Algorithmic
2010 – present. The feed replaces the broadcast schedule. Instagram, Snap, TikTok rank what you see by predicted engagement, not chronological order. Smartphones ship with cameras better than the news crews of the 2000s carried. By 2022, the medium starts generating itself — diffusion models, large language models, synthetic everything.
Part of Media
29 moments in this segment.
- — Spotify and the algorithmic playlist. Spotify launched in 2008, replacing ownership with access to a near-infinite library — and then with algorithmic playlists (Discover Weekly, 2015) that let software curate what you hear.
- — The App Store & the home screen. Apple’s App Store turned the phone into a platform — and the grid of app icons into the doorway to all media.
- — Spotify & streaming music. Spotify made the world’s music an instant, algorithm-curated stream — the post-download model.
- — Facebook's Like button. Facebook added the Like button in February 2009. That single tap turned every interaction into a quantified signal, giving the ranking algorithms the fuel to sort the feed by predicted engagement — the mechanism at the heart of the algorithmic age.
- — Instagram. Kevin Systrom, Instagram co-founder, in 2011 — a year after launching the photo app that Facebook would buy two years later for $1 billion. Kevin Systrom and Mike Krieger launch Instagram on October 6, 2010 — initially as a photo-only app with built-in filters that make iPhone photos look like Kodachrome film. 25,000 users in the first 24 hours. Facebook buys it for $1 billion two years later, on the eve of its IPO.
- — Instagram. launches on 6 October 2010 as a photo-sharing app built around filters and the square frame, hitting a million users in two months. Bought by Facebook for $1 billion in 2012, it makes the visual feed — and the influencer — central to online culture.
- — Snapchat — Ephemeral Media. Evan Spiegel, Snapchat co-founder — pictured in 2013, the year the app shipped Stories and crossed 50 million users. Stanford undergrads Evan Spiegel, Bobby Murphy, and Reggie Brown launch Snapchat in September 2011. Photos that delete themselves after seconds. The form invents Stories (1-day expiry) in 2013; Instagram clones it in 2016. Ephemeral becomes a baseline expectation of social media.
- — Snapchat — Ephemeral Media. Snapchat launches in September 2011 with a radical idea: messages that disappear after they're viewed. Evan Spiegel's app makes ephemerality and the "Story" format mainstream — conventions that Instagram, Facebook and TikTok would all copy.
- — "Gangnam Style" — the viral billion. The first video to a billion YouTube views showed media going global and viral at internet speed.
- — House of Cards and data-driven TV. Netflix's House of Cards (2013) was greenlit and shaped by viewing data, and released all at once for bingeing. Streaming originals rewired how television is made, funded and watched.
- — Netflix & the streaming wars. Netflix’s shift to streaming and original series reshaped television into on-demand, algorithm-served content.
- — The creator / influencer economy. A generation built media careers directly on platforms, bypassing the old gatekeepers.
- — TikTok / Douyin. Zhang Yiming, founder of ByteDance, in 2014 — two years before the company launched Douyin in China; TikTok would follow internationally in 2017. ByteDance launches Douyin in China in September 2016; the international version TikTok ships in 2017 and merges with the U.S. lip-sync app Musical.ly in 2018. The For You Page — a machine-learned feed driven entirely by behavioural signal — becomes the most copied product idea of the decade.
- — TikTok / Douyin. ByteDance launches Douyin in China in September 2016 and the international TikTok in 2017, merging with Musical.ly in 2018. Its recommendation algorithm — serving an endless personalised feed of short vertical video — becomes the defining media form of its era.
- — Cambridge Analytica and the data reckoning. The 2018 Cambridge Analytica scandal revealed how the personal data of tens of millions of Facebook users had been harvested for political targeting — a turning point in public awareness of surveillance-driven media.
- — Misinformation & the feed. Algorithmic feeds optimised for engagement made the spread of misinformation a defining media problem.
- — Deepfakes and synthetic media. From 2017–2019, machine-learning "deepfakes" made it possible to convincingly put words in anyone's mouth on video, opening an era in which seeing is no longer believing.
- — Deepfakes & synthetic media. AI-generated faces and voices blurred the line between real and fake recordings.
- — Pandemic Pushes Everything Online. Times Square in March 2020, near-empty during the early COVID-19 lockdowns that pushed Zoom, Netflix, and Twitch usage to unprecedented levels. COVID-19 lockdowns from March 2020 onward push the entire mediated stack into the home. Zoom adds 300 million daily users in three months. Netflix gains 26 million subscribers in six months. Twitch viewership doubles. The "post-broadcast" media future arrives compressed into a single quarter.
- — Pandemic Pushes Everything Online. The COVID-19 lockdowns of 2020 push work, school, worship and social life onto screens almost overnight. Zoom, streaming and social platforms absorb daily life at a scale that compresses a decade of digital adoption into a few months.
- — The creator economy. By the early 2020s, platforms from YouTube to TikTok to Patreon had built a "creator economy" in which individuals reach mass audiences and earn directly — collapsing the old line between broadcaster and viewer.
- — Stable Diffusion + DALL-E 2 + ChatGPT. Sam Altman, OpenAI CEO — pictured at TechCrunch SF 2019, three years before ChatGPT launched on November 30, 2022 and crossed 100 million users in two months. Three diffusion / transformer models drop within months of each other in 2022: DALL-E 2 (April), Stable Diffusion open-source release (August), ChatGPT (November 30). Synthetic image and text generation cross the chasm into mainstream. The first time the medium starts producing itself.
- — Stable Diffusion + DALL-E 2 + ChatGPT. 2022 is the year generative AI goes mainstream: DALL·E 2 and Stable Diffusion make text-to-image generation public, and OpenAI's ChatGPT — released 30 November 2022 — reaches 100 million users in two months. Media production itself becomes something software can do on command.
- — ChatGPT & generative AI. ChatGPT made conversational AI mainstream, putting a text-generating machine in everyone’s hands.
- — Threads Launches in 5 Days. Meta launched Threads on July 5, 2023 in response to Musk's acquisition and degradation of Twitter — the app hit 100 million signups in five days, the fastest consumer-product growth curve in internet history. Meta launches Threads on July 5, 2023 — an Instagram-anchored text-social app, built as a direct response to Elon Musk's acquisition and degradation of Twitter. 30 million signups in 24 hours, 100 million in five days. The fastest consumer-product growth curve in internet history.
- — Threads Launches in 5 Days. Meta launches Threads on 5 July 2023 as a text-based rival to Twitter/X, and it reaches 100 million sign-ups in five days — the fastest-growing app in history to that point — riding turmoil at X after Elon Musk's takeover.
- — Sora & AI video. OpenAI’s Sora generated photorealistic video from text — the next frontier of synthetic media.
- — Sora, Veo, Runway Gen-3. Sundar Pichai, Google/Alphabet CEO — Veo, Google's text-to-video model, dropped in May 2024 alongside OpenAI's Sora demo (February) and Runway Gen-3 (June). The medium starts generating itself. OpenAI demos Sora in February 2024, Google releases Veo in May, Runway ships Gen-3 in June. Text-to-video generation reaches the quality threshold where individual outputs are indistinguishable from short film clips at first glance. The medium is now producing not just images and text, but minute-long moving footage.
- — Sora, Veo, Runway Gen-3. By 2024 text-to-video generation crosses a threshold: OpenAI's Sora, Google's Veo and Runway's Gen-3 produce startlingly coherent minute-long clips from a prompt. The line between filmed and generated footage begins to blur, raising new questions about what a moving image even is.
