Outline.com

» see outline.com as it lived, on the wayback machine

~ the eulogy ~

You pasted a URL into Outline and got back the same article with everything else removed. No newsletter overlay, no cookie wall, no video that started playing in the corner, no sidebar of things other people were reading. Just the headline and the text, in one column, in a typeface chosen by somebody who wanted you to finish the piece. It also let you highlight passages and keep notes against them. That was the half of the product nobody ever mentioned.

The other half is why it was famous. Outline fetched pages the way a search engine crawler fetches them, and a great many publishers in that era served crawlers the full article while serving humans a paywall. So the clean reading view arrived with the whole story inside it. By April 2017 people on Hacker News were posting links through it without explaining what it was, including links to the Wall Street Journal, which tells you both how fast it got adopted and how obvious the tension was from the start.

No founder ever stood in front of it. There is no launch interview and no about page with a photograph on it. Someone ran it for five years, paid for the bandwidth, kept the parser working against a moving target of a thousand different content management systems, and never put a name on any of it.

~ the dossier ~

~ the trick ~

The parser was the engineering, and it was harder than it looked. Every publisher nests its article text differently and renames its CSS classes on a redesign, then pads the page with promotional blocks that look identical to paragraphs as far as a parser can tell. Outline got it right often enough that people stopped checking. It got it wrong sometimes too, refusing certain articles for no reason a user could see, something several people noted in its own death thread without much bitterness. It needed JavaScript to render, and it exposed an API, so by the end there were browser extensions and bookmarklets and at least one project whose entire purpose was to redirect paywalled domains through it automatically. That is the shape of a utility that has become infrastructure: other people's software depending on your uptime without ever asking you.

~ the week it went quiet ~

The last capture of a working Outline is 28 February 2022. On 5 March somebody opened an Ask HN saying the site had been unreachable for about ten days and the domain still looked fine, and the thread filled up with people naming alternatives instead of answering the question.

One of the guesses was legal takedowns. Another was that a short, valuable domain had simply been sold. On 31 March the first archived capture of the new outline.com is a Sedo listing: the domain, without content, available from its owner, any offer binding for seven days. The second theory won.

I read the English-language technology press from Turkey, where a lot of it has never been purchasable in lira at any price, and where the choice a metered paywall offers is frequently just no. That is the audience Outline served best and the reason its death notice was a Hacker News thread full of people trading workarounds. Nothing was announced because there was no company to do any announcing. There was one person's hosting bill, and then there was not.

~ what people used instead ~

Within a day of the Ask HN thread the replacements were already listed: 12ft.io, which itself closed later, the Internet Archive, browser extensions maintained by volunteers, and Firefox's own reader view for the ordinary case of a cluttered page. The publishers, meanwhile, closed the crawler hole that had made the trick work, so even a perfect clone of Outline built today would mostly return a paywall in a nicer font.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »