Chronology · finding aid 02
This is not a parade of inventions. It is a record of moments when waiting changed shape: from scanned line to institutional queue, from moving head to negotiated link, from visible spinner to invisible policy.
Orientation
A date in this chronology identifies a useful point in the surviving record. It does not automatically identify the first experiment, the first commercial object or the first ordinary encounter. Those moments are rarely the same. Entries use ranges when the evidence describes a transition, and precise years when the cited record supports a specific demonstration, standard, paper or object.
The short paragraph describes mechanism and context. The ruled sentence underneath identifies why that moment matters to human time. Sources are open beside every entry, so an engineering specification never has to impersonate a social history. Related rooms lead to original simulations in the permanent collection.
Condensed register
The table is a compact alternative to the long-form gallery below. It can be scrolled horizontally on small screens; every row links to the full record.
1830s–1959
Before a computer could make a person wait, documents and public information had to become sequences: scanned lines, magnetic tracks and physical characters moving through time.
CH-001
1830s–1840s
Early facsimile work established the durable premise that a surface could be sampled, converted into a signal and rebuilt somewhere else. The historical record contains several experimental lineages, so the museum treats this period as a field of proposals rather than a single immaculate beginning.
The page acquired an order in time: one region could arrive while another had not yet left.
CH-002
1844
The Library of Congress preserves the paper tape associated with the Washington-to-Baltimore telegraph demonstration of May 1844. The object records not merely language but arrival: distant action became an electrical sequence that could mark a local strip and be inspected afterward.
Communication acquired a machine-readable interval between dispatch and material evidence.
CH-003
1860s–1920s
Successive facsimile systems had to coordinate motion at two ends of a connection. A transmitter and receiver needed compatible timing, a shared account of light and dark, and enough stability for a line to remain a line when reconstructed at a distance.
Waiting was embedded in synchronization: fidelity depended on both machines advancing together.
CH-004
1920s–1930s
Long before the office fax machine, facsimile occupied specialist networks and institutional workflows. Its costs, apparatus and operating knowledge kept it far from the casual gesture of placing a sheet in a tray, but the service proved that a distant page could have a measurable transmission time.
The document became an event with a start, duration, operator and destination.
CH-005
1940s–1950s
Expensive central machines encouraged work to be prepared away from the processor and delivered in groups. The computation itself might be brief, but transport, operator handling, queue position and printed return converted a small mistake into a long institutional round trip.
Turnaround time, not processor time, became the unit in which a programmer felt the machine.
CH-006
1949
The Computer History Museum's networking chronology places an early modem in 1949, used to carry digital information over an existing voice-network channel. This was not the familiar home dial-up session, but it established the useful translation at the centre of one: discrete machine states could travel through infrastructure built for sound.
A mature network gained a second life, and data inherited the network's delays, noise and procedures.
CH-007
1950s
Tape and rotating magnetic media stored data as physical position. Sequential tape made duration proportional to how far the medium had to travel; rotating drums and disks made access depend on where a requested record happened to be when the mechanism was ready.
A request for information became a request for matter to move into the right place.
CH-008
1956
An electromechanical character system installed for railway information in Liège belongs to the early history documented for Solari's public displays. Each character changed by advancing physical flaps, making an update audible before a traveller could read its result.
The state of an institution announced itself through the duration and scale of a mechanical change.
CH-009
Late 1950s
Accounts of early time-sharing frame the conventional batch monitor as a costly feedback loop. When diagnosis required a new submission, one small correction might consume hours; the desire for conversational access grew from that mismatch between machine speed and human turnaround.
The important question changed from how fast the computer ran to how soon a person could learn what it had done.
1980–1989
Cassette tones, moving disk heads and installation rituals made computation domestic. The machine was personal, but its time still belonged to motors, media and queues.
CH-020
1980
The ITU history describes Group 3 standardization as a major acceleration, bringing representative page time below a minute under suitable conditions. Digital coding and agreed call procedures made the fax a routine office appliance rather than an exceptional transmission service.
The transmitted page became fast enough to feel immediate and slow enough to remain visibly sequential.
CH-021
Early 1980s
A cassette recorder beside a home computer turned software access into a shared acoustic event. Signal quality depended on alignment, level, connectors and the physical condition of tape; family members learned that an apparently idle machine might be listening with great concentration.
A technical transfer announced itself to everyone in the room, including people who had not asked to use the computer.
CH-022
1984
Commodore's service documentation treats the domestic data cassette as a mechanism requiring alignment, inspection and electrical diagnosis. The manual makes visible how much reliable waiting depended on material tolerances hidden behind the command to load.
User superstition sat on top of a real analogue maintenance problem.
CH-023
1980s
Smaller flexible disks made removable storage ordinary on personal machines. A representative 3.5-inch mechanism rotated at 300 rpm, while a stepper moved the head between tracks; even a request expressed in software remained a choreography of seek and rotation.
Clicks and motor changes let a listener infer that a machine was searching, retrying or finally reading continuously.
CH-024
1985
Brad A. Myers compared an interface with a percent-done indication against one without it in a controlled 48-participant study. The paper reported a strong preference for visible progress, while carefully studying the indicator rather than claiming to invent it.
