Curatorial essay · 02 of 12
An Audible Contract
What a modem handshake made public inside the home
The dial-up handshake was measurement, negotiation and household theatre compressed into a few unforgettable seconds.
Opening scene
A telephone number has been dialled, but no person is expected to answer. The far end responds with a stable tone. The local machine replies. A sequence of pulses, sweeps and granular noise follows, loud enough to fill a bedroom and recognisable from the hallway. Then the loudspeaker falls silent. A small word appears on screen: connected.
To call this sound random is to miss its function. The two modems have inherited a voice channel designed for people and must discover what that particular channel will permit. They advertise capabilities, measure impairment, choose methods, and reduce ambition when the line cannot sustain the preferred rate. What the household hears is an agreement under pressure. It is also a rare moment when network formation is acoustically public.
Calling a network
Before networking could begin, the computer placed an ordinary call.
A dial-up modem entered the network through the public telephone system. From the line's point of view, the connection began as a voice-band call. The number had to be dialled, the remote equipment had to answer, and a usable circuit had to exist between them. Only then could the devices establish a data link. The mundane events of telephony—busy lines, extensions lifted elsewhere, noisy wiring and charging structures—became conditions of computing.
This inheritance shaped domestic behaviour. The household often possessed one line, so connecting the computer temporarily withdrew the telephone from everyone else. A person preparing to go online might announce it, check whether anyone expected a call, and negotiate how long the line could remain occupied. The connection was technically point-to-point and socially many-to-one. A private session on the screen produced public consequences throughout the home.
Training the channel
The tones were tools for discovering a shared operating point.
ITU Recommendation V.8 specifies procedures through which data equipment can begin a session and identify available modes. V.34 describes adaptive modem operation at rates up to 33,600 bit/s. The training process behind the familiar sound was a practical response to uncertain infrastructure. A telephone circuit could contain noise, filtering, echo and distortion. The devices could not assume that a rate supported in principle would survive this call in practice.
The handshake therefore joined compatibility testing to channel measurement. The devices had to recognise one another, establish timing and equalisation, and settle on parameters that both sides and the line could support. A higher number was not automatically a better outcome if it produced errors or repeated retraining. Negotiation converted an imperfect analogue path into a temporary digital arrangement. Disconnecting erased that arrangement; the next call had to discover the world again.
The asymmetric promise
The phrase 56K described a special network geometry, not a universal speed.
Recommendation V.90 formalised a pairing in which the direction toward the analogue subscriber could reach a nominal 56,000 bit/s under required digital-network conditions, while the reverse direction remained limited to V.34-class rates up to 33,600 bit/s. The asymmetry depended on where analogue conversion occurred. It was not simply a faster version of two identical modems talking across arbitrary copper from end to end.
For users, the advertised rate became an aspiration filtered through line quality, provider equipment and regulation. The number reported after connection was itself a negotiated result rather than a guarantee of application throughput. Protocol overhead, errors, server load and congestion remained outside it. This layered uncertainty is why memories of dial-up include both precise figures and profound unpredictability. A status line could say 48,000 bit/s while a web page still appeared one object at a time.
After the sound stopped
Carrier was the beginning of a connection, not the completion of networking.
The Point-to-Point Protocol documented in RFC 1661 shows what followed the physical link. PPP established and tested the link, could perform authentication, and then configured one or more network-layer protocols. To the listener, the transition to silence sounded like arrival. In protocol terms, several forms of agreement could still remain. The modem had made a channel; the computer still had to make that channel useful.
This sequence survives in quieter systems. Wi-Fi association, Bluetooth pairing, TLS negotiation and media capability selection all establish shared conditions before an application can proceed. Their interfaces tend to collapse the phases into a spinner or a single verb. Dial-up exposed them because each stage was slow enough to notice and because part of the apparatus used a speaker. The sound is historically distinctive, but the contractual logic is contemporary.
Open bibliography
Records consulted
The essay above is original interpretation, not quotation. These records establish its technical mechanisms, representative figures and historical boundaries.
- 01
Recommendation V.8: Procedures for starting sessions of data transmission
Documents the signals and exchanges used by modems to discover one another's capabilities before a high-speed session begins.
- 02
Recommendation V.34: Modems up to 33,600 bit/s
The engineering specification behind rate negotiation, training sequences, symbol rates and adaptive operation on telephone lines.
- 03
Recommendation V.90: A digital modem and analogue modem pair
The 56K-era specification: up to 56,000 bit/s toward the analogue subscriber and up to 33,600 bit/s in the reverse direction, under the required network conditions.
- 04
RFC 1661: The Point-to-Point Protocol
Shows that an audible carrier was only the physical beginning: link configuration, optional authentication and network-layer configuration still followed.