video input to determine place and other factors, are gathered up and parsed into tech-useful chunks.

AI layer. The processed audio/video signals get passed to AIs that determine not only their literal meaning, but a range of meta-meanings: who is talking, what is happening in that space at that moment, what is the intent of the request. This AI determines what is being requested, what the source for response should be, how that source should be accessed, etc.

Blockchain layer. Today’s routing—which was crafted largely to have a network that could survive a nuclear attack, not for any aspect of consumer or business convenience—will be replaced by a Blockchain-derived layer that can control interactions, at very small grain, based on high trust. Content bits, payment bits, security bits, essentially anything of value, can be managed by such a construct. Because it has no center, Blockchain is inherently more efficient than today’s server-centric networks.

Network layer. The processing power and vast storage capacity of today’s devices render the issues that loomed large back when TCP/IP was created as, fundamentally, negligible. Depending upon a whole range of factors, any experience, currency, document or data can be routed anywhere it needs to be, or even multiple places simultaneously. Think of the already existing systems that can route texts to the smartphone app where they are most likely to be seen: What’s App today, SMS tomorrow. Or that can route a single message to a set of people who equal “my closest friends” based on its origin and content. Core place-to-place routing, the Internet’s original problem, is now trivial. Routing will become deeply situational and a natural derivative of all the AI-driven and trust-mediated choices made in the upper layers.

Even thinking about these actions in the emerging network as discrete layers may be a bit of an anachronism. They won’t really function in the formal way TCP/IP presumes. Instead, the emerging network will be more like a gyre, with messages and their associated attributes, swirling among these various essential actions, gathering context from the motion.

- Mike Edelhart