Evidence accumulated that explaining a wait could change experience even when it did not shorten the work.
CH-025
Mid–late 1980s
Repeated writing and deletion left files in separated allocation runs on mechanical media. Consolidation tools translated this invisible topology into a task a personal-computer owner could schedule, watch and interpret as care for the machine.
Waiting gained a moral register: time spent restoring order promised future speed.
CH-026
Late 1980s
The GIF format included an interlaced row order that distributed early data across the height of an image. On a constrained link or processor, a coarse whole could become legible before the complete raster had arrived.
The same total data could produce a more useful wait simply by changing its order.
CH-027
Late 1980s
Computer animation workflows treated a moving image as many separate calculations. Frames could be distributed, retried and assembled later; the durable image-sequence method turned one enormous wait into a ledger of smaller completed and incomplete tasks.
The queue became both a production schedule and a recovery strategy.
CH-028
1989
RFC 1122 gathers host requirements for Internet communication, including acknowledgement behaviour, retransmission timing and the TCP TIME-WAIT state. These intervals demonstrate that waiting inside a protocol can protect reliable delivery and prevent old traffic from being mistaken for a new conversation.
Not every network delay is lost time; some delays preserve the meaning of sequence.
CH-029
1989–1991
CERN's history places the Web proposal in 1989, a working server and browser by the end of 1990, and public release in 1991. A page could now collect linked documents over a network and present their arrival through one interface, making remote retrieval an ordinary visual wait.
The screen became a temporary assembly site for objects that did not begin together or arrive together.
1990–1999
A telephone call became a temporary data link. Images learned to reveal themselves early, progress indicators made estimates public, and maintenance turned invisible storage into spectacle.
CH-030
1990
The archived GIF89a specification describes a four-pass sequence of rows for interlaced images. The format made the mechanics inspectable: early passes sample the whole height, while later passes fill the spaces between them.
Progressive revelation was not decoration; it was a property encoded into the file.
CH-031
Early 1990s
Large personal-computer applications crossed several removable disks and several internal phases. The machine calculated costs, copied, registered and finalized, but paused whenever the next physical volume required a human hand.
The user became a peripheral in the machine's workflow: essential, intermittent and slower than memory.
CH-032
1993
A durable usability account distinguishes rough orders of response: an instant that preserves a sense of direct manipulation, a short pause that keeps thought continuous, and a longer interruption that risks losing attention. These are design heuristics, not biological constants.
Delay became something evaluated by what it did to attention, not only by a stopwatch.
CH-033
1994
RFC 1661 specifies the Point-to-Point Protocol as a sequence of link establishment, optional authentication, network-layer configuration and termination. A modem could fall silent after its audible negotiation while several important agreements still remained.
Connection became a stack of waits, only the first of which could be heard.
CH-034
Mid-1990s
V.34-era modems used training and adaptive operation to choose a rate a particular voice-band channel could sustain, up to the standard's upper signalling limits under suitable conditions. Hang up, call again and the answer could change.
The handshake was a temporary measurement contract, not a fixed property printed on the box.
CH-035
1995
The Computer History Museum documents a consumer CD-R drive priced below $1,000 in 1995 and the representative 650 MB medium. At 1× speed, a full disc could occupy roughly eighty minutes; later 52× mechanisms compressed a comparable write into around a minute and a half under appropriate conditions.
The same storage object crossed from leave-it-running ritual to short foreground operation as mechanism speed changed scale.
CH-036
1996
PNG's Adam7 interlace schedules pixels through seven passes across a repeating grid. A decoder can display a broadly distributed preview, then refine it, preserving an early sense of the whole without pretending that the image is complete.
A format could allocate the visitor's attention before it had delivered every pixel.
CH-037
Around 1996
A period W3C account describes progressive JPEG as presenting an early low-detail whole that gains definition as later scans arrive, with browser implementation appearing around 1996. It reached a similar human threshold to interlaced GIF and PNG through different image coding machinery.
A visitor could decide what an image was before the network had finished deciding how clearly it could be seen.
CH-038
1998
V.90 describes a digital-server and analogue-subscriber modem pair with different theoretical limits in each direction under the network conditions it assumes. The familiar '56K' name therefore condensed a directional and conditional system into one memorable number.
Marketing offered a speed; the standard documented a route, a direction and a set of qualifications.
CH-039
Late 1990s
Graphical defragmentation utilities represented allocation units as changing fields of colour. The map compressed millions of moves into a legible surface, encouraging people to watch maintenance they could neither safely interrupt nor usefully accelerate.
An operational report accidentally became ambient entertainment.
CH-040
Late 1990s
Installers, downloads and maintenance tools placed percentage and remaining time in ordinary domestic view. Yet many tasks discovered their total work only while performing it, so the display repeatedly had to renegotiate what 'nearly finished' meant.
A technical estimate acquired the social weight of an appointment.
2000–2011
Queues grew into farms and buffers became reserves of future media. Faster systems did not end waiting; they split it into smaller stages and moved it out of sight.
