Torrent Project

» see torrentproject.se as it lived, on the wayback machine

~ the last words ~

Torrent Project hosted nothing. It crawled the indexes of sites that did, ExtraTorrent among them, and put the results behind one search box with a plain interface and no advertising wall in front of the link. By the middle of the decade it claimed over eight million torrent files, and it had a public API, which meant other people's apps could query it without anybody visiting the site at all.

The Internet Archive has torrentproject.com answering as a torrent search engine in December 2010, so it had been running for years before most people heard of it. What made it visible was the legal attention rather than the traffic. In October 2014 the record labels won a UK High Court order against twenty-one torrent sites and torrentproject.com was on the list, which is a strange kind of promotion: a court telling every ISP in Britain to block an address is also a court telling everyone the address exists.

Then 2016 happened to everybody else. KickassTorrents was seized in July when its alleged operator was arrested in Poland, and Torrentz.eu, which was the other great non-hosting search layer, closed itself in August with a farewell line and nothing else. Torrent Project was one of the few things left that worked the same way, and the Italian and Spanish tech press spent that autumn pointing readers straight at it.

~ the record ~

~ the succession ~

Inheriting an audience is not the same as being able to hold one. The metasearch model had a structural weakness that the hosting sites did not share: if the indexes it read were being taken down one by one, the search layer degrades quietly rather than dying loudly. It also picked up Torrents Time, the browser plugin that streamed a torrent while it downloaded, at a point when that plugin was itself under legal pressure and about to disappear.

The other thing an anonymous operator inherits is the enforcement attention that came with the traffic. Nobody ever published who ran Torrent Project, which is normal for the category and makes the ending impossible to report properly.

~ the last week ~

The archive has the search page intact on 30 August 2017 at 14:18 UTC, input box and all. TorrentFreak wrote on 1 September that the site had been down for nearly three days behind a 403 Forbidden and that the operator was not replying to messages. Several torrent sites were fielding suspicious traffic that week. Whether this one was under attack or had simply been switched off, they said plainly, was unknown.

On 4 September the crawler got a bare nginx 403 page, four lines of default error HTML. Every capture after that is a 403 or a redirect. The rest of the internet settled on 4 September as the date because it is the first day anybody wrote down.

~ where a 403 is ordinary ~

Reading this from Turkey in 2017 gave the ending a flatness it did not have elsewhere. Wikipedia had been blocked in this country since 29 April that year and stayed blocked until January 2020, so an error page where a site used to be was not an event, it was Tuesday. A torrent search engine dying behind a 403 and a site with six million articles being unreachable by government order produce the same screen, and after a while you stop reading the screen as information.

Today torrentproject.se still answers 200. The page is a Swedish-language casino affiliate site, which is what expired domains with residual search traffic get resold into. The index is gone and the address is worth money to someone.

@}-->--

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

« previous  ·  random  ·  next »