Bitzi

2001 - 2013
dead us
before you downloaded a file off a stranger there was no way to know what was in it. bitzi hashed files and let volunteers write down what each fingerprint actually contained, so you could check before spending an hour of modem time on a lie. it never got large and it never made money. two things it built along the way outlived it by decades: the tiger tree hash, and the magnet link that is still what you click to start a torrent in 2026.

~ the last words ~

The mislabeled file was the tax on that whole era. You searched for a song on a network full of strangers, picked whichever result showed the most sources, and started it. On a modem at home a four megabyte mp3 was the better part of an hour, and there was no way to learn until the very end whether you had the song or forty seconds of silence carrying the right filename.

Bitzi's answer was a fingerprint. Run a hash function over the contents of a file and you get back a short string that depends on every byte in it, so two files with matching fingerprints are the same file no matter what either one has been named. Bitzi called its particular combination of hashes a bitprint. Then it did the part that could not be automated, and let volunteers attach a description to the fingerprint: what this actually is, and whether the rip is worth the hour.

The whole thing was a card catalog for a library nobody owned. The data set had a name, Bitpedia, published under a Creative Commons license so anyone could take a copy. An open-source tool called the Bitcollider ran on your own machine, hashed your files and pulled out whatever metadata it could read, so contributing was a matter of pointing it at a folder. LimeWire and Shareaza both shipped a right-click that sent a file's hash to Bitzi and opened the page, and older BearShare builds had one too.

I had the problem and I never used the fix. My downloading years were the internet cafe and the home line between roughly 2002 and 2005, and the method for judging whether a file was worth starting was the file size and the number of sources, which is to say guessing. Bitzi is one of those things that existed the entire time and never reached the people it was built for, because using it meant caring enough to look something up before starting a download, and almost nobody did.

~ the record ~

Born2001, san francisco
Peakthe mid-2000s, when limewire and shareaza both shipped a look-it-up button
Killed31 december 2013
Lifespantwelve years
Made bygordon mohr and mike linksvayer
Killed byits own problem going away, once files stopped coming from strangers

~ what a bitprint was ~

A single SHA-1 over a file tells you that two files are identical and nothing else. Bitzi needed more than that, because the practical question during a download is not whether this file matches that file, it is whether the part that has arrived so far is correct.

So it used a tree. A Tiger tree hash cuts a file into small blocks, hashes each block, then hashes the hashes in pairs all the way up to a single root value. That means any chunk can be verified on its own, in any order, without holding the rest of the file, which is precisely what you need when the bytes are arriving from six strangers at once. Gordon Mohr worked that out at Bitzi. Direct Connect and Shareaza adopted it, and the same structure ended up underneath BitTorrent v2 and IPFS years later.

~ what replaced it ~

The other thing that came out of Bitzi is on more screens today than any product it ever competed with. In June 2002 Mohr published an open specification for a URI that would identify a file by what it contains rather than by where it lives. eDonkey had ed2k: links and Freenet had its own scheme, both welded to one network. This one was deliberately neutral, so any client could read it and any network could serve it. He called it magnet.

Every magnet link anyone has clicked since came out of a file-metadata project that was already struggling to get people to contribute. It has outlived Bitzi by more than a decade and it will outlive most of what is buried on this site.

Mohr's next job belongs in the obituary too. He went to the Internet Archive and became technical lead on Heritrix, the crawler, and on WARC, the format web archives are stored in. The man who could not get enough people to write down what was inside a file went and built the machinery that writes down what is on the web. A good deal of the research behind these graves runs back through his work, which is a strange thing to find at the end of a footnote.

~ what answers now ~

Bitzi shut the website down on 31 December 2013 without much noise. By then the question it answered had largely stopped being asked. Files came from a store or a stream, and when they still came from strangers they arrived with a torrent that carried its own hashes, so the catalog in the middle had nothing left to do.

bitzi.com still resolves in 2026. What answers is a French-language page calling itself a universal digital library with archives and a collaborative metadata catalog, which is close enough to Bitzi's own self-description that it looks deliberate. It is not. It is a generated page sitting on a lapsed domain, with no data behind it and no connection to any of this.

~ leave a rose ~

an anonymous tribute. one rose per visitor per day.

@}-->-- ... roses left here

~ leave a wreath ~

a rose fades by morning. a wreath stays on this grave for good, with your name on it.

from $1, paid once through stripe. it keeps the graveyard running: the hosting and the nightly digging. adopt it and the headstone reads "maintained in memory by" you. the groundskeeper plants new wreaths at the next tending, usually within a day.

» one email a week: new graves and the anniversaries of the dead. subscribe

~ leave a note ~

visitors before you have left these graveside notes. anonymous welcome.

other utilities: Sneak Prevue, Sygate Personal Firewall, RichCopy, Google Swiffy

also died in 2010-2014: Nexus Q, Is Anybody Down?, Pixorial

« previous  ·  random  ·  next »