CH-041
2000
The V.8 recommendation documents procedures by which data-terminal equipment advertises and discovers communication modes at the start of a telephone-network session. Even as the consumer experience was becoming familiar, connection still depended on a carefully ordered exchange.
The ritual sound belonged to interoperable procedure, not to random electronic distress.
CH-042
Early 2000s
Production rendering distributed frames and subtasks across many processors. Capacity grew dramatically, but so did the appetite for geometry, lighting and revision; the result was not the disappearance of waiting but a more sophisticated institution for deciding what would render next.
Parallel work made the queue scalable rather than obsolete.
CH-043
2003
A Pixar production paper reports that, around the peak of Finding Nemo work, roughly 3,000 processors could still face more than 10,000 tasks in the queue. The figures belong to that production environment, not to a universal measure of animation.
More machinery changed the scale of waiting while preserving priority, failure and morning inspection.
CH-044
2003–2005
Published revisions of T.4 and T.30 document the representation of facsimile pages and the procedures surrounding a call. Their continued formal maintenance shows that an apparently familiar office ritual was an interoperable system of negotiation, training, transmission and confirmation.
Mature waiting is often the most standardized waiting: reliable because every stage has a name.
CH-045
2000s
Streaming media asked a player to consume content while transport was still delivering it. A reserve of future material could absorb ordinary variation; if consumption caught the available data, the interface stopped and made the invisible shortage visible as buffering.
The present depended on a small stockpile of the future.
CH-046
2000s
Filesystems, scheduled maintenance and changing storage media reduced the need for a person to watch blocks move. Consolidation remained meaningful for some mechanical volumes, while flash storage required different maintenance choices and made indiscriminate rewriting undesirable.
A wait can vanish from culture while the underlying housekeeping continues under another policy.
CH-047
2000s
Faster links did not make partial images irrelevant. The older logic of interlacing—provide distributed meaning early—survived in previews, placeholders and staged rendering, separating the time at which a visitor can decide from the time at which every byte is present.
Perceived completion and technical completion became explicitly different milestones.
2012–2026
Adaptive delivery, accessibility guidance and formal metrics treat waiting as something to shape. The remaining delays belong as much to institutions and expectations as to raw speed.
CH-048
2012
A large observational study of online video sessions associated startup delay and interruptions with measurable changes in viewer behaviour. Its conclusions belong to the studied service and dataset, but the work helped make waiting a business and experience metric rather than an anecdote.
The spinner stopped being only a technical symptom and became a quantifiable point of abandonment.
CH-049
2013
RFC 7005 describes reporting for de-jitter buffers that deliberately retain media before playout. The buffer's delay is not simply a fault: it exchanges immediacy for resilience against variation in packet arrival.
Some waits exist because removing them would make the experience less continuous.
CH-050
2017
RFC 8216 describes HTTP Live Streaming as segmented media with playlists and variant streams. A client can choose among representations as conditions change, often reducing visible stalls by changing image quality instead.
The obvious wait was exchanged for a quieter compromise that could happen while playback continued.
CH-051
2022
RFC 9317 frames streaming as an interaction among transport, buffer and media consumption. The document gives operational language to a domestic experience: the player must keep enough future content on hand that ordinary delivery variation does not interrupt the present.
Buffering becomes a managed system state rather than a mysterious wheel on the screen.
CH-052
2022
RFC 9112 specifies HTTP/1.1 message syntax, including chunked transfer coding for content whose complete size need not be stated before transmission. Length-delimited pieces can arrive and a final chunk can mark completion, allowing useful delivery without a known denominator.
A transfer can be rigorously complete even when no progress bar could honestly predict its percentage.
CH-053
2023
WCAG guidance asks interfaces to let people disable non-essential animation triggered by interaction when that motion can cause discomfort. A loading or progress display must therefore preserve information without assuming that movement is harmless or necessary.
Designing a wait now includes deciding what not to animate.
CH-054
2024
The living HTML standard continues to distinguish a progress element with a known value from one whose duration is unknown. Contemporary software may be faster, but the semantic problem survives: sometimes a system can measure completion, and sometimes it can only testify that work continues.
Honesty about uncertainty remains an interface capability, not an admission of failure.
CH-055
2026
The Museum of Waiting rebuilds selected behaviours with original code, language, diagrams and synthesized sound. Standards establish mechanism; institutional histories supply context; interface research tests some human consequences; uncertainty remains visible where memory and measurement cannot be made equivalent.
A vanished inconvenience can be studied as a relationship among time, machinery and trust without impersonating the product that once contained it.
Catalogue note
Devices and practices overlapped for decades. A date at which one standard was published is not the date every office, home or country adopted it. Older waits often continued beside newer ones.
Transfer rate, media condition, line quality, workload and local practice vary. Representative figures remain attached to the sources and contexts that make them meaningful.
Legibility did not make delay pleasant, and familiarity did not make old systems equitable. The museum studies what a wait revealed without asking visitors to want it back.
This is a revision-controlled public finding aid. Corrections should identify the accession, the contested wording and an institutional or primary source that improves it.