No baby boomers had been born when Congress enacted Title II of the Communications Act in 1934 as a means of regulating the Bell telephone monopoly, and the first Millennials were in elementary school when that monopoly was broken up in 1983. Title II was set to die along with plain old telephone service until the Obama administration decided Title II should be used to implement net neutrality—the principle that consumers should have reasonable access to internet functionality. Title II is wholly unsuited to this task, because it doesn’t apply to Silicon Valley companies that control access to many of the internet’s core functionalities.

Title II doesn’t apply because Congress never intended to apply Title II to internet transmissions. Congress intended to limit Title II regulation to transmissions that are interconnected with the public switched telephone network—i.e., to the plain old telephone service that was once provided by the Bell monopoly.

Congress limited the application of Title II to transmissions that meet the statutory definition of “telecommunications,” which is defined as “the [1] transmission, [2] between or among points specified by the user, [3] of information of the user’s choosing, [4] without change in the form or content of the information as sent and received.” The conjunctive elements of this definition clearly describe transmissions on the public switched telephone network, but clearly do not describe internet transmissions.

The ability (or inability) of a user to specify the “points” of a transmission is an unambiguous factual distinction between plain old telephone calls on the publish switched telephone network and internet transmissions. A user making a plain old telephone call specifies the call’s points merely by dialing a telephone number. Conversely, a user cannot specify any definable points for internet transmissions. A comparison of network topologies and addressing systems for the public switched telephone network and the internet illustrates the distinctions.

Topology of the public switched telephone network

The public switched telephone network is based on centralized switching in which each residence and business (or “customer premises”) is connected by a dedicated line (or “loop”) to a switchboard (or “switch”) located in a facility (“central office”) near the center of a local network. When a user makes a phone call, the switch connects the callers’ loop to the called party’s loop to establish a dedicated circuit for the duration of the call.

For example, if Ajit called Mignon on the local telephone network depicted above, the switch would connect Ajit’s loop to Mignon’s loop to set up the call. During the call, the circuit formed by their loops would be dedicated to their conversation only. The switch would disconnect their loops only when Ajit or Mignon hangs up.

Centralized switching works well in relatively small geographic areas, but in areas that require long loop lines, centralized switching is uneconomical. The Bell System’s solution was to keep the areas served by central offices relatively small (about three miles in radius) and to interconnect central offices with “trunk” lines. Whereas subscriber loops are typically dedicated access circuits that connect customer premises to the central office, trunks are shared because only a small percentage of loops are typically in use simultaneously. Though trunks are shared, each call still receives a dedicated circuit. When the central office switches a loop to an unused trunk line, it reserves (or “seizes”) the trunk for the duration of the call. When the call is disconnected (because a party hangs up), the trunk is “released” for use in another call.

Communications between central offices are connected by switching the loop of the calling party at the “initiating central office” to a trunk line connecting to the “terminating office,” which then switches the trunk to the loop of the party being called.

For example, when Ajit calls Mignon using the network depicted above, Central office #944 is the initiating central office, which seizes the trunk line highlighted in red and switches Ajit’s loop to it. Central office #331 is the terminating central office that switches Mignon’s loop to the trunk highlighted in red to complete the circuit. This circuit will remain dedicated to the call between Ajit and Mignon for its duration.

In larger cities, multiple central offices are connected by “tandem offices” (or “tandem switches”) that only switch (tandem) trunk lines and cannot originate or terminate a call. The use of tandem offices avoids the need to directly interconnect all central offices in a local area, as depicted below.

The centralized local switching topology described above is typically used within a city and its immediate area. The network within such an area is known as an “exchange.” Calls made within an exchange area are usually included within the basic telephone subscription price (i.e., there is no additional charge to make a local call within the same exchange area).

Calls between exchanges (known as “toll,” “long distance,” or “interexchange” calls) use the “long distance” network, which is an extension of the “local tandem” network topology used in larger local exchange areas. Exchanges are connected to other cities using long distance (or “toll”) trunk lines. An exchange sends a long distance call through the local “toll office” (or “interexchange office”) which handles billing for long distance calls (because long distance calls were traditionally not covered by the basic subscription price). The toll office (which might be merely a specially equipped portion of the switchboard in an exchange with a single central office), then switches the call to the appropriate intercity line, either directly or through an “access tandem.” A simple long distance network is depicted below.

A user specifies the points of a plain old telephone call merely by dialing the number because telephone numbers function as the addresses of customer premises (i.e., specific loops) and individual subscriber telephones (i.e., a different telephone number is assigned to each mobile device), or other telephony endpoints.

The format for a 10 digit telephone number is divided into 3 parts represented as NXX-NXX-XXXX. The first three digits are typically assigned to a specific exchange area (like the one depicted in Figure 3) and are commonly known as “area codes.” The three digits following the area code are assigned to a specific central office within an exchange (office code), and the final four digits (line or station code) are assigned to a specific local telephone line or mobile device (such as a cellular phone or tablet). A telephone number thus specifies (1) a specific exchange area (via the area code), (2) a specific central office (via the office code), and (3) a specific local loop or mobile device (via the line or station code) that is dedicated to a specific customer premises (e.g., a particular residential address) or a specific device (e.g., a particular iPhone).

