StumbleUpon

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

~ the eulogy ~

stumbleupon launched in november 2002 as a browser toolbar. you installed it on firefox or internet explorer. you picked some categories you were interested in (technology, photography, weird news, whatever). then you clicked the stumble button at the top of your browser and you were taken to a random web page that someone else had liked in those categories. you clicked the button again and you went to another page. you could thumbs-up or thumbs-down each page to refine the recommendations.

it was the first product to do this kind of random-web discovery at scale. it pre-dated reddit's rise (founded 2005, took a few years to get big). it pre-dated twitter (2006) and definitely pre-dated the social-media-curated-recommendations model. for users in the mid-2000s, stumbleupon was how you found new websites that nobody else knew about.

i used stumbleupon between roughly 2007 and 2010. it was the procrastination button on my browser. essay due tonight, but first one quick stumble. one quick stumble would turn into 30 minutes of clicking through random pages. a strange flash animation. a niche photography blog. an academic page about old typewriters. each click was a small surprise. some of the pages i remember from that era i would never have found any other way.

~ the dossier ~

~ how it worked ~

the technical model was a recommendation engine plus a toolbar. users rated pages thumbs-up or thumbs-down. the engine learned what each user liked. when you clicked stumble, the engine picked a page from its database that other users with similar tastes had thumbed up.

the database grew through user submissions. anyone could submit a page. the system would show submitted pages to early users in the relevant categories, and if those users thumbs-upped the page it would propagate to other users with similar interests. this was a pure user-curated web index, with the recommendation layer making it personal.

the model worked because the early-2000s web had a lot of small, idiosyncratic sites that nobody could discover any other way. google was still mostly about answering specific queries. stumbleupon's role was different. it was discovering small sites you would never type into a search box because you did not know they existed.

~ the curtain ~

ebay bought stumbleupon in may 2007 for 75 million dollars. the integration was bad. ebay tried to monetize through paid placements, which damaged the recommendation quality. founder garrett camp left in 2009 (he later co-founded uber). the original product team scattered.

in april 2009 the founders bought stumbleupon back from ebay. they ran it as an independent company for the next nine years. during this time the web changed. social media became the main way people discovered new content, and reddit and twitter handled the "random interesting thing" use case. stumbleupon's user numbers declined every year after 2012.

on 23 may 2018 the company announced that stumbleupon would shut down on 30 june 2018. the existing user base would be migrated to a much weaker successor called Mix, also from the stumbleupon team. mix tried to be a more social and curated version. nobody used it. mix is technically still alive in 2026 but it has almost no traffic.

~ what it left ~

the random-web concept. stumbleupon's button is the ancestor of every cross-site "random page" feature now embedded in modern projects. cloudhiker and the wiby search engine are two small modern attempts to revive the random-web discovery experience. neither has reached stumbleupon's scale.

the lesson about acquisition damage. ebay bought stumbleupon for 75 million dollars and made it worse. the founders bought it back for less than the original sale price. this is one of those rare cases where startup founders undid a bad acquisition by buying their company back. the more common version is that the acquirer kills what they buy by changing the incentives.

the missed opportunity. stumbleupon had a 40-million-user audience in 2012 and could have pivoted into a serious content discovery product. they did not. by the time they tried with mix it was too late. the random-web button should have evolved into something competitive with reddit but never quite did.

@}-->--

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

« previous  ·  random  ·  next »