Industries / Media, arts and recreation / Media and entertainment
Streaming and distribution
This is the work of getting finished programs to audiences: broadcasting over the air, cable, cinemas, home video and streaming over the internet, along with the captions and descriptions that let more people follow them.
7
Milestones achieved, from 1920 to 2025.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
In depth
The same ground is charted in more depth, as industries of their own.
The road
AchievedWatch forOpen
Achieved
1920
Radio on a schedule
On 2 November 1920 the Westinghouse station KDKA in East Pittsburgh began scheduled programming by reading out the Harding–Cox presidential election returns, sending 100 watts on a 360-meter wavelength from a shed on a factory roof. The IEEE, the engineers' professional body, calls KDKA "a world pioneer of commercial radio broadcasting" and says the idea grew out of engineer Frank Conrad's weekly experimental broadcasts from his amateur station, 8XK.
Credit
Westinghouse; Frank Conrad; C.P. Davis, who conceived the service
1980
Captions for deaf viewers
On 16 March 1980 PBS, NBC and ABC broadcast programs with closed captions: text carried hidden in the TV signal that appeared on screen only through a decoder box, sold by Sears. At launch the networks captioned 16 to 20 hours a week, mostly prime-time entertainment, by the account of Barry Cronin of the National Captioning Institute, written that year.
LessonCaptions reached few people while viewers had to buy a separate box; the US Television Decoder Circuitry Act of 1990 required decoders built into sets with screens larger than 13 inches.
Credit
National Captioning Institute; PBS, NBC and ABC
1984
Home taping ruled legal
On 17 January 1984 the US Supreme Court ruled in Sony v. Universal that recording a broadcast at home to watch it later, which the Court called "time-shifting", was fair use, so Sony was not liable for selling Betamax video recorders. Both sides' surveys of Betamax owners found that time-shifting was the main use.
LessonThe studios predicted harm, but the trial court found they had not proved any likely harm from time-shifting, and the Court found that it enlarged the TV audience.
Credit
US Supreme Court; Sony
2007
Movies streamed, not mailed
On 16 January 2007 Netflix, then a DVD-by-mail rental company, added a "Watch Now" tab that played about 1,000 movies and TV series in a web browser on a personal computer, starting in 10 to 15 seconds without first downloading the whole file. Viewing hours were capped by plan, at six hours a month on the $5.99 plan.
Credit
Netflix
2012
Cinemas switch to digital
In July 2005 Digital Cinema Initiatives, a body formed in 2002 by Disney, Fox, Paramount, Sony Pictures, Universal and Warner Bros., approved a common specification for showing films from digital files. By the end of 2012, more than two-thirds of the world's nearly 130,000 cinema screens were digital, by the count of David Hancock of IHS Screen Digest, and a wide US release that once needed 3,000 to 5,000 film prints shipped a few hundred.
LessonConversion cost up to $100,000 per screen, as film scholar David Bordwell reported; studios helped pay through a "virtual print fee" on condition that cinemas removed their 35mm projectors, and small independent cinemas had to raise the money themselves.
Credit
Digital Cinema Initiatives; the cinema chains that converted
2020
Quibi closes in months
Quibi, which cut its shows into 10-minute episodes for people to watch while commuting or standing in line, raised about $1.8 billion and announced on 21 October 2020, seven months after launch, that it would shut down. Its founders, Jeffrey Katzenberg and Meg Whitman, wrote that either "the idea itself wasn't strong enough to justify a standalone streaming service" or the timing was wrong.
LessonMoney for programs did not create demand; the app research firm Sensor Tower estimated that only 8% of free-trial users went on to pay.
Credit
Quibi (Jeffrey Katzenberg, Meg Whitman)
2025
Streaming passes TV channels
In May 2025 streaming made up 44.8% of all TV viewing in the US, more than broadcast (20.1%) and cable (24.1%) combined, the first time it had done so in Nielsen's monthly report, The Gauge. YouTube alone had 12.5% of TV viewing.
Credit
measured by Nielsen; the streaming services
Watch for
Broadcast TV's next standard
The US Federal Communications Commission voted unanimously on 28 October 2025 to propose rules that let TV stations decide for themselves when to stop broadcasting in ATSC 1.0, the digital standard most sets receive today, and move to ATSC 3.0, sold as NextGen TV, without a firm end date or a rule that new sets carry 3.0 tuners. The National Association of Broadcasters says NextGen TV is on the air in more than 80 markets and has asked the FCC to set shutoff dates, starting with the top 55 markets in February 2028.
Achieved whenThe FCC adopts final rules, and the big-network stations in at least one US market switch off ATSC 1.0 with an outside count of the viewers who kept or lost free TV.
Claimed by
FCC and National Association of Broadcasters, October 2025
Auto-dubbing at YouTube scale
YouTube says its automatic dubbing, which translates the speech in a video into another language with a synthetic voice, became available to all users in 27 languages in February 2026, and that in December 2025 more than 6 million viewers a day watched at least 10 minutes of dubbed video. It is also piloting lip sync, which adjusts speakers' mouths to match the translated words.
Achieved whenAn outside measurer reports how much dubbed video is watched, or a published panel of native speakers rates YouTube's dubs against human dubbing of the same videos.
Claimed by
YouTube, February and May 2026
Open
01
Describe public-domain films
Audio description, the spoken track that tells blind and low-vision viewers what is happening on screen between the lines of dialogue, for full-length feature films that are in the public domain. An AI model drafts each description and places it in a gap in the dialogue, a trained describer edits the draft, and the finished tracks are published free alongside the films.
The briefCredit · unclaimed02
Map where free TV reaches
A measured map of broadcast TV reception across one US metro area. Volunteers place a network TV tuner with an antenna at their homes, and a small program logs the signal strength and quality of every station in both the ATSC 1.0 and ATSC 3.0 standards every few minutes for a month. The map compares the measurements with the coverage the FCC's model predicts. It logs signal quality only and does not record programs.
The briefCredit · unclaimed
Open 01 · the brief
Describe public-domain films
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
Audio description, the spoken track that tells blind and low-vision viewers what is happening on screen between the lines of dialogue, for full-length feature films that are in the public domain. An AI model drafts each description and places it in a gap in the dialogue, a trained describer edits the draft, and the finished tracks are published free alongside the films.
Why it is still open
US rules require description only from the stations of the four big networks in the top 90 markets and from large cable and satellite systems, at 87.5 hours per quarter per channel. The authors of the ADQA benchmark (October 2025) write that automatic description research mostly works on clips a few seconds long and that current methods "lag far behind human-authored ADs". YouDescribe, a crowdsourced description site, added AI-drafted descriptions that volunteers edit in 2025, but it serves user-made YouTube videos, and its researchers say only 7% of requested videos have been described. This research found no library of public-domain feature films described this way.
Built with what exists
Public-domain feature films from the Internet Archive; a vision-language model that reads long video, such as Gemini or Claude Fable 5.2 (see the Note "Three working devices"); speech-to-text with word timings, such as ElevenLabs Scribe v3 (see the Note "Fewer word errors"), to find the gaps in the dialogue; Able Player, an open-source web video player that reads WebVTT description tracks aloud with the browser's speech synthesis.
First step
Pick one public-domain feature, transcribe its dialogue with word timings, have a model draft descriptions for the first 20 minutes, and time how long a describer takes to fix them.
Done looks like
Ten feature films described, with the describer's editing time logged per film; for two of them, the same describer also writes a track from scratch, and at least 12 blind or low-vision viewers answer plot and visual questions in the style of ADQA after watching, scoring within 10% of the hand-written track.
Open 02 · the brief
Map where free TV reaches
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A measured map of broadcast TV reception across one US metro area. Volunteers place a network TV tuner with an antenna at their homes, and a small program logs the signal strength and quality of every station in both the ATSC 1.0 and ATSC 3.0 standards every few minutes for a month. The map compares the measurements with the coverage the FCC's model predicts. It logs signal quality only and does not record programs.
Why it is still open
Public TV coverage maps are predictions. RabbitEars.info, the most used station-lookup site, computes them with the FCC's open-source TVStudy software and topographic data, and its own guide notes that trees and buildings in an antenna's view "aren't shown on any topographic database". The FCC's October 2025 proposal asks how to keep free over-the-air TV available in all homes during the switch to ATSC 3.0, and this research found no published crowdsourced reception measurements to answer it.
Built with what exists
SiliconDust HDHomeRun tuners (the FLEX 4K, $199.99, receives both standards) and indoor or rooftop antennas; SiliconDust's open-source libhdhomerun library, which reads each tuner's signal strength, signal-to-noise quality and symbol error quality; a Raspberry Pi per home for logging; the FCC's TVStudy software for the predicted coverage; QGIS for the map.
First step
Set up one HDHomeRun and a logging script in a single home, record a week of readings for every local station, and compare them with that address's RabbitEars prediction.
Done looks like
At least 100 homes in one metro area logged for four weeks; for each station, the share of homes where measured reception agrees with the predicted service, published with an open dataset, and the comparison repeated after any station in the area switches off ATSC 1.0.