In the example depicted above, Ajit specifies Mignon’s vacation home as the end point of the call when he places a long distance call to telephone number (803) 555-2010. In this example, Ajit would first dial a 1 to signal that he intends to make a long distance call. Central office #944 would then switch the call to the toll office in Ajit’s local exchange. That toll office would switch the call to the toll office in area code 803, and the toll office in area code 803 would switch the call to central office #555. Central office #555 would then switch the call to line #2010, the local loop that terminates at Mignon’s vacation home.

Congress designed the Communications Act around the fact that a users specifies the known points (locations or devices) of a call merely by making the call. Since its adoption in 1934, the Communications Act’s most fundamental jurisdictional distinction has been based on the plain old public switched telephone network’s inherent ability to specify the end points of a telephone call. Specifically, Congress granted the FCC authority to regulate “interstate and foreign” communications while expressly denying FCC authority with respect to “intrastate communication service.” According to the Supreme Court, this “system of dual state and federal regulation over telephone service” requires federal and state regulators to exercise their jurisdiction based on the geographic location of the end “points” of “telephone communications.” The Supreme Court recognized that Congress based this jurisdictional separate on the fact that “[t]he end-to-end geographic locations of traditional landline-to-landline telephone communications are readily known.”

Topology of the internet

In a series of decisions determining that voice-over-internet-protocol services should not be regulated under Title II, the FCC relied on the fact that the points of internet transmissions are “difficult or impossible to pinpoint.” The internet’s defining characteristic is its ability to transmit information without specifying a particular path (i.e., dedicated circuit) or transmission points.

The internet protocol suite has separate systems for identifying and addressing devices on (1) local networks and (2) devices on different networks. Each internet device on a local network has a unique number known as a “hardware address” (or “MAC address”), similar to the way in which each mobile phone is assigned a unique telephone number. Unlike telephone numbers, however, MAC addresses are used only for transmissions between devices that are directly connected on a local network.

Separate “IP addresses” are used to connect to devices on different (non-local) networks. IP addresses are independent of particular hardware (i.e., logical) and are used to create a “virtual network” for indirect transmissions between or among local networks (i.e., “internetworking”). The use of “logical” IP addresses solves the basic problem of connecting different networks: the problem being that transmissions between devices use MAC addresses, but each device on a local network only knows (or can directly discover using “address resolution protocol”) the MAC addresses of the devices that are directly connected to that network.

Unlike a 10-digit telephone number, which identifies a specific loop or mobile device, an IP address does not identify a specific device; an IP address identifies only the interface (“host” or “network interface”) between a specific device and the internet (other networks). The main components of an IP address are a “Network Identifier” (or “Network ID”), which identifies the network where the host is located, and a “Host Identifier” (or “Host ID”), which identifies the host on the network. Each internet router maintains a “routing table” that maps different Network IDs and the other routers to which the router is connected. Each entry in the table contains information about one network (or subnetwork or host) indicating the routes that lead to that destination. Each time a router receives a packet, it compares the destination IP address to the entries in its routing table to decide where to send the packet next. The process of routing is what allows a device to send transmissions to any other device on the internet without specifying a particular point or even knowing where a particular point is.

The use of “logical” IP addresses solves the basic problem of connecting different networks: the problem being that transmissions between devices use MAC addresses, but each device on a local network only knows (or can directly discover using “address resolution protocol”) the MAC addresses of the devices that are directly connected to that network.

For example, assume Ajit is using an xDSL broadband connection and wants to access information that is associated with the URL http://www.fcc.gov. As depicted below, Ajit cannot directly connect to the server that has this information even if Ajit knows the server’s MAC address, because the server is on a different network, and neither Ajit nor his computer knows where the server’s network is located. Ajit must instead send his message using the server’s IP address, which enables routers to forward the message from one physical network to the next, one step (or “hop”) at a time. At each hop, a router determines where to forward the message next until it reaches the host, which knows (or can discover) the server’s MAC address and forward the message to its final destination. Note that, if a particular route becomes congested, Ajit’s transmissions may take multiple paths, and there is no way for Ajit to specify or even know the points of his transmissions.

Why this distinction matters

The distinction is critical to the future of internet regulation. When a statutory definition is unambiguous, Congress’s intent is clear and the statutory definition controls. The unambiguous fact that internet transmissions don’t fall within the statutory definition of “telecommunications” thus means the FCC cannot regulate internet service providers under Title II.

Though a federal appellate court upheld the FCC’s 2015 order applying Title II to internet service providers, the court didn’t address the meaning of the term “telecommunications” because the parties in the lawsuit didn’t raise the issue. This means the FCC is free to decide that it lacks legal authority to apply Title II to ISPs in its current proceeding to reexamine its net neutrality rules.

If the FCC does that, and a reviewing court affirms the FCC’s decision that the term “telecommunications” is unambiguous, a future FCC won’t be able to change its mind. No matter how important the policy issues raised by net neutrality might be, the FCC has no power to change the unambiguous meaning of the statutory definition of “telecommunications.”

An FCC decision holding that the term “telecommunications” doesn’t include internet transmissions would thus end the current game or regulatory ping-pong with respect to net neutrality and put the issue squarely where it belongs: in Congress.